Which is of course crucial because as we know, the best workout routine is the one you’ll actually do. Because willpower is dumb and also fake.

Why do I dislike most resistance training routines?

Aversive Micro-Interactions In Different Resistance Training Regimens

Barbell Weights

Okay so this is, I understand, theoretically the best option for building muscle and strength (according to Starting Strength and friends), and you can do the Big 3 powerlifts (squat, deadlift, bench press) using just a barbell and weights. You can go quite a long way with just a barbell and weights!

Barbells are, however, a royal pain in the ass. Why?

• Crush risk. Both squats and bench press position the bar above you so that if you fail the set the bar naturally would like to go through your body. In squatting, you can push the bar back off you, which is noisy and bad for the floor but probably safe enough; in bench presses, you’ll really want a safety bar. Who has time to set up a safety bar?

• The bottom of the motion is where all the effort is concentrated. I find this more aversive with deadlift and squat than bench press. That moment in time when you’re just starting a deadlift and the bar is barely moving and you just have to put in more effort? Hate that shit. And with the squat I feel like I’m trying to find the happy middle ground between “can do the rep” and “still going down all the way.”

• Setup time. You gotta take the old weights off. You gotta put the new weights on. What, you think I’m at the gym to do work? Get outta here.

• Floor space + expense means this basically just happens at the gym. Disregard all of the above; barbell racks are enormous and costly. Which means we have Gym Startup Time as a consideration.

• Putting on/taking off weights when the bar is on the floor. Gotta maneuver the bar into the plate’s hole juuuuust right and probably you’ll have to adjust a few times so it goes in smoothly (insert mildly crude sex joke here.)

VERDICT: Theoretically optimal; pain in the ass and I don’t like doing it. Also, empirically I won’t do it.

“Is it possible you’re just a wuss?”

Yeah, whatever, maybe. Who cares; if that’s my problem, it did not get resolved by months-long semi-consistent barbell routines launched at various points in my life.

Dumbbell Weights

This resolves some but not all of the problems above (in particular, the crush risk is eliminated). We now have:

• Squats aren’t practical at full weight. The bar is (excuse the Claudeism but the pun is worth it) load-bearing here.

• With dumbbell deadlifts I feel vaguely like I’m about to crush my testicles at the top of every rep. I dislike this.

• Also with deadlifts the dumbbells require you to get much closer to the floor than the barbell does. This is because the weights you use on the barbell are much larger in radius than the typical dumbbell is.

• Lugging weights into position is still annoying and tiring without being a particularly good training stimulus. Idk dude when I was working with a trainer I’d have to lug these enormous 80 pound dumbbells to the bench so I could do a set of bench presses and frankly the process of doing this was maybe half as much work as doing the actual bench presses.

Verdict: achievable at home, still kind of annoying to do the relevant exercises with.

Bodyweight Routines

“Who says you even need equipment?” Hey, valid! Let’s talk about it.

• Pushups require me getting into a weird horizontal position that is mildly unpleasant. The blood always rushes to my head during pushups. Tolerable, but annoying.

• Most exercises require quite a lot more reps which means now you’re doing cardio as much as resistance training. Ideally resistance training for strength and hypertrophy (muscle gain) has you doing a limited number of hard reps (e.g. 6-12 reps in a set.)

• Pull-ups/chin-ups require either large freestanding equipment or a doorway you trust to hold a pullup bar. (I don’t trust my current apartment’s doorways for this.)

Verdict: this is probably better for cardio than muscle growth, and I’m less interested in cardio (I use salsa dance for that).

Also I hate cardio.

But machines! Or cable stacks!

Full disclosure: I love me a good cable machine.

Most of the actual problem here is that unless you’ve got a really spacious place and a shitload of money to blow then you’re definitely only going to be using these at the gym, and that means you’ve got to Go To The Gym, which means you have to change clothes and drive to a place and park and check in and now you’re allowed to use the machines and/or cable stacks.

That’s so much labor! Uncompensated, even!

God Bless Band-Weighted Barbells

I find that the Big 3 lifts have quite accessible and pleasant analogues here. The tradeoff is that you’re getting way less resistance at the bottom of a given movement. But the prize?

• Bench presses are pleasant and can be done standing up. Excellent, I can watch TV with the roommate. (The same goes for basically all these, really. Also: no noise.)

• Deadlifting is feasible and surprisingly pleasant. The decreased resistance at the bottom of the movement means that starting a rep doesn’t have the irritating period of “I’m pulling really hard but this thing isn’t moving” feeling that the ordinary barbell deadlift has.

• Squatting feels good also. Again, decreased resistance at the bottom of the squat, but in exchange I don’t feel like I have to exert Constant Vigilance or else potentially tip over. (Yes, yes, skill issue.) Also if I tip over the bar is like ten pounds, who gives a shit.

• Various accessory exercises are easy and straightforward. Sitting bench. Overhead press. Bicep curl. These all seem tunable to be difficult while still allowing the full range of motion.

• Joint safety, supposedly. I read somewhere that joints are most vulnerable at the start of most lifts, which is where load is lightest in band exercises. In theory this has safety implications and at the age of 38 I am now older than God so that seems good.

But the costs are real:

• No number-go-up. Going until failure is fun and reasonable here, but the thing about bands is that they don’t really tell you how much you’re lifting because bands exert way more force the further they’re stretched. Strength gain is more subtle here; you won’t be able to track progress near as easily.

• Decreased resistance at the bottom of the rep: threat, or menace? There exist claims (which I decline to evaluate) that the struggle at the bottom of the rep is where most strength gains are to be had. That said, there are enough resistance band enthusiasts that I expect this is not that big a deal in practice.

• The limiting reagent in deadlifting becomes “grip strength”. That’s not the worst thing imaginable, but it is noticeable.

• Bands snap sometimes, and the whiplash can hit your eye, rendering you blind. Don’t wanna go blind. Accordingly: wear safety glasses and replace the bands when they start to show tears.

Anyway, if you’ve been wanting to get into resistance training but find that weight lifting sucks to do and also you hate it, try band-weighted barbell systems along with perhaps a comfy pair of safety glasses. The cheapest ones are on Amazon and they can be had for $60 bucks or so; if you hate it, send it back, and if you do like it, hey! Now you have a workout you’ll actually do. Also apparently you can spend up to several hundred dollars on very fancy band-weighted barbells, which starts to make a lot more sense once you’ve worn out the cheap Amazon ones.

Will I still think this in a few weeks after the novelty’s worn off? Idk really but I feel optimistic.

After the final workshop ended, it was a pretty emotional experience for most of the men; I was crying and hugging one or two of the models involved while, in the background, he was having a heated argument with one of the more-prickly models about— something, I’m not quite sure what exactly, but it ended with the girl yelling “I AM NO LONGER GETTING PAID FOR THIS CONVERSATION SO IT IS OVER.”

Anyway, that was fun!

Likewise, my first workshop with that group featured me very very nervous with the models. I think I had some impression, not consciously noticed, that if they came to a good conclusion about me that meant I was good and otherwise I was bad. Accordingly every single interaction became deeply stressful; every single thing I said or did passed through the filter of “what perception will this cause in the girl and is that desired”.

One time we were doing soulgazing— literally just staring into a model’s eyes for several minutes at a time— and the model began to tell me that I was clearly trying. She began to snap her fingers in front of my face, saying “stop trying. Stop trying.” I remember being both frantic and also baffled as to what she was asking me to do; every single way i could contort or slacken my face resulted in her declaring that I was still trying.

(This did not help, but i think it was diagnostic.)

There are equal and opposite errors someone can make when faced with critical feedback. The first is to immediately reject it: you don’t have a problem, this other person is the one with the problem. Maybe he sucks!

And the second error is to declare yourself as wrong, immediately, to try and immediately swap out your original self-narrative that’s now under threat with the different and perhaps more easily-defended self-narrative as “guy who takes feedback well.”

I think these are both, basically, just manifestations of social anxiety. Because they are both fundamentally impression-management driven: in the first case you’re defending your original fortification and in the second you’re hiding in a different, newly-constructed fortification because your first is under attack. But either way you’re still hiding; you’re still acting as though your existence is contingent on either the other person either (1) liking you or (2) being BAD AND WRONG. Someone who is fairly decent and logical having a negative opinion of you? THIS CANNOT STAND.

Hence the traditional refrain: “yeah all my exes were crazy/evil.”

At some point-- actually, during that same workshop series-- I realized that I actually didn’t need to either immediately accept or refute any specific piece of feedback. I could just, when someone would criticize or comment upon my conduct in a way I disagreed with, instead just nod and say “noted.”

Which I think is the most honest response: I was, in fact, noting what they were saying without making a particular commitment to act on it in any specific way. Actually I think this is doomed to be the fate of most real life social feedback. There is a saying: if one guy says you’re drunk, maybe he’s just an asshole, but if three guys say you’re drunk then maybe call an Uber to give you a ride home. The question of whether a person is the first-of-three or just some asshole is, regrettably, unanswerable in the moment.

The flaw the rejector and the capitulator both share is the inability to let the criticism’s status go unresolved.

The Test Is “Why Is Arguing This Point Useful”

In the dating workshop, the guys were getting correction all the time. Hell, at Slutcon (happens again in a month!) I saw all sorts of guys get correction from the women. And there was a frequent compulsion that the guys would feel to explain themselves and put themselves in a more reasonable light after receiving critical feedback. (“You seemed people-pleasy there.” “I wasn’t people-pleasing, I was just…”). I’m sure I did it once or twice; it’s harder to catch yourself in the act than you might think, if you haven’t tried.

The reason why this was an anxiety tell is straightforward: there’s no reason independent of impression management why arguing the point (“no, that wasn’t hover-handing, because…) is useful at all in the dating workshop environment. You’re receiving feedback on how your actions were perceived in the moment of you having done them; your explanation will not turn back time; therefore your explanation is, basically, irrelevant noise even if it is entirely correct.

Compulsive People-Pleasing Vs Taking Feedback Well Vs Stonewalling

I think Taking Feedback Well is a skillset that is, in many ways, dependent on you having a locus of correctness that is independent of what other people say and think.

One definition of “security”— a definition that I think is deeply good and useful— is “being able to exist in a room with someone who has opinions about you that are importantly contrary to your self-narrative without needing to either repair their opinion or expel them from the room.” (I think this is indistinguishable from what some people call ‘grounding’ or, alternatively, ‘healthy masculinity’. ‘toxic masculinity’ is, among other things, the absence of this virtue.)

This is a stance only available to you if your self-narrative is not deeply reliant on the opinions of arbitrary other people in the room with you right now. If you have come to a considered decision about The Good and have aligned your actions to that, I think this gives you a sorting mechanism that lets you separate out Important Critical Feedback from Irrelevant Noise.

You might still take feedback about the aesthetics of what you say and do from complete randos, but this ought to be considered tactical and amoral in a similar way as commentary re: your wardrobe. You might also notice that someone makes a good point: that your statements and actions don’t line up with your real values. In which case, it is worth bringing some thought to the question of correcting this.

Other feedback?

Eh.

“Wait, but reputation sometimes matters materially! Surely I should argue my point then?”

Perhaps rarely, but much less often than you might think. That’s because “defensiveness” is a very well-trodden behavioral pattern and immediately after you receive feedback, anything you say by way of context or explanation that can be rounded off to “defensiveness” will be. “Fielding a criticism” is basically the maximally hostile interpretive frame for the feedback recipient.

If you have externally-verifiable information that reshapes everything by all means say it, but short of that, any kind of nuancing is more likely to get you placed in the “can’t take feedback” bucket than it is to land whatever point you wanted to make. I think “noted” is frequently the optimum in terms of making it clear to the other person that you’re at least going to wait and consider before completely rejecting their feedback. (Is it possible you’re just saying ‘noted’ and will never consider it again? Sure, probably. The difference in-the-moment is inconclusive to onlookers; most of the function here is to avoid the Obviously Can’t Take Feedback Bucket.)

If the situation is truly vital and warrants it— a pretty rare case, frankly— then you can engage in the repair attempt with justifications after perhaps a day, after the Danger Zone has passed where reputation repair gets reflexively rounded off to “probably just defensiveness.”

And if the situation is not one where your physical or monetary security critically relies on this specific person’s private opinion of you… or if the faux pas or whatever you’re worrying about isn’t truly apocalyptic… well, now you’re trying to solve a problem that both can’t be solved reliably and which doesn’t particularly need you to solve it.

Let the person’s new opinion be what it is.

Also one-offs are structurally unimportant in iterated interactions

Basically nobody you’ve spoken to any significant number of times will write you off due to a single unfortunate event (barring, again, truly apocalyptic doings.) You are the weighted average of someone’s interactions with you.

That’s why, if someone already dislikes you, arguing the point about specific incidents generally won’t move the needle: you’re like someone trying to argue someone out of the politics they’ve spent a lifetime building their opinions on.

And if someone has just met you and formed a negative first impression, that’s regrettable; however, either they will have additional interactions with you (and thus the impression will gradually average out to whatever it ends up being) or they won’t have additional interactions with you, in which case their negative opinion of you shouldn’t motivate any particular urgent action on your part.

I’m not sure how to think of Admonymous and friends

’s You Can Just Do Things advocates for using anonymous feedback tools to improve the impression you’re giving off.

I’m undecided about how to think of this. The benefits— getting concrete information about others’ perceptions, like weird odors or whatever, which you would not have otherwise had— is real.

And if you’re afraid people are going to write shitty things about you, it is deeply worth interrogating why you fear this, because there is no a priori reason for you to fear that and excavation is likely to be useful for you. (I suspect a partial answer is that you take other peoples’ opinions too seriously because you have not come to a firm conclusion about how you are aligning your actions to the Good, which means you’re outsourcing this to other peoples’ opinions which are intrinsically unstable.)

But on the other hand, I suspect anonymous feedback forms are likely to make the question “how do my actions look from the outside as evaluated by randos” feel far more salient to you, and social anxiety writ large is basically the disorder of “the opinions of others are wildly too important to you at baseline” so if this is a problem you have maybe you shouldn’t encourage that tendency.

Here’s my test: am I checking feedback to get wardrobe-importance-tier information about how I’m coming off, or am I checking feedback to Know Whether I Am Okay? If the former: excellent, I shall carry on. If the latter: this is more like I’m an addict taking a long drag off a desperately-needed cigarette than it is like self-improvement. (This also applies to non-technology-based forms of feedback-seeking, on reflection.)

And how do you know which bucket you’re in? Simple. Imagine the moment of opening up the feedback inbox. Does it feel scary, to you? Does it feel like you’re opening up your acceptance or rejection letter to the college of your dreams, and you’re not sure which it is yet? Do you find yourself avoiding opening up the feedback inbox because of what you might find? ‘Cuz you’re not going to feel that way if you’re getting wardrobe-tier information. You’re only feeling that way if it feels like the inbox contains a verdict that must be taken seriously.

(Hilariously this is one of the few places where I think the avoidance instinct— “don’t make a form for feedback bc it’s maybe negative and that’s scary”— is coincidentally guiding your actions correctly. It’s just correct for totally unrelated reasons than its actual felt reasons: you shouldn’t open the inbox up in this state because that’s a reliable indicator that you’re going to use Admonymous or whatever as validation that your existence is okay, which is an abuse of the system. If you can fix that, do so and then be free: if you can’t, probably don’t use an anonymous feedback form.)

But I am uncertain about where the average person falls on this spectrum. If you’re not predisposed to social anxiety such practices seem like an easy win, at least.

We are now one month away from Slut Con, the relationship bootcamp for nerdy guys who want to level up their skills, and I cannot wait. The media is already abuzz in speculation.

Who are the women who will be volunteering their time and energy as flirt girls? Are they professionals? I can’t speak for everyone, but overwhelmingly, we’re just flirty women with professional degrees. Engineers, analysts, mothers, medical providers, writers, and researchers. We run the gamut.

Why would we do this? I can’t write this out on my personal substack. I don’t talk like this in my normal life. If my coworkers knew what I was doing next month, ooof. But I need the guys who are going to Slut Con to know: I want to fuck. I like nerdy guys. I’m a nerdy girl. I like men who can match my intellectual level. Slut Con is THE place to be for this. Flirt girls are not expected to sleep with the attendees. We are not expected to even let a man touch us. Any physical escalation is entirely on our own discretion. That being said we are not discouraged either.

I want everyone to have successful and fulfilling relationships, be they for the conference, or the rest of their lives.

To that end, here are some pointers from an anonymous flirt girl on how to take me to bed.

1. Get tested. The social expectation to have sex with me is to have a standard 4 panel + HSV within the last 3 months. Ideally within the 2 weeks before the event. A standard 4 panel includes HIV, syphilis, chlamydia, and gonorrhea. Have the test results on your phone because I’m going to ask to see it.

2. Have protection. Make sure it fits. One of my guy friends recently complained that he disliked condoms because he finds putting them on to be adverse to maintaining his erection. Oh, he was safe, but he just didn’t particularly like them. Reader, he was wearing too small a size and never knew. I provided him with a larger size and much fun was had. If you’re unsure of what size you need, bring boxes of several sizes. We flirt girls signed up to help out you guys. If we’ve elected to go to bed with you, we are not going to disparage you for not knowing what your optimal condom size is. Also make sure you are keeping those condoms safe. Don’t keep a condom in your wallet or car to get degraded and ruin its structural integrity.

3. Practice mindful escalation. Don’t just grab the flirt girls. Let me repeat myself. Do NOT grab the flirt girls. Even if you think it’s okay. Even if one girl looks like she just got groped by a stranger and you think gee, maybe I can grope her too! You don’t know the full backstory. I have a flirty relationship with a lot of the male volunteers. My consent to my close male friends does not automatically extend to all Slut Con attendees. Approach each flirt girl as an individual and escalate appropriately. If you’re unsure if you can touch her, just ask. This is a safe space for asking. There will be training on how and when to approach flirt girls at the event.

4. Kindness will get you further than you can possibly imagine. We flirt girls have a whisper network. Some of you reading this will worry you’re being watched, and in a sense, you are, but not in the way you fear. What travels through the slut cloud is narrow and protective: if you’re exceptionally kind behind closed doors, that gets passed along in pro-social contexts. If you make one of us feel unsafe, that gets passed along too. All disclosures are kept to the minimum needed for mutual respect and safety, and nothing else. In all my time in the slut cloud, I have never once heard a disclosure about a man’s anatomy or health; no one is comparing notes on erectile dysfunction, no one is making a disparaging remark. That is not what this is for.

   What it is for is kindness, because kindness is the currency here. I have been known to ask my favorite partners if I can recommend them to other flirt girls, and I will set you up if that’s what you want and you so impress me. That doesn’t mean I expect perfection; missteps at these events are expected. A good lover is one who takes feedback with grace and is joyfully present. How you handle the misstep is where your worth lies.

5. Aftercare! This need will fluctuate wildly between flirt girl to flirt girl. Be open-minded to all needs. I like hugs and snuggling. That doesn’t mean I expect us to be boyfriend/girlfriend, but I do expect de-escalation appropriate to the intensity of the interaction.

6. Don’t write about me/take my picture without my permission. I have a life beyond this event. I have a professional career that does not involve flirting with cute nerdy guys. I won’t dox you for attending. Don’t dox me for volunteering.

STI Addendum, written by Aaron:

Note that Planned Parenthood (and, I assume, some other places) won’t test for herpes (HSV) 1 or 2 unless you’re showing symptoms, but some girls really still prefer an HSV test in hand. I am also told that https://prodxhealth.com/products/pro-13-home-collection-kit?variant=42594725298278 has most of the panels somebody would conceivably want (or really you can just use just any PASS-certified test) EXCEPT herpes and ureaplasma. To repeat, the linked kit doesn’t include HSV or ureaplasma, so you’ll have to get that separately if you want it. The girls prefer this to be done <2ish weeks before the event if possible.

A list of things to be tested for if the possibility of sex at SlutCon is something you value:

• Critical panels: chlamydia, gonorrhea, syphilis, HIV (HIV-1 and HIV-2). (These are what you get if you visit a planned parenthood and ask for an STI screen)

• Others panels commonly requested: hepatitis B, hepatitis C, trichomoniasis, Mycoplasma genitalium, Herpes Simplex Virus (HSV-1 and HSV-2), and ureaplasma.

See Also:

“Sir?” The inscrutable gadgetry which Jeeves had recently installed into his head gave a sound similar to the ticking of a regrettably-irregular clock. I attempted to ignore it.

“These trousers are perfect. Look at that crease. You could shave with it! A man could go out into the world and whatever befalls him, whatever errors he made, he would know that, at very least, his legs were above reproach.”

“Thank you, sir.”

“I should say— if you could ensure that every day’s trousers were equally as splendid that would be absolutely the bee’s knees.”

I could hear a sound rather like a well-bred bumblebee clearing its throat as Jeeves fixed me with that newly-uncanny gaze of his. “Every day, sir? These trousers are quite exceptional; it may take some effort to ensure that each day you can wear such well-creased trousers without rendering your wardrobe repetitive.”

I would not be deterred. “Spare no expense, Jeeves! A man’s crease is, as they say, his castle.”

“Just so, sir. I shall give the matter some thought.”

I did not, at the time, ask how much thought. This was an unfortunate miscalculation. One does not, dealing with an begadgeted valet of Jeeves’ surpassing skill, leave the quantity of thought unexamined.

But the morning was fine, the trousers were magnificent, and I went off to the Drones with a song on the lips.

Some days later, I received a letter from my bank. I read it twice, then a third time, in the displeased manner of a chap who suspects he is being had on.

“Jeeves,” I said, “there appears to be a deal less money in my accounts than there was.”

“Yes, sir.”

“A rather substantial sum less, in fact.”

“Quite so, sir.”

“Jeeves, I believe the bank has perhaps misplaced a sum of my money.”

“No, sir.”

I paused. “No?”

“You had to withdraw money to fund the purchase of a tailor.”

I felt similarly to the time I woke up hungover after the yearly boat race and was, gently, informed of a series of entertaining pantomimes I had apparently given. I lowered the letter and fixed Jeeves with the terrible Wooster Stare. “I own a tailor?”

“Hepplewhite and Dunne, sir, of Savile Row, as of the tenth. I telephoned you at the Drones before signing, whereupon you said— and I have it verbatim— ‘Jeeves do not bother me with this, I am at poker and taking Tuppy Glossop for all he is worth’.”

The Wooster Stare intensified; I wonder that I couldn’t see a spot on his forehead begin to smolder. “Jeeves, why do I own a tailor.”

“For the trousers, sir.”

“Elucidate.”

“Certainly, sir. You will recall a desire for such trousers daily?”

“I do.”

“The trousers came from one Hepplewhite and Dunne; no other house can make them. However, the firm is booked out seventeen months ahead and Mr. Hepplewhite could not be persuaded to move you up the order book.”

“So you bought the firm?”

“So you bought the firm, sir.”

I was conscious of a sinking feeling somewhere around the third waistcoat button.

“Jeeves, can I afford this?”

“Yes, sir. Mr. Hepplewhite, while gifted as a tailor, was rather less so as a businessman; accordingly the tailor could be purchased on a discount. Some minor improvements to the business should result in you making the money back within the year, plus a small consideration extra.”

It was rather like discovering that the tiger who has just eaten your uncle had, on reflection, decided to pay your rent by way of apology.

“…ah. Good. Carry on, then, Jeeves.”

I wish I could say that I put a stop to it then. I wish I could report that Bertram Wooster, sensing the approach of something large and ominous, planted the feet, squared the jaw, stuck out the hand, and said “no further, Jeeves.” But the sheer, blasted fact of the matter was that the trousers were uncommonly good, the money was quite welcome, and when a chap’s income has apparently increased in the course of improving his wardrobe it takes a sterner man than I to call a halt.

So I let it ride. And for a season it seemed to have settled.

Then Aunt Dahlia telephoned.

“Bertie,” screeched my dearest aging relative, in a voice to flush terrified squirrels out into the hunting field, “what the absolute devil is going on in Bradford?”

“Bradford?”

“Bradford, yes! I’m talking about the mills, you abysmal fathead! Tom’s just come back from his club in the state, saying some concern has bought up three of the largest worsted mills in the West Riding, and when I requested Seppings to look into it he said the ‘concern’ is you.”

I covered the mouthpiece. “Jeeves.”

“Sir?”

“Do I own three worsted mills in Bradford?”

“Technically only two of the mills are in Bradford, sir. To answer what I suspect was your intended question, you now own four mills in total; the fourth you purchased two days ago.”

I uncovered the mouthpiece. “Apparently it’s four.”

“Bertie, I’m fond of you— probably the same reason one grows fond of a particularly irritating but persistent seagull— but in God’s name, explain to me why you have four mills?”

I thought quickly, mind racing like Tuppy in pursuit of the last sausage at breakfast, and suddenly with the classic Wooster resourcefulness I came to a solution. “Hallo? Hallo? Operator?”

“Bertie, I know you’re still there, you pie-faced young blot! I’ve seen more intelligence in a retriever! And the retriever was dead!”

“I say, are you still there, my dearest aunt?”

“Don’t ring off, you knock-kneed—”

I replaced the receiver with the tender care of a man setting a lighted firework on a mantelpiece.

“Jeeves,” I said carefully, “please explain yourself.”

“You recall you requested that you be able to wear trousers with the perfect Hepplewhite crease each morning?”

“Yes?”

“Mr. Hepplewhite’s gifts, sir, while considerable, are quite at the mercy of the mills which produce his cloth. The Bradford product of late has been uneven; upon reflection it seemed simplest that you should own the mills.”

“Jeeves, why was I not informed of these purchases?”

“You were, sir.”

“I was?”

“Yes, sir. The latest purchase I proposed to you of while you were in the middle of a crossword; you responded with precisely the words ‘excellent, make it so, Jeeves’ whereupon you requested a seven-letter word for ‘surprising.’”

A treacherous memory held up a card. “Well… blast it, that doesn’t require three mills. I only have two legs!”

“Four mills, sir. It is uneconomical for a tailor to produce trousers for only yourself; in order to run the business at a profit we need to guarantee the order book, which itself requires us to supply other tailors, sir, whereupon we can realize certain economies.”

“Can I afford four mills?”

“No, sir. You have taken out a bank loan using these businesses and your family estate as security.”

“I asked you for trousers!”

“You asked for these trousers with this exceptionally fine workmanship, sir, and I endeavor to provide satisfaction.”

There was a silence. Perhaps reading the expression on my face, Jeeves said, “May I offer a whisky sour, sir?”

“Please do, Jeeves.“

Somewhere around the purchase of the coal mine (and the less said of that sorry affair, the better), the thing took on a momentum that I shudder to think of. One evening in November a narrow-faced harried-looking man showed up upon my doorstep.

“Mr. Wooster,” he said. “ I’m terribly sorry to call on you unannounced, but I was rather hoping for a word about some of your, ah, business interests?”

I took a deep breath to call upon Jeeves, but on turning round he was already there, silent, like an unusually helpful but ominous cat. I turned back toward the guest. “Yes, quite so. What do you need?”

“Mr. Wooster. It’s the Gold Standard, you see. The Chancellor is concerned that… which is to say, the Treasury has reason to believe you may have a position—”

“Hold up. Jeeves,” I said cautiously, “do I have a position on the Gold Standard?”

“You have been obliged to form one, sir.”

He sympathetically offered me a whisky and soda. There is a tide in the affairs of men, as the fellow said, which taken at the flood leads a man on to… something. Anyway, by way of girding my spirit to take on that flood, I took the whiskey.

“And, ah, what is my position on the Gold Standard?”

“I should prefer not to say in front of a representative of His Majesty’s Treasury, sir.”

The man made a small noise.

“Jeeves,” I said, in the dull tones of a man looking down a firing squad, “is the crease in my trousers affecting the monetary policy of the British Empire?”

“Only at the margins, sir.”

I woke up the next morning resolved. There was a Wooster at Agincourt, or possibly Crecy— Aunt Dahlia has the painting— and I came down to breakfast feeling much the same as my ancestor must have felt, staring down the army of that Napoleon chap.

For my resolution was this: I would sell it, the whole lot. The mills, the ships, the mine, the tailor, all of it. I would go back to being a chap with a flat and a car and if the trousers suffered, I would suffer with them like the Woosters before me, a long and noble line of sufferers of mediocre trouserage. (Troussage? I should have to ask Jeeves later.)

I put it to him over eggs.

“Divest, Jeeves. Sell it all, and sell it today.”

“Very good, sir.” He did not move. “Might I raise a consideration before I instruct the brokers?”

“You might raise one, but I warn you, it is like to be stomped right back down again. The Wooster mind is made up.”

“Quite so, sir. I shall observe that selling it all at once may result in your holdings taking some losses.”

“Some losses?”

“Yes, sir.”

“…how much in losses?”

“We would be down to twenty per cent of your current holdings by my estimation; perhaps seventy per cent of your original holdings from before the tailor acquisition.”

The blood rushed from my face like Catsmeat from the Drones after too much vermouth. “Why, damn it all?“

“Slippage, sir.”

I buried my face in my hands, feeling rather like I expect Jesus did before that Judas fellow boarded him up in that tomb. “Explain, Jeeves.“

“One cannot, sir, sell four mills on Wednesday and expect a fair market price.”

“I don’t care, blast it. This dashed business has gone on long enough, and however poor the Wooster— as they say, however poor the Wooster he still has the old Wooster dignity.”

“Quite so, sir. If I may suggest another consideration before I inform the brokers of your decision?”

“Suggest away, blast you.”

“To sell, sir— more specifically, for you to be seen to sell— would be felt in the City. And rather beyond it, I suspect.”

“Beyond it? You mean beyond London?”

“Beyond England, rather.”

“And… how strongly felt, would you say?”

“I should expect the word ‘calamity’ to feature in the newspapers, sir.”

There was a long silence. There suddenly rose before me frightful images of angry letters in the post.

“Jeeves,” I said carefully. “I can’t get out, can I?”

“You can, sir. At your word I shall inform the brokers immediately.”

I looked down at my trousers. They were, I noted, immaculate. The crease ran from hip to instep like God Himself had drawn the line, with the stitching done by none other than the Archangel Wossname. They were the best trousers any man had ever worn, and as my eggs grew cold I knew I would wear them for the rest of my life.

“When you said, back at the beginning, that you would be giving the matter of my trousers some thought.”

“Yes, sir?”

“How much thought have you given this matter?”

The contraption built into Jeeves's left hemisphere flickered and made a noise like walnuts being poured down the stairs.

“All of it, sir.”

See Also

First: anxiety disorders (such as social anxiety) are generated directly by dysfunctional attempts at achieving an impossible or unverifiable goal; the specific disorder you “have” is downstream of the specific dysfunctional goal you’re habitually attempting to achieve. So far so good.

And second: this effort consists primarily of directed mental effort, not physical or verbal actions as-such. Physical compulsions matter chiefly as evidence for the mental compulsions which spawn them. (See also: RF-ERP.)

This second axiom, taken seriously, is a huge problem because not only do we not have full conscious control over our minds, we also don’t have any way of measuring what our minds our doing and especially can’t tell what other peoples’ minds are doing. “Just stop” is an instruction that most people will fail utterly at. Mainstream clinical psychology in the form of CBT has, as far as I’m aware, observed this and mostly responded by only measuring and trying to prevent the physical tells (“safety behaviors”) of anxious people, which doesn’t work particularly well as judged by measured remission rates.

We are like the drunk searching for his keys under the streetlight when he actually dropped them in the dark, where "the light” in this case is “behavior we can see and measure and intervene in directly.”

The RF-ERP school of thought is what you get when a clinician notices this and is like “excuse me I’m pretty sure the keys are over there where it’s dark” and now we have a whole different set of problems which are basically just the problems mainstream CBT research has decided to punt on. Like: how do you even track whether this very specific internal phenomena is happening? And how do you stop it reliably? We are now inheriting those problems by virtue of not punting on them.

So. What do you do, if you’re trying to help someone stop trying to solve an invisible and impossible problem in the privacy of their own head?

Buddhism: Or, The Religion Where All Human Problems Are OCD-Shaped

Buddhism is, partly, the claim that (1) suffering is caused directly by ongoing mental effort toward impossible goals; that (2) this effort is mostly invisible and compulsive for the untrained person; and (3) that there are trainable methods for making it visible and dropping that compulsive effort. (There’s also a whole cosmology that goes along with this, which I’m less interested in.)

That means Buddhism has a couple millennia worth of head start on this problem, with the primary difference between Buddhism and third-way CBT approaches (eg ACT, MCT) being a massive difference in the scale of their proposed interventions, since Buddhism claims that “impossible” describes, on some level, the entire collection of goals that humans hold.

If you take this claim seriously and are looking for techniques for dealing with this, you will end up at something that’s basically meditation, and probably sooner rather than later.

The Insight Approach (Theravada vipassana)

The idea here is that if you can see the grasping happen inside you at a fine-enough resolution then this, itself, is adequate for it to cease. The natural human instinct is to orient yourself at the thing being grasped. The insight here, if correct, is that this instinct is in many cases misguided: the thing you actually should attend to is the impulse to grasp itself. Devote deep, inquisitive attention to this. Poke at it.

It’s basically the penny arcade comic.

I think this might be the one I basically subscribe to? Noticing the grasping and then making a decision about whether the grasping is truly necessary/useful is most of what I do, and I find it works quite well for me. The bet I’m making with Abandon Hope is that if I can build a powerful enough microscope with language and hand it to people then this will, for a subset of them, be adequate to make them SEE THE THING and then— the next, equally-tricky step— HALT THE THING.

Which seems to work for a subset of them, shockingly enough. My coaching practice right now is basically trying to iterate on instrumentation to move others over the line.

And part of it— an important part, I think— is just prescribing extremely tedious meditation-flavored exercises, because the grasping impulse operates continuously at extremely low levels and, I’m starting to realize, is similar in quality to other sorts of grasping-style phenomena, such as “horniness” and “wishing your roommate would stop showing you YouTube clips so that you can get back to reading an article” and “checking your phone in an elevator” and “watching a YouTube short you don’t give a shit about to see what happens at the end” and “trying to 100%-completion a game even though you’ve stopped actually enjoying it” and “wanting to give an article you’re writing just one more pass through Claude, bro, just one more, just to make sure there’s nothing wrong with it, just one more time through Claude will 100% ensure that nothing is bad about it that you didn’t know before publishing, just one more pass, maybe we schedule the article to go live tomorrow to give us more time for checking behavior editing.”

So What Is The Grasping Impulse, Exactly?

I’d say it’s basically just two parts:

1. A sense that the present moment is incomplete or subtly wrong. You’re bored right now: you could be entertained. The song you’re listening to is okay: but it could be a different, better song that you’re listening to. You’re reading a book; but you could be reading a better book.

2. A sense that you could address this wrongness, and then the moment would be Complete, and (implicitly) you would be happy at last.

It feels, day to day, like… a lot of stuff, really? Staring at the open fridge while you’re not really hungry in the hopes that you’ll want to eat something inside. Browsing around Netflix hoping something will catch your attention. Feeling as though you ought to be doing something different than what you’re currently doing.

“Anxiety” is what you get when the grasping impulse is noticeably way too strong, for whatever reason, such that it commands an enormous amount of your cognition: so much that it’s noticeable as something you wish you weren’t doing rather than just the unsatisfying texture of standard human experience.

What if GLP-1s are just generalized anti-grasping medication?

GLP-1s help with alcohol addiction (a fact which is unsurprising to me; I accidentally dropped my once-a-day drinking habit when I went on GLP-1s, which I only noticed like a month after I stopped touching my stash of White Claws.)

GLP-1s also help with gambling addiction, apparently.

GLP-1 use appears to be correlated to lower anxiety. (This is extremely vulnerable to various confounds so hold this one loosely.)

There is a syllogism, here, which I don’t think anyone has tested formally.

1. If anxiety disorders are, basically, the universal grasping impulse escalated to a clinically-noticeable impairment, and

2. If GLP-1s are generalized anti-grasping medications, then

3. GLP-1s might be effective anti-anxiety medications in addition to everything else

Note that I think this formulation implies that GLP-1s— if they work for this purpose at all— should only work for patients where they feel they can’t stop grasping, who can’t stop themselves from trying to avert/monitor whatever feared outcome, rather than for patients where the grasping consciously feels like responsible or mandatory behavior. (This is, for readers who have been following along, literally just the positive vs negative metacognitive beliefs distinction in Metacognitive Therapy theory: we’re hitting the subconscious felt pull of the fix-it impulse, not consciously-held beliefs.)

This also assumes the unification of appetitive-grasping and aversive-grasping into the same biological phenomena; this might just be false on a neurochemical level and GLP-1s hit only the appetitive grasping, in which case this would be a no-op. (The fact that SSRIs can improve worry for some patients while also ballooning appetite functions as some evidence against this unification of appetitive-grasping and aversive-grasping. Tough to say.)

This all speculative, of course: I’m neither a therapist nor a doctor. On balance I give it maybe 30-40% odds this actually pans out in a useful way.

Even so, this feels worth testing.

See Also

Anyway, I saw this Twitter thread a bit ago and I had Thoughts.

The Twitter discourse almost solely orients around the asker’s benefits from the ask, and the askee’s costs. But this isn’t like begging where it’s a totally one-way transaction. Both careers and romance are matching problems. In either of those contexts, the person you’re about to make an ask of has, potentially, a problem that you are putting yourself up as the solution to. If you’re seeking a job then if you’re pinging a prospective employer, this can only be because you suspect the employer has a problem for which you— a possible employee!— might be a solution.

There is no other possible reason for this interaction to exist: if you have 100% certainty that the employer won’t be interested then there’s no reason for you to reach out in the first place. Your benefit only comes conditional on the employer’s benefit.

It’s true equally for the romantic bid. If you’re making a pass at someone, then you have generated the hypothesis: this interaction may be mutually rewarding. The mutually-rewarding version of the interaction is, indeed, the only possible rewarding version for the asker. In the rejection case, the asker is paying a similar cost to the askee: an unrewarding and probably awkward interaction.

Pickup artists and con artists: the exceptions that prove the rule

The reason pickup artistry has acquired such a poor reputation is because it breaks the underlying symmetry: negging, “last-minute-resistance techniques”, all that stuff can turn a symmetric win-win (interested, romance for all) or lose-lose (disinterest, no romance for anyone) proposition into win-lose (she’s uninterested but has sex with him anyway.)

But these aren’t the typical version of even maximally audacious asks, and I dislike the current cultural framing where they are. The modal version is much more like: the asker is imposing a transaction on the askee: the askee bears the cost of performing a rejection in-person (or letting the message die quietly in the inbox.) The cost is, when done skillfully, small and bounded. When done unskillfully: somewhat higher, but still bounded.

The askee also receives a benefit, which is the option, if they so choose, to do a business or do a romance with this person.

The askee’s interests

The median value of the option is roughly zero; the average value of these options is— for people who are not popular enough to be drowning in unsolicited asks— often quite high.

Which is why the discourse looks like it does: the median ask is probably just pure cost (since the median ask is rejected), and if you cut off the tails where benefits actually accrue it just looks like pointless waste to everyone involved.

So let’s flip it around and discuss the askee’s interests plainly. How many unwanted asks would you allow to be imposed on you, in exchange for the single desired ask that leads to a happy relationship, or for a lucrative business partnership?

Your future husband or startup cofounder is on the auction block, and his price is denominated in awkward suitors you, personally are going to have to reject, at times and places not of your choosing. What do you bid?

Is it a lot?

Do you bid, perhaps, alarmingly high?

I’ll be real with you: I think the answer should be “yes.”

In full (saved you a click!):

Woke: Deconstructive meditation is after a pretty specific and intense goal that typically requires a lot of work—stepping out of limited notions of identity, a process which creates radical shifts in perception, affect, and, often, one’s view of what life itself is. However, given that, in the moment, limited notions of identity are often constructed on a stance of “problem” or “solving,” a powerful practice is to temporarily adopt “nothing needs to be done” or “what if there were no problem to be solved” as experiential stances—this might, in fact, be the most powerful and central practice. Paradoxically, this “no problem” practice does, in fact, lead to progressive shifts leading towards a goal that can be achieved, a goal often adopted in response to practical problems, like, wow, my default psychology is super painful. It’s a mountain that you have to walk up backwards.

I am, in particular, a really big fan of “what if there were no problem to be solved” or (my preferred variant) “what if I simply didn’t have to solve this problem?” in response to really most forms of worry.

I think that’s for a few reasons:

• The hypothetical framing lets me not have to immediately adjudicate whether this is true or not. It’s sort of an invitation: what would life feel like if I simply didn’t have to solve this problem?

• It invites me to feel into the contrast: this is me, working the problem. This is me, not working the problem. The second state is mostly a lot more pleasant.

• It invites the question: is the experiential cost of working the problem worth the benefits, in expectation?

• Are there benefits?

• If I’m not sure: why do I feel like I need to work this problem? What underlying purpose is being served? What underlying, deeper problem is this instrumental to solving?

• What would it be like if that, also wasn’t a problem I had to solve? What if I decided to neglect it entirely? What would happen?

Some problems are somewhat obligatory (gotta make rent!) but I suspect the vast majority of unpleasant problems I spend my time noodling on, and most of the ugh-fields I develop around whatever topic, are basically just self-narrative-driven with no real payoff.

Oftentimes real problems that I am obligated to work (a coaching client has a problem I don’t know how to solve!) come secretly bundled with fake, impossible problems that can simply be declined (“I MUST PERFORM KNOWINGNESS AND HAVE ANSWERS TO ALL PROBLEMS AS THEY ARISE.”)

The true answer, as it frequently does, dissolves into: give precisely my model of the problem in precisely the way I am thinking of it, state exactly what I’m uncertain about and where my model does not provide actionable advice, and provide that mismatch as an open question which invites noodling on it together along with a tentative way forward.

Which if you push it a meta-layer up: it dissolves into process goals. Instead of trying to find the Good Future (or avoid the Bad Future) and furiously backchain from it, ask “what is the decision procedure one executes in response to this class of problem?” Then follow the procedure.

See Also

I was reviewing some code that Sean had written and I noticed that in his code he had computed an average (arithmetic mean) composed of three different, heterogenous measures of quality for [complicated thing that’s outside the scope of this essay.]

This number he had generated— let’s be clear— was entirely, necessarily meaningless, as meaningless as the sum of “five kilograms plus three dollars”, and he had added a column in our database to track it. I gently pointed this out to him and asked what the deal was.

He shrugged. “I dunno. The project manager asked for it.”

“…why did he want the number?“

He shifted uncomfortably. “I dunno.”

“But it’s clear to you the number is meaningless, right?”

“I guess so? The project manager did say he wanted it.”

And thus, I had to explain to him the thing that I had to have explained to me when I was an intern, and which I feel like a lot of interns have to have explained to them: that work is different from school.

School, see, has a very interesting property compared to the real world, a property that almost nothing in the real world shares: assignments cannot be wrong. That is to say, if you are given an assignment in school, your job as a student is to do the assignment as written, ideally without question. If you’re in the more forward-thinking sort of school you might be able to ask without rebuke “why are we being given this assignment” but make no mistake: the assignment and its rubric won’t change because you asked this.

That’s because assignments in school are devised for one of precisely two reasons: “so that you learn the relevant material” and “so that it can be graded, to rank you against the other students.” There is no other reason for any school assignment to exist.

And that means the specific form of the assignment is very nearly irrelevant. “Learn how addition works” is achievable by any series of addition problems; one addition problem isn’t really “better” than another. And for the purpose of grading, the specific assignment given matters still less: anything with a rubric will do.

That’s why even naturally-curious students lose the impulse to ask “why are we being given this assignment.” From their perspective, and from the perspective of the school, the question is pointless: one assignment is as good as another and a student asking “why this one” sucks up class time while never producing useful information.

Enter work

Now enter the working world.

Suddenly, assignments aren’t arbitrary. Frequently they’re underspecified; occasionally they’re entirely incorrect. That’s because the goal of the work assignment isn’t “turn this in for a grade at which point you the work product is hurled into a dumpster;” it’s “make money for the business” or “ensure we are in compliance for xyz regulation” or “ensure we are secure against cyberattacks.“

If your manager asks you to create a database and the thing he actually wants is to be able to stop losing track of inventory for unclear reasons, then the assignment he gave you— “create a database”— is correct precisely insofar as it lets him achieve his actual goals, and “checking the explicit spec against your manager’s actual goals” is one of the things differentiating senior engineers from interns.

That’s one of the most important things, in fact.

This cannot be taught in schools

Any time you are doing work that is for a distinct and non-interchangeable ultimate goal, you are not doing schoolwork. You are doing work. Alternatively, you might be doing a very complex hobby.

For that reason, I think people who mourn that students aren’t getting taught critical thinking in school are making a category error: to the extent that critical thinking is about “understanding the relationship between means and ends and what the right thing to do is in a given situation” as opposed to just “identify the logical fallacies present in this one-page opinion piece,” we’ve officially left the realm of school and entered the realm of every non-school domain in life.

We have 25-year-olds who can’t do basic forms of means-end reasoning because we suck up so much of their time in school-shaped learning that the first time kids ever have to do means-end reasoning is after graduation, well into adulthood.

I have no specific policy prescription here, beyond that “make schools teach critical thinking” is very nearly a category error.

Becuase school may be, in fact, the only institution that can’t teach critical thinking.

“But what about school-work hybrids where like you’re doing apprenticeship-style labor under a supervisor?”

I think the operative question is always this:

“Can the worker be dinged (even implicitly) for doing precisely what the supervisor told him, and not inferring that something different was needed?”

If the answer is “yes,” then sure: you have a genuine situation where you’re able to teach the meta-skill of evaluating the spec you’ve been given and checking it against whether it’s what is actually needed. I’ll note that the real world does give you a “yes” answer to this question: in the above anecdote, the intern did in fact get (very gently) told off for doing what he had been told, because the job spec of software engineer is not “do what you are told.” It is “satisfy the business objective.”

If the answer is “no,” then you’re just doing a very gussied-up version of the prototypical school-shaped job: “you did good if and only if you did precisely what you were told to do.”

“So… couldn’t a school just do that?”

Not generally, no.

Consider: students can appeal grades. They can do this because grades are meant to be “fair”, and “fair” means “consistent with what was announced in advance as the grading criteria”.

What you are doing, structurally, in work-shaped environments is that you are disregarding fairness in evaluating work outputs. “Good” is defined as “the job got done in a way consistent with the meta-goals of the job regardless of what the supervisor told you directly.”

Now compare this with school.

Each step follows directly: grades have to be fair → fairness requires criteria you’ve announced in advance, which applies to each student individually → therefore the evaluation has to be according to these criteria → therefore the criteria cannot be wrong.

And why does school have to be fair? Because schools are institutions where a primary purpose is evaluation, and anything with that primary purpose will end up school-shaped in this way.

You can see this operate from the other direction in the workplace: all the megacorporations I’ve worked at— Microsoft, Amazon, Expedia— have crafted elaborate pseudo-objective grading rubrics for precisely one part of their work environment: promotions and performance evaluation. This is, not coincidentally, the part of the job that is the most obviously farcical: almost none of what engineers do can be objectively evaluated, so these rubrics mostly serve to launder a huge variety of subjective judgments made by the engineers and managers on the promotion panel. But large companies sense that the appearance of fairness is deeply necessary in evaluative environments, which is how these rubrics get devised in the first place.

(Notably, the reason this is allowed to be clearly a farcical veneer over subjective judgment is because at work there is no appeal. You can’t drag your promo panel in front of an adjudicator and be like “PER CRITERION 3B I DEMONSTRABLY HAVE MET THE BAR FOR SENIOR ENGINEER.” The rubric structures the conversation but it doesn’t bind anyone. Which is good because I think basically everyone at these megacorporations knows that objective promotion criteria can and do get gamed: anyone remember that brief period where giant companies were rewarding employees directly for ai-tokens-consumed? Extremely funny, very predictable results, no notes.)

Of course, that just means the thing that everyone already knew: promotions are, always and everywhere, popularity contests. The primary difference between healthy and toxic work environments is whether popularity is meaningfully connected with how well you helped achieve the organization’s objectives.

“But what about assignments in school that aren’t graded? Couldn’t those be work-shaped?”

Okay, yeah, you got me. Those are exempt and can be work-shaped.

(Good luck getting the students to do them.)

What is guaranteed in a therapist?

If you hire a therapist, you know several facts about them already, because “therapist” (and the thing they do, “therapy”) is an extremely specific legal category. The license any therapist will have tells you these things:

They have gotten a relevant certification + associated training

There are something like five of these. Licensed psychologist, LMFT, LCSW, LPC, etc.

They will, as part of receiving this, have gotten something around 2000-4000 supervised hours with patients depending on jurisdiction.

Note that the term “supervised hour” is itself a bit of a term of art here. The typical minimum ratio for the supervision is one hour per week of seeing patients, which is to say, of perhaps 30 clients the therapist-in-training might talk to in a week, they might discuss 3 of them in detail with their supervisor over the course of an hour. This implies a certification guarantees perhaps ~100 hours of expert consultation on ~3000 hours of actual practice, devoted to some mix of adminstration, liability management, and actual therapeutic technique.

They are accountable to a board for misconduct

You can report therapists to a licensure board, and they can revoke the therapist’s license depending on whether they’ve engaged in egregious-enough misconduct. This includes stuff like sexual misconduct (no fucking patients!), exploitative dual relationships (e.g. entering business deals with clients), breaches of confidentiality, or failures of mandatory reporting.

They can bill insurance and take Medicaid

If you have insurance, therapists can bill it! It generally does not pay them well; this is especially true if the therapist takes Medicaid, which gives Chipotle-service-worker levels of hourly compensation once you take into account no-shows and time spent outside of session.

Guaranteed you-can-sue-them confidentiality

There are laws with teeth that the therapist won’t talk about your shit to other people. Law enforcement itself (with the exclusion of mandatory reporter obligations) cannot force a therapist to talk about what his patient has told him.

These are all real guarantees, and they matter for a lot of people, especially for the emotionally vulnerable and the money-conscious.

What is NOT guaranteed in a therapist?

That the therapist is doing any specific form of therapy.

Most therapists practice some eclectic blend of therapeutic modalities: a dash of CBT here, a bit of ACT there. Supportive talk therapy, freed of any specific modality or theory-of-change, is quite normal. (I don’t recall my past therapists doing anything that resembled any specific therapeutic modality at all.)

Effectiveness at treating any specific condition

I suspect eclectic supportive talk therapy is the modal therapy situation, and in terms of effectiveness, “supportive conversation from a person with a credential” is likely to be similar to getting supportive conversation in any other context.

Note that “years of experience” and “type of credential” very weakly predict therapist effectiveness.

Results demonstrated a small, but significant relationship between therapist experience and internalizing client outcomes. There was no relationship between therapist experience and outcomes in clients with primary anxiety disorders. In samples of clients with primary depressive disorders and in samples of clients with mixed internalizing disorders, there was a significant relationship between experience and outcomes.

Famously, claimed modality also doesn’t predict therapist effectiveness (the famed “Dodo Bird Verdict.”)

To be clear, it matters enormously which therapist you end up with! Basically all the literature I’ve read agrees the best therapists are much much better than the worst therapists. It’s just not predictable which therapists is which using years of experience, credential type, modality, or anything else easily legible to prospective patients. Even star-based patient reviews are likely to systematically overvalue warmth and rapport as opposed to “did this person actually resolve my problem.”

Wait, why wouldn’t years of experience predict therapist quality?

That is because people frequently assume that “time spent doing a thing” equates to “practice” and it’s not. “Practice” requires that you have a feedback loop where you can discover what you did that worked and what you did that didn’t work, which allows you to improve over time. Years of experience will predict quality in only those fields where there is a feedback loop: where the practitioner is monitoring things and seeing what works and what doesn’t.

However, there’s no specific mechanism by which therapists are required to do this, especially after they’re trained: therapists have to rig up their own outcome monitoring practices and devote time and effort to feeding that information back into how they interact with patients. This takes time and effort and therapists are, frequently, starved for both time and energy. So they might do this, they might not.

There’s a saying in software-engineering-land: that there’s a difference between “ten years of experience” and “one year of experience, repeated ten times.” I think that’s kinda what we’re seeing here: therapists exist of both types and you won’t be able to see which kind they are from a “years of experience” blurb.

Caveat Emptor

That means, unfortunately, that we are in a similar place with therapy as any other small service business in capitalism, like contractors or auto repair. Some therapists are quite good! Some are basically not giving therapy in any except the strictest regulatory sense. Years of experience and credential type don’t really tell you to what degree this will be true for a given person.

A pretty strong implication is that you should have a good idea of what you want from therapy before you arrive for your first session. Do you basically want somebody supportive to dish with once a week? Okay, valid. A support structure while you break up with your abusive spouse? Excellent, carry on.

But if you’re looking to deal with a specific psychological problem you want treated, like anxiety or depression or ptsd, I think it is reasonable to figure out jointly with the therapist what a reasonable timeline for improvement looks like and what their stated mechanism of change for that would be. And then figure out, in advance, what alternative approaches look like if the current approach doesn’t pan out by some particular date. (I think it is very very common in both medicine and psych for the doctor or therapist to delay the intellectual labor of “figuring out a plan B for if this doesn’t work” for as long as possible; this tendency sucks.)

And then, if the timeline elapses and you haven’t improved along whatever dimensions you care about, do whatever the next thing in your flowchart was, which possibly involves finding a different therapist. Our goal being to avoid the traditional failure mode of “you’ve been trying xyz for a year so let’s give it a few more months” since that treats everyone’s time as incredibly cheap.

See Also

This is for a variety of reasons, most of them sympathetic. Rejection hurts; sometimes you’ll discover that by making a bid you made someone uncomfortable, which is unfortunate when it occurs but which cannot be prevented reliably, since the same thing that gets you information about whether a bid is likely to succeed is, itself that bid.

Reinforcement Learning From Hot Girl Feedback sessions are offered at a lot of Lighthaven conventions. A facilitator and a group of pretty girls stand on a stage, only one of which interacts with the male participant at any given time; it’s a little bit like one-way circling, where the girl expresses whatever it is she’s feeling in-the-moment, positive or negative. “I’m enjoying this,” “I felt attraction there,” “you feel like my little brother,” “I’m getting serial killer vibes.” Whatever she’s feeling.

I see a lot of behaviors in Reinforcement Learning from Hot Girl Feedback which I’d term as disapproval-avoidance behaviors. These all represent ways of avoiding making a legible romantic bid while still engaging with the event; the verbal channel must be occupied by something and the “something” men choose varies wildly.

I have seen the following, and I classify all of these as fundamentally avoidance behaviors:

1. Men doing things that are transparently silly and jokey in a way that is intended to read as unattractive: goofy flailing motions, exaggerated facial expressions.

2. Men talking about how weird and artificial the RLHF experience is. This is accurate-- it is weird and artificial-- but this is also beside the point.

3. Men ratcheting up the conversation to a higher level of abstraction, wanting to talk about the container itself or about whatever intellectual interests they have.

It’s like convergent evolution in ways to not flirt during exercises that specifically revolve around flirting.

The common thread is that these are all ways of not making a romantic-sexual bid that could be rejected. Someone called out (1) as simple self-sabotage (can’t feel bad about failing if you never really tried), which is a possible explanation but which feels less true to me than the explanation where they have substituted the correct-but-feels-risky behavior (expressing desires that could get them judged) with a behavior that feels-safe but isn’t really the point of the exercise (anything else.) Maybe it’s both, I dunno.

This tendency manifests in real life… mostly via that same kind of avoidance. It only registers in RLHF as avoidance instead of “ordinary social existence” because the point of the exercise is literally to improve at flirting: men love to take a format whose entire goal is expressing desire and then filling it with patter instead until they’re ushered off the stage.

I think these behaviors occur for really straightforward reasons. Making a pass at somebody is, whatever else it is, a status bid: you are implicitly claiming that you are the sort of person that is allowed to make romantic bids. You can see echoes of this fact in popular media: the smooth-talking jock who makes a pass at the female lead and gets told off is someone going through a very unpleasant status correction in real time. People hate getting downward status correctives.

The fear, I think, is that if you make a pass at the girl on stage then you’re setting yourself up to be humiliated in front of an audience. Which is true! This is a possible outcome, mostly because no matter how gentle everyone involved is being the activity itself is deeply vulnerable. They feel the sting of the implicit status-lowering from rejection; they feel the rush of status increase from Doing A Good Job. It feels like a high-stakes performance and the problem is that attempting to treat it as such is what generates the behavior stream that makes dudes fail at RLHF in the first place.

Which is a microcosm of romantic bids in real life. What fun!

The Theory Of Change

I think the girls who are running these RLHF sessions are doing god’s work; but I also think that when they critique specific surface behaviors they’re mostly missing the underlying generator. It reminds me strongly of when my old dating coach would run model work.

It’s still useful, though, because I think the more bites at the apple the guys get the more opportunity they have to crack the underlying principles involved.

See Also

When does active investing make sense? The responsible answer in the professional-managerial class— of which I am a part— is “basically never.” What, you think you’re smarter than hedgefundies who are valuing stocks as their full-time job? The arrogance. The unmitigated hubris. Besides, the Efficient Market Hypothesis has in its relatively strong form “the market has priced in all future inputs to a stock’s profitability, and therefore if you think you have identified such an input by noodling for ten minutes then it is almost certain it’s priced in.” Instead, one invests in a broad-market index fund where you get maximal diversification and (in expectation) lower volatility compared to an arbitrary-selected basket of fewer names.

But I (and the rest of the rationalist community) had, over the last few years, several formative experiences where some people in the community were like “hey isn’t that a hundred-dollar bill laying on the ground” and others were like “NAH PROBS NOT, EMH BAYBEEEEE” and then the hundred dollar bill was snatched up by a pigeon and then everyone was like “oh yeah I guess that was a hundred dollar bill huh, guess that’ll never happen again.”

Repeat 4 or 5 times and you have a pattern and honestly around time #2 a few years back I said “fuck the EMH I’m gonna do some conviction-based thematic investing.”

Edge

What is edge?

Edge is “non-consensus but correct beliefs about the future.” Crucially, it is not necessarily secrets— the theoretical arguments for the AI buildout were being publicly discussed for years prior to Nvidia exploding.

Anyway, in order to establish that edge exists for you personally, you have to (1) figure out what consensus beliefs about the future are, and why they are held, and then (2) you need to know why they have substantial likelihood of being wrong. The Ideological Turing Test is, I think, pretty fundamental investing technology.

What is my edge? Well, I do know a lot about social anxiety as a disorder. Not super investable. More relevant is probably just that I am a part of the rationalist community and have the appetite to read long posts by e.g. Gwern and friends.

The rationalist community has a very good track record of being very early to things. The following is a list that influential members of the community were pushing as huge deals many years ahead of the mainstream:

• AI and the scaling laws (he said so, so obviously.) See Gwern in 2020.

• Covid

• Bitcoin

• GLP-1s.

• Various altcoins during the post-covid cryptocurrency boom (oops, this one may have been a miss)

• Effective Altruism (not really investable but still very important)

Of course, “AI is a huge deal” is now more-or-less consensus; what isn’t consensus is the ripple effect that’s likely to have throughout the economy since Gary Marcus types still perceive AI as being mostly an unproductive curiosity, of interest only to tech nerds, and Serious Economists view AI as “mildly positive in expectation, nothing special” (Acemoglu to my knowledge still has not revised his opinion that AI will produce an increase of ~0.5-1.5% GDP over the span of a decade.)

More-generally, rationalists like to extrapolate trend lines on graphs and are willing to do so even when this results in insane-sounding outcomes.

The Characteristics Of Fake Edge

Quoth HPMOR:

Mr. Malfoy is new to the business of having ideas, and so when he has one, he becomes proud of himself for having it. He has not yet had enough ideas to unflinchingly discard those that are beautiful in some aspects and impractical in others; he has not yet acquired confidence in his own ability to think of better ideas as he requires them. What we are seeing here is not Mr. Malfoy's best idea, I fear, but rather his only idea.

I’ve had good investment ideas; I’ve had bad investment ideas (oh, Ethereum); I’ve had investment ideas where they were quite close to being good ideas but then weren’t.

Guess who bought puts on Shutterstock on the strength of AI image models just before the stock spiked for unrelated reasons? Expiring, of course, when the puts were still out of the money. RIP.

As they say: “right on direction but wrong on timing” is, in finance, synonymous with wrong. Hubris!

More specifically I think the problem here was that I had a thesis— “this industry is structurally doomed”— but puts are mostly kinda mediocre as speculative instruments honestly? Especially if (as in my case above) the market for SSTK puts didn’t really exist for anything more than six months out, since if you’re holding onto puts that close to expiry you’re bleeding out the put’s value extremely quickly (“theta”). And shorts are even worse: you might get to profit on those eventually but you can’t eat “eventually,” especially not if you get margin-called and your account liquidated.

I think on some internal level a lot of why I ended up making the Shutterstock bet regardless was because it felt too clever to not make. Of course, I’m not so much a fool as to burn a huge chunk of my portfolio on that kind of bet, but I lost thousands of dollars regardless.

Lessons! Many lessons.

Where does the rat community still have non-consensus views?

• The Strong AI Thesis: AI 2040 came out recently; its main purpose is as a policy whitepaper advocating for a global slowdown treaty. It also has an economics supplement, where it points out that conventional economists (Acemogle, Goldman Sachs) view AI as a fundamentally less impressive technology than the writers of AI 2040 did; and I accord the writers of AI 2040 significant credibility since they not only are the same people who did the quite prescient AI 2027 whitepaper but also were of the community that predicted the scaling hypothesis and its implications as far back as 2020.

  • The Strong AI Thesis also predicts the obsolescence of most white-collar labor that isn’t legally required to be done by a human in the next few years. In a brute structural way the most valuable asset, by far, that any white-collar worker owns is his years remaining as a viable worker. If Strong AI makes the value of this asset go to zero, AI investments may be required as a hedge.

• IVF and polygenic screening technology. PGS has gotten really, really good. You can (at the cost of doing IVF + PGS) avoid having the worst of the genetic disorders today and once in-vitro gametogenesis takes off you’ll be able to cheaply sequence hundreds of embryos which can then be selected between for health and (more controversially) IQ. This will also allow for not doing the expensive and inconvenient IVF egg extraction process. IVG doesn’t yet work today, but Metaculus claims the first human baby from IVG is likely to come in ~2032. Which, uh, is somewhat after our AI thesis. Also the IVG companies are all private, so this is mostly non-investable anyway. Even so!

• AI Will Likely Kill Us All. This is noninvestable; or rather, this might be investable in the sense that a large chunk of futures where a massive AI buildout happens humanity perishes as a result, implying that some unknown percentage of “AI explodes” should be totally devalued in terms of expected portfolio value.

What Remains Of The AI Thesis?

Or equivalently: “where does the rat community still hold non-consensus views as regards AI, and what are the investable differences between the market-projected predictions about the world and the rat-community predictions about the world?”

• Interest rates due to AI-driven productivity. Most versions of the big-AI thesis involve productivity exploding, which drives the present value of money through the roof, which implies interest rates have to explode upward accordingly. Which implies somehow going short long-duration bonds, either via direct shorts or via long-dated OOTM puts on (probably) TLT. This is probably the most direct and within-community consensus trade, though it has problems in the possible worlds where he have a horrible non-AI disaster and the world goes tits-up since that is the world bonds are intended to hedge against. Which probably implies OOTM long-dated puts on TLT instead of shorts.

• Broad-based economic productivity from AI translating into general stock price increases. The flip side of the interest rate trade is “long-dated OOTM calls on the S&P 500”; SPX has them up to five years out, which is nifty. The market is pricing such calls as lottery tickets if you crank them OOM enough; the argument that can be made is that they are exceptionally well-priced lottery tickets. The main problem: if interest rates skyrocket then the S&P 500 gets crushed. Ooops.

• The current AI trade, but moreso. Semiconductor supply chains, power generation (especially portable power generation). This is already close to mainstream consensus, but if you take Ratworld seriously then likely then it’s almost certainly not consensus enough. At least over long enough time scales; note that AI 2040 buildout projects don’t really stop us from having an 2000-style economic meltdown such that current companies get hosed even as the underlying technology scales up as expected.

• Megacorporations firing most of their workers and pocketing the difference. This mostly accrues to preexisting big tech players, ignoring business models. I’m fairly confident this occurs but less confident about whether this has a huge impact on prices such that it’s attractively investable; on the other hand, “invest in QQQ” has a fairly low burden-of-proof on it (also QQQ is strongly correlated with SPY which makes sense given that it is most of SPY), and that seems like the natural instrument for this sub-thesis.

• Humanoid robotics. The better and cheaper AI gets, the more you want AIs to be able to do stuff in the real world. And the world is, mostly, built for things that are human-shaped.

• ….land? They’re not making more of it. Regrettably annoying to invest in directly unless you’re literally buying a house, and houses are illiquid and lumpy goods which also imply onerous maintenance requirements. Bawwww. That said, there are companies out there that mostly hold timberland and other low-value property.

What are the arguments that the rationalist community is just wrong about this stuff at a broad level?

I did, after all, point out the Ideological Turing Test bit above so I really should be able to answer this at a reasonable probability level.

Leg 1: AI Isn’t Good At Stuff

It hallucinates frequently; many of my techie friends hold the strong opinion that AI is not good at even coding stuff and is likely to never be.

Additionally, most jobs have as a crucial component that the person must be able to learn the job over time. AI right now is like a genius savant with amnesia: great for discrete tasks, terrible for any project that has to last longer than a context window.

I note only that feedback loops are getting better and better for AI stuff. “Make my website” followed by “break my website, then fix the breaks” works pretty well right now; God knows what it looks like a year from now. There is at least one good and useful website I know of (Glosso!) that was entirely vibecoded.

And if the “amnesia” part is solved somehow, that either leads directly to economic boom or to everyone dying. Or both!

Leg 2: AI Companies Are Super Leveraged Right Now And Thus Are Vulnerable To Economic Shocks

What if Trump starts another war? What if this is structurally similar to the dotcom bubble— where the tech was real but everyone was just too levered up and it all topples over?

The stock market can be resilient but is not automatically so. Even if AI is a real technology that upends the world over the next few years, that doesn’t help anyone if the major players are going bankrupt left and right, leaving the data centers and GPUs to be bought up at pennies on the dollar by whoever is left.

I think this critique is basically accurate and is why my portfolio isn’t stuffed to the gills with SOXL and TQQQ.

Leg 3: TSMC getting destroyed in a war sets the world economy back by a decade or something

Isn’t it kinda ghoulish to bet on stuff you think might contribute to ending the world

Probably! I basically stand by my Three-Futures Model here:

That is: you have the

• Future Where Gary Marcus Is Right And AI Is A Bubble And Can’t Be Used To Do Stuff,

• Singularity-Future (preceded by massive instability and likely economic boom)

• Non-AI Disaster Future where AGI isn’t a thing because we couldn’t get there because the relevant labs have been destroyed for ~unrelated reasons. See also: war with China.

• Pause AI Future (just added!) where people get freaked out enough to SHUT. DOWN. EVERYTHING. Nvidia hardest hit.

• Some kind of hybrid future where AGI is created and is hugely relevant economically but money is still a thing.

I’m mostly discounting the first Gary Marcus Is Right future’s probability because [waves hands at current events]. The singularity future, meanwhile, is basically only investable in the short-term.

Investing in AI stuff, I think, kinda requires that hybrid future if it’s to be a good idea: where AGI is relevant, but we all live long enough to collect on our bets. But it’s still important to note that future #3 (non-AI disaster, including financial) is still a live option which is why I have a fair amount of portfolio ballast, mostly in short-dated treasuries (cash-adjacent!) and gold.

Really taking the ASI thing dead seriously might mostly imply an extremely boring portfolio that has an upsettingly large amount of gold and cash, living alongside megacap+AI-exposed equities as a hedge against AI-driven worker obsolescence, living alongside a small tranche of long-dated put options on long-duration treasuries? Wow, there’s a downer of a thesis to end on. I was hoping for a get-rich-slightly-quick scheme as opposed to a ride-out-doom-or-the-singularity-with-hedging-against-career-destruction scheme.

Even so.

Anyway! Bet-sizing is, of course, equally or more important than individual stock picks; for that I have hunches and modeling and prayer. Frankly I still go back and forth on whether I am being too aggressive or not aggressive enough in my allocations, which I understand is traditional.

Godspeed. Not financial advice.

“Abandon hope” is a hard instruction. If you don’t know the mental motion involved I suspect my instruction to do so is, uh, similar to in this parable:

“Sir. Imagine you are in the driver’s seat of a car. You have been sitting there so long that you have forgotten that it is the seat of a car, forgotten how to get out of the seat, forgotten the existence of your own legs, indeed forgotten that you are a being at all separate from the car. You control the car with skill and precision, driving it wherever you wish to go, manipulating the headlights and the windshield wipers and the stereo and the air conditioning, and you pronounce yourself a great master. But there are paths you cannot travel, because there are no roads to them, and you long to run through the forest, or swim in the river, or climb the high mountains. A line of prophets who have come before you tell you that the secret to these forbidden mysteries is an ancient and terrible skill called GETTING OUT OF THE CAR, and you resolve to learn this skill. You try every button on the dashboard, but none of them is the button for GETTING OUT OF THE CAR. You drive all of the highways and byways of the earth, but you cannot reach GETTING OUT OF THE CAR, for it is not a place on a highway. The prophets tell you GETTING OUT OF THE CAR is something fundamentally different than anything you have done thus far, but to you this means ever sillier extremities: driving backwards, driving with the headlights on in the glare of noon, driving into ditches on purpose, but none of these reveal the secret of GETTING OUT OF THE CAR. The prophets tell you it is easy; indeed, it is the easiest thing you have ever done. You have traveled the Pan-American Highway from the boreal pole to the Darien Gap, you have crossed Route 66 in the dead heat of summer, you have outrun cop cars at 160 mph and survived, and GETTING OUT OF THE CAR is easier than any of them, the easiest thing you can imagine, closer to you than the veins in your head, but still the secret is obscure to you.”

The instruction to “GET OUT OF THE CAR” is, in this parable, a remarkably useless instruction, because it does not break down the mental motion into its constituent parts.

I aspire, here, to at very least gesture at the car door handle.

The Core Problem With In-Person Exposures

The feedback loops are agonizingly slow.

Like: I’m a Greenberg stan, as you may recall. Which means the core thing exposures are meant to do is teach a mental motion— the mental motion of noticing and cutting off the problem-solving impulse which constitutes anxiety— rather than “suffer through a stressful experience.” The exposure to the feared situation is basically only useful insofar as it’s able to summon up the compulsion to reason about how to fix it, which is why (in this framing) they ought to be very low-grade.

But if someone gets to the exposure, attempts to ABANDON HOPE, and fails (as indicated by the continued presence of stress) then there’s not a lot for them to do but go back and schedule another call, which sucks, because now we’ve burned like two weeks and haven’t accomplished anything beyond finding one thing that doesn’t work. Which isn’t nothing but it’s also not, y’know, great.

This dynamic shunts us toward exposures which have to happen in-call. There’s a few different things I’ve been trying out in this week’s client meetings:

Imaginal Exposures

So. Our goal here is, ultimately, to teach you the mental motion of unclenching: letting go of control over outcomes. As I’ve discussed previously, I think this both involves disclaiming control over the feared outcome, and providing something to take its place. I’m trying to go deeper into that here.

The idea here is that you think of a situation you’re habitually anxious in: like you’re just about to cold-DM someone on LinkedIn, or just about to cold-approach a girl at a bar.

The idea is we’re going to now freeze you in the moment of time where you’re considering doing the scary thing. Nothing about the situation will change; our goal is only for you to try different mental framings around it to see which ones activate which sensations in your body or mind. The idea is that we have you practice inhabiting the relevant feelings— the feelings you’d prefer to have in the situation, which ideally feel true enough to you to actually inhabit.

Identifying the starting intention

I’m going to use “I want this person I’m cold DMing to think I’m worth their time.” This actually works in both personal and professional contexts, and crucially, it’s almost maximally uncontrollable: the reasons why someone might not respond to your cold DM are as vast and endless as the sea.

Disclaimed Intentions

First: we need to disclaim your responsibility to control the outcome of this situation (whether or not you get a positive/negative/no response). Situations are, mostly, uncontrollable; this is the human experience; our goal is to make you viscerally feel that rather than merely understand it intellectually.

I’m going to suggest a few framings, in order of most sweeping and most likely to be flagged as ‘lies’ to least sweeping and most inarguable.

This person thinking I’m not worth their time is an acceptable outcome. This is the best option if you can build a felt sense of it being true. Equanimity in one shot, if it works.

“Them thinking I’m not worth their time” is not a problem I am obligated to fix. This is a weaker statement than the above; a situation can be bad (thus you feel like saying it’s “acceptable” is a lie) but also not obligate you to solve it. If a meteor is going to crash into the earth in a year then this wouldn’t be acceptable as such but devoting cognitive effort to preventing it would be pointless waste.

The bad outcome is a problem I am not going to spend effort in preventing or fixing. “I am not obligated to fix this” is ultimately still a factual statement in some loose sense, which can be argued with. Here we are doing away with the factual statement; regardless of whatever else someone might think about your being obligated to do this tedious job, you are declining to do it nevertheless. This is only true or not-true by your actions; you can make it true.

“None of these sound true to me.”

Okay, here’s what I’m thinking.

You’re reading this essay and doing this exercise for a reason. There is clearly some sense in which you’re aware you’ve chosen a poor tradeoff on the Anxiety Production Frontier because that’s why you’re doing all this in the first place.

The jiggling around of statements, in this framing, isn’t persuasion as such. There’s a faction inside you that already notices that doing an IRL version of Nathan Fielder’s The Rehearsal hasn’t actually gotten you any of what you wanted. On some level this has always been a waste; you already knew that.

The goal of the above exercise is to find the wording that is, for you, literally true: where we are making a concrete statement about where we explicitly state that the tradeoff we have made kinda sucks, and we are now going to decline to make it, and then we are going to enumerate, in detail, what it would concretely look like to stop making it. The reason we’re hunting for a specific statement that feels true is because I can hardly expect you to act on one that’s false, and a true such statement definitely exists because that’s what got you in the room.

The “click” sensation I’m waiting for you to notice is one of recognition.

“But sometimes I can improve outcomes by trying really hard.”

This is true.

Often it’s not— most live-fire social situations, for instance, get much much harder to deal with the more system-2 cognition you recruit in an attempt to achieve specific social outcomes. But sure, possibly micro-editing a DM for hours would in-expectation make the other person respond better.

The main problem with this line of reasoning becomes clearer when you zoom out: men who endlessly scheme about individual dates and messages and women are, almost universally, terrible at dates and messaging and interacting with women, because their anxious vigilance results in them actually going on very few dates (which they then pre-emptively overplan enormously, a la Nathan Fielder’s Rehearsal.

People who vigilantly ruminate about all their networking experiences do very little networking and thus are very bad at it. “Getting reps while terrible at the thing” is, always and everywhere, what makes you good at the thing. You can rubber-stamp this basic pattern on almost all the domains people are anxious about: they minmax their attempts at winning individual battles at the cost of losing the war.

The question is not whether you can improve your results in one interaction (the answer is, always and forever, “maybe”), it’s whether we can improve how you do in the aggregate by changing how you treat these events as a class.

If you like, we can frame this as an experiment. Try letting go of outcomes (via doing the above in any situation where you feel anxious) for, perhaps, two weeks, or however long it takes you to get perhaps ten or fifteen reps under your belt. Set a time on the calendar, and prior to that time be utterly monastic about not doing anything to influence social outcomes positively in your direction. Be merciless. Uproot the impulse root-and-branch from your psyche, and act only upon consciously-endorsed values which do not refer to others’ perceptions of you.

Then, after that two weeks, look back. How was it?

Noticing Tension Stored In The Body

If a statement above intellectually seems clearly true but doesn’t feel true in your gut, I suggest holding it in your mind while scanning your body for any place that you’re holding tension.

And while holding the statement of anti-intention in your mind, relax that body part.

(This is basically as woo as I get. The reason I ask you to relax is because I suspect that some part of interoceptively believing the statement of non-intent involves your body inhabiting a posture that doesn’t imply readiness to leap into action.)

If you’ve chilled the fuck out we can move on to positive intentions that are to take the place of the anxiety-producing outcome-grasping.

What-To-Do: Trying On Intentions

Framings which sometimes work are:

• Experimentalist-Curiosity: “I wonder what will happen if I [send this DM]/[say hi to this girl]”

• Factual: “I am going to [send a true DM that, by my standards, is a quality message]/[say ‘hello’ to this girl] and they’ll say what they say.”

• Prosocial: “I am going to [offer this person the opportunity to engage with me professionally]/[offer this girl a genuinely-felt compliment], and they respond how they respond.”

• Brute-Force: “My only goal is to say literally the minimum viable greeting that gets me into a conversation with this person or group (probably it’s “Hi. I’m Aaron.”) If I have done that I have succeeded and I am not going to lift a single finger to make the rest of the interaction go well.”

(The last one probably sounds weird but I think for me it’s the most-reliable option for accessing equanimity in the starting-conversations domain? I think because it is mutually-exclusive with any of the scheming I instinctively want to do around starting conversations. It might be less-applicable for domains without a natural brute-force action available.)

If you’re still stressed, try (again) noticing where there might be tension in your body: some kind of somatic, body-related indicator that you’re still trying to do a thing that isn’t just “blandly accept the outcome.” Then try releasing that tension while also holding the given intention in your mind.

At some point, it might or might not “click,” providing some felt sense of “this can be a true statement.” You’ll know if it’s clicked because you no longer feel intense stress and resistance at the thought of doing whatever the thing was you were anxious about. If it feels like a lie then we have not yet succeeded; you need to jiggle things around a bit more and possibly use a different framing. (“Jiggle things around a bit” is where I, too, can merely point at the door handle.)

When you’ve got this down once, the rest is (as they say) reps.

Outcome Equanimity Training

Here is a similar imaginal exercise.

You imagine your feared outcome— the LinkedIn person doesn’t respond and secretly thinks you’re a fool, or the girl at the bar says “nah” and turns away.

The idea is that you say the words (or at least hear internally the words) “that is an acceptable outcome,” or whatever the anti-intention statement you arrived at above was. And you basically jiggle things around in your mind until “that is an acceptable outcome” becomes a straightforwardly true statement. Because… it kind of is acceptable? Or at least, it is an outcome that cannot be reliably prevented and “not accepting outcomes that cannot be prevented” is foolish by any standard.

Because, of course, “that is an acceptable outcome” is utterly bloodless and indicates merely a lack of inclination to modify the verdict or indeed to do anything about it whatsoever. “That is an acceptable outcome” is a phrase denoting equanimity. “Her loss”, by contrast, is different: it accepts the rejection only by re-litigating whose judgement was actually correct, and this litigation is forbidden cognition: its only purpose is to impress an imaginary jury, which is basically just social anxiety turned inward.)

There are a whole zoo of rejection-responses which are basically fake equanimity, structurally identical to “her loss.” Such as:

• “They’re probably not hiring anyway”/”she probably has a boyfriend”: This wasn’t really a rejection because reasons.

• “Everything happens for a reason:” Demanding compensation; the outcome is accepted only in exchange for some narrative or developmental payout.

This won’t mean rejection won’t suck, of course. If I stub my toe then this will be painful; if I spend hours ruminating about how to avoid stubbing my toe I will, crucially, still not avoid stubbing my toe but I will have now paid a substantial cost that’s totally distinct from the toe-stubbing and very likely to have been worse.

Narratives Considered Harmful

I want you to also notice that internally you might have a rejection-narrative rolling: a thing that asks what it means that you were rejected. The above items are, basically, answers to this feared rejection-narrative, and narrative meaning-making around rejections represents compulsive rumination effort that should be killed when you detect it. Rejections are quite normal and could mean literally any number of things.

The important thing is that you have acted according to values you endorse.

Wait, but what about learning from my mistakes?

To the extent you can learn things in terms of object-level social skills, you will mostly be learning them implicitly from reps. Things that tend to go well, things that tend to go poorly: it’s hard not to notice these things if you’re paying attention, and the sin of the socially anxious person was never not paying attention.

If you conclude you did not act from values in a situation, I encourage you to run a structured postmortem, derive action items, and then let it go without attempting to transform it into a self-narrative.

Self-Application: Cold Approaches

A few days ago I was kicking around the idea of doing cold-approaching again just for the sake of doing test-runs on my hypotheses re: achieving equanimity, since I never did get to where cold approaching felt good or fun; but it’s been at least a year since I last did any cold approaching whatsoever and I have significantly better introspective access now. A friend of mine helpfully piped in on Discord that she’d like to join me, which sounds like fun (also is a helpful precommitment mechanism since this is 100% the kind of thing I procrastinate on); she and I will be doing that in a couple of days during a bar crawl if we can actually get our schedules to align.

Anyway, figured in the meantime I’d do some imaginal exposures and probably the hardest things concretely about doing them is resisting my deeply-felt impulse to simulate how things go. Like: the Curiosity framing for me fails here because I keep trying to evaluate the things that I’m curious about to see whether they function as effective openers and whether the hypothetical imagination-girl would react well; accordingly, this is terrifying. I find myself critiquing the things I want to ask about for how likely it is to land well.

Vulnerable Honesty/Selfish Impulse is similarly colonized here, I think because both of these contain degrees of freedom: there are a large number of true things one can say to open a conversation at a bar. Almost every smoother-feeling approach has the same core problem, which is that it invites the grasping impulse.

The only thing that’s actually worked for me even in my imagination is the brute-force approach: my job is to say, literally, “Hey. I’m Aaron” and then jesus take the wheel. (Is this optimally effective behavior? Oh, almost certainly not. It’s barely usable. That’s somewhat the point!)

I note this probably rhymes the most closely with how my very anxious past-self achieved zenlike calm in his dating workshop; the exercises were well-specified and the correct responses per the exercise were overdetermined (say the true thing, let them hate you); the only thing that ever made them feel difficult was the strongly-felt impulse to manage the models’ reactions to these true statements, which was not endorsed by the coach, and dropping this made an extremely narrow behavioral specification which I think was the active ingredient.

So it is here, I think. I’ll let you guys know how it goes!

fin

[Also, extremely gratified to see all the interest in coaching! What fun.]

Here’s what I think: mindfulness-meditation-style attentional training might be necessary as a prerequisite for at least some people to do effective Exposure Therapy and Response Prevention in anxiety disorders.

A problem I’m tackling right now is that a lot of people do experience their anxiety as a sort of unpleasant somatic fog, not as a set of implicit propositions that can be accepted or rejected. Which means they’ll white-knuckle through exposure after exposure without really learning the mental motions involved in not grasping-at-outcomes. That’s bad.

My model (which you’ll know if you’ve been following me for a while) is that this felt sense of “my anxiety is a diffuse black-box of uninterpretable emotion” is false, and importantly so; and that the problem here is actually just lack of introspective access to their own felt anxious impulses.

I’ve been reading up on various methods of achieving this, and I think I’ve found a few that seem deeply relevant for what we’re trying to do here. Namely, the Trace and Surrender—“response prevention”, if you’re from CBT land— require access to a few capacities. Here they are and the exercises I suspect function acceptably as prerequisites:

Tier 0: Figuring Out Your Root Obsessions

Read through the Obsessions list in Part 2.

The Obsessions list is likely to have at least one item that you strongly identify with. For some men it's creepiness-avoidance; for many people (men and women) it’s awkwardness-avoidance or cringe-avoidance; many women experience make-sure-I-don’t-look-weird-or-ugly as a deep, intense felt imperative.

Collect whichever items feel relevant to you.

Tier 1: Hypothesis Testing On What Anxious Problem You Were Solving

You were, earlier, very anxious in some situation. You’re not exactly sure why.

This is about reconstructing anxious cognition in the past. It has two steps.

Step 1: hypothesis generation

You ask: “suppose I was analyzing a stranger wearing my skin, who did exactly the things I did. What, concretely, did I do? What are possible things that stranger might have been feeling, and what urges might have been driving them? What goals might they have held that make their actions rational-feeling in the moment?”

It is important here to treat yourself as a sort of exotic beetle: you’re not here to praise or condemn what you were doing. You’re here to understand why you were doing it.

Go down the list you wrote in Tier 0. It is close to axiomatic that the behaviors you were engaged in were in the pursuit of at least one of these differing objectives; select the one or two options you generated above that feel the most true. Sorry, this is extremely vibes-based.

At this point, we have finished and have done something between “confabulation”, “rationalization” and “explanation.” There is a folk belief that this is a skill that develops over time— if you’re going at it with scientific curiosity, then you’ll probably notice patterns in the explanations you gravitate toward and be able to identify those impulses in the moment of firing. This is basically a specialized version of a practice called “Gendlin’s Focusing”.

Tier 2: Mindfulness Meditation + Noting For Developing In-The-Moment Introspective Access

Mindfulness Meditation

I suspect this problem is due to inability to decouple “having a thought” with “engaging with the thought.”

Here, I’m going to steal an exercise from Acceptance and Committment Therapy:

For this exercise, begin by getting comfortable in a quiet space and closing your eyes. With practice, you will probably be able to do this exercise during your daily activities.

Once you are comfortable, visualize a parade of tiny soldiers marching in front of you. Each soldier is carrying a sign, and each sign has one of your thoughts written on it. Each new thought goes on a sign in this never-ending parade of tiny soldiers. The signs can carry words, images, even sounds and voices. Whatever your mind produces can go on a sign.

If you prefer, your thoughts can float by on leaves flowing in a stream, as clouds in the sky, as credits on a movie screen, or as widgets on a conveyer belt. What matters is that you imagine watching your mind’s activities from a distance as the thought parade goes by.

When you notice that you’ve forgotten what you’re doing and you become attached to a particular thought, simply climb back into the bleachers and let the parade resume (adapted from Hayes, Strosahl, and Wilson 1999).

That’s the easy part. The hard part is in simply observing your thoughts without trying to change them or make them go away. It may help to remind yourself that none of the thoughts are facts, even if they seem compelling or if it seems that you must do something with them (like make the parade go faster so a particular thought will disappear).

You may also notice that you’re experiencing judgments about the thoughts. I shouldn’t be thinking that or Only a crazy person would have that thought. Take those thoughts, put them on a sign, and add them to the parade. They are not facts, and you need not respond to them.

The dude who wrote this claims that doing this for maybe ten minutes a day for a week or two should open up your ability to decouple yourself from the thoughts that you are having in-the-moment.

Shinzen-style “See Hear Feel”

Helpful PDF here.

The core of See-Hear-Feel is that you spend some time doing some otherwise-mindless activity (walking, driving) while running a specific metacognitive exercise. Every five seconds or so, you note and speak aloud (subvocalized is fine) what kind of thing your attention is oriented toward. The categories are See-Out (external visual stimulus), Hear-Out (external audio stimulus), Feel-Out (body sensation caused by specific external thing like the cold), See-In (internal mental image), Hear-In (thoughts you hear as voices or sounds), Feel-In (gut feelings not obviously caused by specific external stimuli.)

There’s a bit more to it (see the PDF above), but this is the core of the exercise.

This is somewhat unlike the above Thought-Parade practice in this doesn’t involve you sitting still. You’re doing stuff, but you’re jamming a shim in between any mind-event that comes up and your engagement with it; before engagement you’re expected to have Noted it, and this Noting is what’s expected to improve ability to refuse-to-act on that mind-event, because the Noting is what trains you in recognizing your internal senses as mental events in the first place. Can’t refuse a demand you don’t notice as a demand.

The mechanism by which this works is, I suspect, basically the same as the ACT-style mindfulness exercise: sustained metacognitive attention is more than anything an active habit, and habits can be trained, and having a certain amount of this one allows you to progress to situations demanding more of your attention.

Tier 3: Developing Introspective Access In Mildly Stressful Social Situations

So you’ve been doing the mindfulness stuff for a little bit. What now?

Well, I think you need reps running the same basic metacognitive skill in social situations, where most of the stray internal images and audio will be related to your feelings of social anxiety (or more concretely, your concerns about what other people are thinking of you in the moment, and your scheming about how to make them think better things.)

The rules here are the same as in tier-2: when thoughts or images cross your mind about how you appear, you are to Note them (“See In” or whatever) and cease engagement unless they actually deserve your engagement. Awkwardness-avoidance thoughts in particular are almost universally suspect.

I like the bar for this; you might be fine at the bar but have other social environments that feel (mildly) scary to you. Feel free to do this in whatever your low-to-moderate-stress environment you feel like doing it in.

The Bar Is An Inert Social Environment

If you’re at a bar, most of the time people won’t go out of their way to talk to you. They could do this; but they mostly won’t.

The Trace is hard to perform in active social settings because you’re actively being called upon to do stuff and engage with the world. Also, most social settings have somewhat higher perceived stakes: in the bar the stakes are, necessarily, nothing at all. You have no social capital riding on this. Necessarily, any fears you have revolving around existing in a bar are the products of compulsive cognition instead of anything real. Which is great because for this exercise you don’t have to be distracted analyzing “but what if the anxiety is pointing at something actually important,” which is a pretty big distraction for ordinary social situations.

However, you’re still in a social environment. People will observe you. They might actually form judgments about you in the privacy of their own head. And now we do the above Thought Parade exercise, except within a passive but mildly-anxiety-inducing social situation where you can be still for a few minutes without getting hassled. The core concept is that you’re just doing the exercise, but in this context (I assume) many of the thoughts you end up putting on the parade floats are thoughts about how you are being perceived. I ask that you label these thoughts as you go and not otherwise engage with them, either positively or negatively. We are accepting whatever thoughts observers have about us without attempting to alter them in any way.

This trains both the metacognitive skill of “noticing anxiety as implicit demands” and the skill of “noticing that these thoughts will arise but cannot force specific actions from you” and the skill of “disengaging from the attempt at managing others’ perceived impressions of you.” These are good fucking reps.

Maybe someone thinks you’re weird sitting alone, and maybe they don’t. The important thing is to accept this outcome and do nothing to prevent it. Because if someone looks at you and is like “WOW THAT GUY LOOKS LIKE SUCH AN ASSHOLE, STARING OFF INTO SPACE PERFORMATIVELY LIKE A HIPSTER” the consequence of this happening is nothing.

Which I suspect makes “sitting at the bar while not managing others’ perceptions” a pretty good beginner-level place to practice the STAR technique, specifically the Trace and Abandon Hope elements. In RF-ERP terms, we’re going to a place where the compulsion is likely to fire and then not doing the compulsions. In particular the mental problem-solving compulsions around “figuring out how awkward you look at this instant and then attempt to reduce that amount of awkwardness.”

Over the last few months I’ve gotten a fair few responses to my work; sometimes it’s just appreciation, sometimes it’s critique or advice requests, and I’ve done my best to respond usefully where I can.

And after Manifest, I ended up with a few people with social anxiety or related issues who volunteered to be guinea pigs for the very specific kind of introspective skills training that I am trying to do. (Love you guys!)

So far, it’s been useful for me and, I believe, for them! I have three people who I’m meeting with reliably. But I need more. This is for a few reasons:

• As I discussed previously, I need to collect outcome data, which is hard to come by otherwise: there have been quite a few people who told me, unsolicited, that reading my stuff produced significant and lasting change for their social anxiety. Which is great! But I lack a denominator— the set of people who bounced off my work. And because these snippets come without context it’s difficult for me to tell the difference between (1) genuine lasting change by stated mechanism, (2) genuine lasting change for some other reason, (3) transient insight high from reading a (if I do say so myself) absolute banger of a piece of writing which doesn’t translate into behavioral improvement, and (4) transient but real improvement followed by regression. At least some of these comments claim to be from months after reading, which at least removes (3) and (4) as possibilities for those people; at Slutcon I saw either (1) or (4) and because of my lack of persistent relationship with those people I had no way of distinguishing “lasting change” from “fun weekend”. I’ve definitely seen at least one case of (2).

• For failure cases, I need to have more information about what caused the failure.

My theory of change around coaching is extremely concrete and specific. I suspect people who have trouble following my protocol are mostly running into one of the following issues, which can occur separately or together in any combination:

• they lack introspective access into their own anxious cognition. Many people do seem to perceive it as an unpleasant somatic fog instead of something with correct-in-its-frame logic. I think there are a lot of ways of training this introspective capacity (which goes under the name “mindfulness” in some traditions.) Need: improved introspective access, which I’m pretty sure is quite trainable, and which I have a number of ideas in how to train (post incoming).

• They notice that the anxious thoughts are not ones to identify with, but have trouble acting on this knowledge. Need: improved introspective access + improved rumination control (also a learnable skill; metacognitive therapy speaks of this, as do Michael Greenberg essays).

• They are in situations where culturally the territory is contested about what you can and should do and why, which makes it difficult to tell the difference between anxious and healthy cognition (see also: social-romantic anxiety). Need: improved introspective access + defensible principles + very specific skills training + knowledge of how different environments differ.

• They lack baseline social skills (comorbid autism is a common reason for this), and anxious modeling of other people is a method of coping with this lack of natural room-reading ability. You’ll know if this is you because people will, on a regular cadence, chew you out for somehow upsetting them. Need: concrete process-oriented social skills training, plus the other stuff. Resolving your social anxiety is going to be tricky if you frequently have people chewing you out for random things at irregular intervals.

Unfortunately I don’t have a strong model of how to train up baseline social skills for the autism case; by all means schedule a call, but know that here I will be noodling through an unsolved problem with you as opposed to executing on a model I feel confident in. But the others I can probably help with! I think a reasonable indicator for if you should reach out is if you felt seen by one of my essays, since that tells you at very least that I’m working from an accurate model of your problem as seen from the inside. (If, on the other hand, you read Abandon Hope and was like “wtf this is bullshit” then tbh i’m a bit surprised you’ve read this far.)

I predict having nonzero success, and I also think the failures will be informative in future writing. (Also, I’ll only write about any particular person’s experiences anonymized and with their permission. I am very likely to talk about the results from a particular approach I’m taking in general, without discussion of any particular person for whom it went well or poorly.)

I’m offering this at a sliding scale starting at (for now) 150 dollars an hour and ending (for now) at “free;” I’d like to make this accessible to people regardless of ability to pay, at least until my schedule gets crowded or I start really needing money. [Edit: given the number of responses I’ve received thus far, the “schedule gets crowded” clause is likely to kick in sooner instead of later.]

I believe I am quite knowledgeable on the topic of:

• Social anxiety

• Its almost-unstudied cousin, social-romantic anxiety

• Building a social and romantic life

• Partner dancing (and specifically getting through beginner hell as a lead)

I’m willing to talk about whatever, but outside my stated range I may or may not have actionable advice.

Note that I’m not a therapist and this isn’t therapy, and indeed I cannot actually treat social anxiety or anything else; this is, instead, training for a very specific set of introspective skills that I believe are load-bearing in resolving social anxiety and possibly other forms of anxiety. If you’re in acute distress I recommend seeing a therapist.

I’ll say this: any time I am asking you to do a thing I’m going to have an extremely clear mechanistic theory of change around that thing, which I will be willing to explain (and, if you like, defend) at length. I feel like “do this mysterious inner work thing because it will help somehow, trust me” is a weirdly common norm that has developed over time and I dislike it; I will never just ask you to trust the process. “Trust the process” is silly, anyone with a process and a worldview should be able to explain their theory of change and what specifically they are doing concretely, in terms a reasonable person would understand.

So reach out and I’ll shoot you a calendly link! (Also reach out if you just want to talk about your experiences with the protocol/social anxiety whether or not you need advice; I’m always fascinated to hear how people have done with it and in what contexts.)

A few years back before I had my MOMENT OF SATORI I was doing a model-work-heavy workshop with Fully Known (I was, at the time, still trying to chill the fuck out around attractive women) and I got hella chewed out by one of the girls, who said something like:

“You speak fast and you move fast and I find that very unattractive. I like men who are slow and purposeful, and this fast movement you are doing… just extremely, tremendously unattractive.”

(At a different point one of the facilitators— I think it was Nick— told me that I reminded him of a snake being charmed by a flute.)

At the time I was like “eh, women like what they like,” but on reflection what was actually being conveyed was the status tells around extremely fast and uncertain movements. Fast, jerky, self-interrupting movements imply “I am continuously monitoring for your approval in real-time and moving to adjust,” which is the primary low-status meta-behavior.

For which the book Impro, by Keith Johnstone, is just loaded with incredible material; the author describes himself “becoming” high status and low status via body language in front of students, who then try to figure out how he is conveying this:

I change my behaviour and become authoritative. I ask them what I’ve done to create this change in my relation with them, and whatever they guess to be the reason—‘You’re holding eye contact’, ‘You’re sitting straighter’—I stop doing, yet the effect continues. Finally I explain that I’m keeping my head still whenever I speak, and that this produces great changes in the way I perceive myself and am perceived by others. I suggest you try it now with anyone you’re with. Some people find it impossible to speak with a still head, and more curiously, some students maintain that it’s still while they’re actually jerking it about. I let such students practise in front of a mirror, or I use videotape. Actors needing authority—tragic heroes and so on—have to learn this still head trick. You can talk and waggle your head about if you play the gravedigger, but not if you play Hamlet. Officers are trained not to move the head while issuing commands.

Keith also talks about the role of perceived relative status in making improv scenes come alive; that most people have a “preferred” role they like to play in scenes— high status or low status— and that he conceives part of his job as Improv Coach as being to force players out of their preferred roles:

In the professional theatre actors divide up roughly into high-status and low-status specialists. The actors’ directory Spotlight used to have the high-status specialists at the front (called ‘straight’), followed by the low-status specialists (called ‘character actors’), followed by children, and then dogs. This isn’t as bad as it sounds, but it’s symptomatic of the tendency for actors to overspecialise. A proper training for actors would teach all types of transaction.

Some problems: there are students who will report no change of sensation when they alter their eye-contact patterns. If you observe them closely you’ll see that the ones who always play low status in life won’t ever hold eye contact long enough to feel dominant. When high-status specialists break eye contact and glance back, they’ll be holding the glance back for at least a second, which is too long. You may have to precisely control the length of time that they look before they experience the change of sensation. Then they’ll say, ‘But it feels wrong.’ This feeling of wrongness is the one they have to learn as being correct.

I remember a girl who always played high status in improvisations, and who had never experienced safety and warmth as a performer. When I asked her to put a short ‘er’ in front of each sentence she used the long ‘eeeerrrr’ but denied she was doing it. When I asked her to move her head as she talked she moved it in an abstract way, as if watching a fly circling in front of her. I asked her to play low status with an expert low-status improviser, but she held on to herself tightly with her arms and crossed her legs as if refusing to let her partner ‘invade’ her. I asked her to unfold and then to tilt her head and suddenly she was transformed—we wouldn’t have recognised her. She became soft and yielding and really seemed to enjoy the feelings that flooded into her, and she acted with feeling and rapport for the first time. Now that she’s learned to play low status with a low-status partner she can learn to play it with a high-status partner.

Another student refused to play high status in anything but a wooden manner. He said that he lived in a working-class area and that he didn’t want to be stuck-up. I explained that I wasn’t trying to remove his present skills, which were very necessary, but only to add a few new ones. He believed that it was necessary to play low status within his working-class community, not realising that you can play high or low in any situation. His problem is that he plays low status well and he won’t experiment with other skills.

I asked him to play a scene in which he was to tell his father he had VD. I chose the scene in order to stir him up, and involve his real feelings. All young men have anxieties in that area. He acted out a scene with no conviction at all, and tried to think up ‘clever things’ to say.

‘I’ll give you the dialogue,’ I said. ‘Enter. Go to the window. Look out, then turn and say you’ve got VD.’

He did this. He looked out of the window and immediately made trivial movements, and dropped down in status.

I stopped him. I explained that if he turned from the window, looked at his father and didn’t move his head, then he’d experience exactly the sensations he was trying to avoid. I said that he mustn’t try to suppress the head movements but that he must be aware when he does them, and then somehow feel so dominant that he no longer needs to make them. When he repeats the scene it’s his father who breaks eye contact and starts to crumble. From this beginning the student can learn to play characters of any social class, and make them high or low.

This is surprisingly similar to the bits about white-knuckling in exposure therapy: it’s not actually that useful to Just Do The Things while terrified. You have to figure out why it is you are doing safety behaviors compulsively, and then make whatever mental motions are required to stop this from making internal sense; going directly to “stop doing the safety behaviors” is mostly just suffering with extra steps.

Similarly, I noticed in a different workshop run by Lynn that the things she would try to correct in the men were, mostly, lots of little low-status tells. But this direct-suppression-of-signals was ineffective holistically; because the low-status tells (rapid head movement, appeasement-oriented verbal behaviors, crossed arms, looking down) were all part of a unified motor-emotional loop, so she ended up playing this frustrating game of whack-a-mole.

The Snapping Game

Keith also has weird little games he invents that shoehorn players into one role or the other, like the Snapping Game: one player is the Master, and the other is the Servant. The rule of the game is simple: if the master snaps his fingers, the servant flops to the floor and dies on the spot and another servant takes his place. And the master is given an instruction: if the Servant ever annoys him in carrying out (or failing to carry out) his duties, if even the slightest twitch of vexation crosses the mind of the Master, he is to snap his fingers and murder the servant; and that players take to their roles surprisingly quickly. (Also the scenes generated thereby are reportedly hilarious.)

My suspicion is that the Snapping Game makes status undeniable and immediately consequential; players don’t have to act high or low-status in this game, they are high-status or low-status because of how the container works and the feelings flow from there.

Spontaneity As Introspective Training

Less often discussed is the Spontaneity chapter; in it, Keith describes how difficult (but necessary!) it was to get his students able to say the first thing that came to mind, without any sort of censorship. It’s hard to turn that off.

‘I saw your hand reject an earlier shape. What did you want to take?’
‘A tin of sardines.’
‘Why didn’t you?’
‘I don’t know.’
‘Was it open?’
‘Yes.’
‘All messy?’
‘Yes.’
‘Maybe you were opting for a pleasanter object. Mime taking something else off a shelf.’
His mind goes blank.
‘I can’t seem to think of anything.’
‘Do you know why?’
‘I keep thinking of the sardines.’
‘Why don’t you take down another tin of sardines?’
‘I wanted to be original.’

I ask a girl to say a word. She hesitates and says ‘Pig.’
‘What was the first word you thought of?’
‘Pea.’
‘Tell me a colour.’
Again she hesitates.
‘Red.’
‘What colour did you think of first?’
‘Pink.’
‘Invent a name for a stone.’
‘Ground.’
‘What was the name you first thought of?’
‘Pebble.’

Normally the mind doesn’t know that it’s rejecting the first answers because they don’t go into the long-term memory. If I didn’t ask her immediately, she’d deny that she was substituting better words.

‘Why don’t you tell me the first answers that occur to you?’
‘They weren’t significant.’

I suggest to her that she didn’t say ‘Pea’ because it suggested urination, that maybe she rejects pink because it reminds her of flesh. She agrees, and then says that she rejected ‘Pebble’ because she didn’t want to say three words beginning with ‘P’. This girl isn’t really slow, she doesn’t need to hesitate. Teaching her to accept the first idea will make her seem far more inventive.

This is, in fact, the same introspective exercise as I’ve discussed previously in identifying anxious cognition. You do a thing in the heat of the moment that you don’t precisely endorse; and then the facilitator (or yourself) asks, “why did you do that thing you don’t endorse, instead of the actual exercise?” Then the student gives the answer they give; introspective access is used as way of dredging up the mental block the student has installed. Only after the student recognizes the feeling of having performed the mental block can he learn that specific mental motion of routing around it. The facilitator, in this model, is there to construct places where the harmful cognition will fire legibly-enough to be identifiable for the student.

Get Thee To The Improv Studio

Improv comedy teaches tremendously useful skills (or at least it can, if you make an effort to learn them.) This activity takes a close second to dance in “activities I full-throatedly endorse for anyone who is socially anxious or wants to get better at dating.” Ideally an improv studio that flows from the Keith Johnstone lineage.

See Also

When people attempt the Abandon Hope protocol (STAR) and it fails for them, what has happened and why?

Right now I’m vending this material out through a blog. I think having the blog as primary transmission medium is important because 1:1 coaching can’t ever scale beyond a single practitioner; but I’ve started doing 1:1 coaching anyway (hit me up, I’ve got ~3 people so far) and frankly most of the reason I’m doing so is because my async feedback channels via writing mostly suck.

Like: there’s going to be a ton of people who read my writing and some proportion will bounce off my homegrown protocol and some proportion (in absolute numbers, more than I expected!) will claim it one-shotted them in some important and beneficial and lasting way. I have no real way of calculating the ratio because “people who write in to say it worked for them” doesn’t include the denominator of those for whom my stuff failed. 1:1 coaching is a feedback channel that lacks these specific selection effects, though sadly it does have the selection effect of “this is a person for whom reading the protocol was inadequate to allow for solo execution.” Nothing’s perfect.

And yeah blog comments are fine, but I don’t love them as a conversational medium and there’s a lot of important context that I’d be getting in a 1:1 coaching relationship which will simply be absent in a blog comment.

Anyway, with 1:1 coaching I can get a better idea of what exactly the problems are that people run into, and see what framings do and don’t work in trying to teach people the very specific metacognitive skill that I think is important here. I intend to propagate everything I learn from 1:1s back into the Abandon Hope post (or other blog posts, depending.) The failures, in particular, are critical information for improving the framework.

The point is to make myself entirely unnecessary in this process. Which is probably unachievable in full generality but I bet I can get most of the way there.

Some More General Open Theory Questions

Why are some impossible-to-control outcomes stressful to engineer and others aren’t?

The goal of “make this person feel welcome at my meetup” is a goal about someone else’s internal state, the same as “make this person like me.” Clearly the difference between those is that one of them invokes self-focused attention (or more specifically, attention on oneself as a social object); but why self-focused attention anxiety-generating in general? I don’t stress about one in the same way that I stress about the other, even though the same core issues appear in each (the inability to read minds, the inability to reliably engineer internal states in others, the potential for failure to reflect poorly on me as a person.)

Why do League of Legends games become ~90% less stressful if you’re playing on a team against an AI rather than against other humans?

The temporary Doombots League of Legends variant (RIP) featured teams of players against extremely formidable AI opponents. I notice that people were wildly less stressed out when we started losing against these AI opponents, who had a bunch of complicated bespoke advantages that made them entertaining to play against. The same people who would lash out at teammates when losing regular matches would, in Doombots Mode, just happily be like “aw shucks them’s the breaks.”

You can see this in other games also: PvE is just generally less stressful than PvP even when players can’t speak across teams. Why would this be the case?

What is the maximally anxiolytic attentional focus for introductions to groups and new people?

There’s still always a twinge of nervousness when I introduce myself to a new person ex nihilo, or try to join a group conversation already-in-progress. Which implies I still have impression-management goals I default to in these situations. Intolerable!

The anxiolytic goals I have noticed are

• Experimentalist: “what will happen if I say this thing?”

• Beneficence-oriented: “I can probably make this person’s day better by introducing myself.”

Damn it, probably I do need to start cold-approaching in bars again; can’t make headway on a theory without experiments, and I’m my own best subject. So it goes.

This has been bugging me for a while. The DSM has basically three categories I’m interested in: Anxiety Disorder, “Related To OCD” Disorder, and OCD Proper.

I think this divide does not stand up to any kind of principled scrutiny, which I gloss over briefly in one of the Abandon Hope appendices but didn’t really justify thoroughly.

Let’s discuss.

The Anxiety Disorders Chapter

This includes

• separation anxiety disorder

• selective mutism

• specific phobia

• social anxiety disorder (social phobia)

• panic disorder

• agoraphobia

• generalized anxiety disorder

“OCD And Related Disorders” Chapter

• obsessive-compulsive disorder (OCD) proper

  • Subtypes: Contamination ocd, “gay ocd”, symmetry OCD, scrupulosity OCD, moral OCD

• body dysmorphic disorder

• hoarding disorder

• trichotillomania (hair-pulling disorder)

• excoriation (skin-picking) disorder

Common Presentations For OCD Proper

• Contamination / cleaning: fears of germs, with washing or cleaning compulsions.

• Harm / checking: fear of being responsible for harm (fire, burglary, accidents), with repeated checking and reassurance-seeking.

• Symmetry, ordering, and “just right”: a need for things to be even or to feel correct via careful arrangement.

• Unacceptable/taboo intrusive thoughts: distressing aggressive, sexual, or religious (scrupulosity) obsessions. Because the compulsions here are primarily mental (rumination, praying) rather than visible, this is colloquially called “Pure O.”

How Were These Categories Arrived At Initially?

For the DSM-I: Freud. It was Freud.

They bought his stuff pretty much wholesale at the time; they labeled OCD as “obsessive compulsive reaction“ (later, “obsessive compulsive neurosis”) filed alongside “anxiety neurosis”, “phobic neurosis”, and “hysteria” within the broader family of “neuroses.” They lumped symmetry-based presentations of OCD with fear-based presentations (eg contamination) and then sort of unreflectively carried those into all future DSM versions.

BEHOLD, an AI-generated diagram of the relationships between various disorders per the DSM-I:

1980: The DSM-III Cancels Freud

The committee writing the DSM-III had decided the Freud was on the outs and attempted to purge the DSM of all of his latent influence, and that meant abolishing “neurosis” as an organizing category entirely, since “neurosis” has a technical meaning in Freudianism, namely: disorders born of repressed desires.

The reasons for this were numerous but one of the big ones was that they wanted to do their categorizations not on the basis of presumed origin but on typical presentation of the course of the disease.

The members of the “Neurosis” category were dispersed based pretty much on surface affect: anything that caused anxiety went in “anxiety disorders” (including OCD since most OCD presentations involve lots of anxiety). Depression went into “Affective Reactions” class, which was considered a subsection of “Psychotic Disorders” alongside “Manic Depressive Reaction” and “Pychotic Depressive Reaction“.

The DSM-5: Moving OCD Out Of Anxiety Disorders

Later instances of the DSM made small changes, sometimes based on neuroimaging studies or responses to various medications, but nothing that actually challenged the core categories discussed above; OCD as an illness predates most of what we think of as “modern psychology.”

The 2013 reorganization, after a bunch of argument between psych nerds, pushed OCD out of anxiety disorders and into its own chapter. It shuffled in which chapter OCD got placed, but both sides of that debate took for granted that OCD is a coherent, unitary thing uniting symmetry and harm presentations which belongs in one list together: contamination washing is filed with symmetry ordering and scrupulosity, as a package.

There Is No Consistent Definition For OCD That Reliably Distinguishes It From Anxiety Disorders

The DSM defines OCD as the following:

DSM-5 Diagnostic Criteria for Obsessive-Compulsive Disorder (300.3)

Presence of obsessions, compulsions, or both:

Obsessions are defined by (1) and (2):

1. Recurrent and persistent thoughts, urges, or impulses that are experienced, at some time during the disturbance, as intrusive and unwanted, and that in most individuals cause marked anxiety or distress.

2. The individual attempts to ignore or suppress such thoughts, urges, or images, or to neutralize them with some other thought or action (i.e., by performing a compulsion).

Compulsions are defined by (1) and (2):

1. Repetitive behaviors (e.g., hand washing, ordering, checking) or mental acts (e.g., praying, counting, repeating words silently) that the individual feels driven to perform in response to an obsession or according to rules that must be applied rigidly.

2. The behaviors or mental acts are aimed at preventing or reducing anxiety or distress, or preventing some dreaded event or situation; however, these behaviors or mental acts are not connected in a realistic way with what they are designed to neutralize or prevent, or are clearly excessive.

Let’s go through these criteria! I contend that any of these, applied consistently, would either move one of the OCD-spectrum disorders into the anxiety disorder category, or vice-versa.

“OCD features repetitive compulsions, whereas anxiety just features avoidance.”

Not so!

First, social anxiety sufferers frequently seek out reassurance that their feared outcomes have not come to pass; this comes both in the form of seeking verbal reassurance and also comes in the form of compulsive rumination about past social events. Second, social anxiety features the vast number of compulsive tics laid out in the SAFE:

Generalized Anxiety Disorder also features these repetitive mental compulsions (which would put it in the colloquial “Pure-O” category.)

Health Anxiety features repetitive doctor visits and, I assume, checking WebMD for if you have cancer. (Answer: maybe, always maybe.)

“OCD compulsions aren’t realistically connected to avoiding the feared outcomes, whereas anxiety disorder safety behaviors are.“

Example: symmetry OCD has no intrinsic feared outcome. Which: fair enough.

Counterexample: contamination OCD compulsions— washing, cleaning and masking up— are genuinely related to the project of not getting sick; they’re merely done to life-limiting excess.

Likewise, checking the stove for the 9th time is related to the project of not-accidentally-leaving-the-stove-on.

“OCD-spectrum disorders involve thought-action fusion, whereby anxiety disorders don’t.“

Thought-action fusion is a concept where some people believe that some of their thoughts— as in, totally internal events that happen inside their head— have genuine moral or practical significance and thus must be controlled. This is dysfunctional because you don’t have direct control over your thoughts (see also: “don’t think of a pink elephant.”)

This is only true of some types of OCD— harm OCD and scrupulosity OCD (where the patient is worried about going to Hell as punishment for blasphemous thoughts) are variants featuring this. Contamination OCD doesn’t have this feature; nor does hoarding, nor does checking OCD.

“OCD obsessions are ego-dystonic, but anxiety disorders aren’t.”

Contamination OCD is often ego-syntonic. People legitimately want to avoid getting sick, and sufferers often (though not always) perceive themselves as washing to get the germs off, as is reasonable.

“Okay, but they know their degree of washing is irrational!”

That doesn’t save you: social anxiety sufferers also know that their safety behaviors are irrational (“it’s a nightclub, I’ll never see any of these people again, which means there can be no real social consequences, so why am I still freaking out and doing all the safety behaviors.”)

“You have to ban rituals to treat OCD, whereas you ban safety behaviors to treat anxiety disorders”

“Safety behaviors” are literally just rituals, except rebranded to fit anxiety disorders. They protect against the feared outcome in exactly the way OCD compulsions do: which is a little bit, at great cost.

The True Differentiator Is Whether The Feared Outcome Is Unfalsifiable, And Almost None Of Them Are

The inhibitory learning model of social anxiety goes like this: the Conditioned Stimulus (you see a dog) gets, at some point, paired with an Unconditioned Stimulus (the dog bites you.) You seeing a dog is not intrinsically aversive; it only becomes so after you’ve paired the two events in your mind. The dog bite is intrinsically aversive (thus, “unconditioned”— you need no conditioning to not want to get bitten.)

The exposure therapy model constructed out of this requires that the patient be exposed to the Conditioned Stimulus (the sight of a dog) without the Unconditioned Stimulus (the bite) enough times where they learn that these two things aren’t paired to each other.

In simple phobias (like fear of dogs), this occurs easily: it is demonstrably true that you’ll leave the exposure session safe and unbitten by the dog. Learning can occur.

The problem that OCD-spectrum disorders— and, I argue, most anxiety disorders— all share is that the Unconditioned Stimulus cannot be reliably avoided during exposures, because the Unconditioned Stimulus is a mental event in the patient’s head.

Let’s take a few examples:

Social anxiety

In social situations most people don’t tell you that they think you suck, even if this is truly the case. You infer that they think you suck; which means you cannot reliably get an exposure where you receive the Conditioned Stimulus (being at the party) with a clear absence of the Unconditioned Stimulus (other people thinking you suck.)

Which means you’ll have people consistently getting exposures to social stimuli, countless exposures, multiple times a day, where everything went fine on the surface but where this didn’t resolve their problem in any way because they are still in the habit of summoning up the Unconditioned Stimulus (“people think I’m so cringe oh god”) all on their own.

Contamination OCD:

You can’t check whether something is “contaminated.” You can check whether you’re sick, but if you do this then chances are really good you’ll turn up some kind of evidence. My throat feels funny— did it feel this way a few days ago? I coughed just there. Maybe it’s covid?

Which means you can’t run clean exposures that disconfirm the thing you’re fearing; if you touch the doorknob, you can’t prove that didn’t damage you in some invisible way.

Generalized Anxiety Disorder/Hoarding:

The problem here is that the feared event is in the future, and there is an indefinite amount of future which is unknowable to the patient. Their fear is that something bad will happen that they need to prepare for now: either via constant contingency planning (GAD) or hoarding items in case they are useful later.

You cannot run an exposure where nothing bad happened in response to throwing a thing out. Because the fear isn’t something bad will happen now, it’s that something bad will happen… someday.

(Actually this one isn’t as clean as what I’m describing here; I know of at least one hoarder who I think mostly did it due to extremely poor executive function, and I’ve read that some hoarders develop emotional attachments to all their stuff that doesn’t exactly fall into what I’m describing here. It seems plausible that there are a few mechanistic reasons someone might end up looking like a hoarder from the outside.)

Religious Scrupulosity OCD:

I don’t know what coherent exposure protocols for this could even look like. “Think blasphemous thoughts and notice that you’re not in Hell yet?” As far as I know treatment is mostly sending such people to religious authorities who can talk them down.

Gay OCD:

A mildly hilarious (sorry, but it is) form where a man gets intrusive thoughts of other naked men, and then is like “OH NO I NEED TO STOP THESE INTRUSIVE THOUGHTS OR ELSE IT MEANS I’M GAY”

This one is non-disprovable because… well, there’s no amount of not-getting-a-boner-around-guys that disproves you’re gay because what if you just haven’t met the right guy yet? What if you’re repressed? (More broadly, “gay” and “not gay” are just descriptive labels we attach to people based on their typical patterns of attraction rather than, like, a thing you can get blood-tested for and which is writ upon your soul.)

The True Divide(?):

Straightforward Phobias:

• arachnophobia/fear of dogs/etc

Intolerance-Of-Uncertainty Disorders:

• Anxiety-Based OCD subtypes (contamination, checking)

• Social phobia, generalized anxiety disorder, health anxiety

Craving For Symmetry/Completeness:

• Symmetry OCD

• Skin-picking

• Hair-pulling

• Nail-biting (presumably)

“Expectancy Violation” is the wrong model

If you’re guarded from falsification of your fears by the structure of reality, then the target of exposures cannot be “expectancy violation” as such: it’s “learning to stop doing the things that are meant to guard against the feared outcome, because they are limiting your life in ways that have led to you seeking treatment for a named disorder.”

Which almost necessarily involves letting the feared outcome happen more often.

Social anxiety safety behaviors are a pretty good way of not causing people to hate you, at the cost of never developing new human connections. Contamination OCD rituals are a pretty good way of not getting sick, at the cost of not leaving your house or doing most interesting activities. Generalized Anxiety Disorder’s relentless planning activities are a pretty good way of not getting blindsided by reality, at the cost of being kind of low-key anxious all the time.

That’s why the treatment isn’t about arguing with the disorder; because the disorder follows from (in most cases) reasonable premises. It’s about mechanically shutting down the habitual activities done in service of avoiding this unfalsifiable feared outcome.

Thus, response prevention.

Who Cares?

OCD and Anxiety Disorder aren’t really natural categories; I think stuff in those categories largely ended up together by historical accident.

If they were merged, this would force unification between OCD-style response prevention principles and CBT-for-anxiety principles, since the latter includes prominent reassurance components but the former doesn’t and I suspect the former is correct for the reasons discussed.

I’ve listened to the Stratechery podcast for a long time; Ben’s a terrific analyst and has made me a substantial amount of money on Meta specifically, for which he has my gratitude. Nevertheless, I think he is confused with regards to Anthropic. He has written, regarding the recent Fable standoff:

Here’s the thing about these safety justifications: I think they work because, to Anthropic, they aren’t justifications. The company really believes that they are the only ones who believe in super intelligence, and thus are the only ones who are sufficiently concerned about the dangers. That excuses decision after decision, policy after policy, and confrontation after confrontation that, to people on the outside, look like a bizarre combination of cynicism and naiveté.

The contrast to OpenAI is massive: I think that one way to understand how and why OpenAI lost its lead is that, in the years following the release of ChatGPT, the company has been at war with itself internally as what used to be a research lab was suddenly seized with the burden of being the accidental consumer tech company; to the extent OpenAI solved that conflict, it was by bleeding huge amounts of talent to Anthropic in particular.

Anthropic, on the other hand, has perfect alignment between talent and mission and business. The company gets to sell to researchers the creation of a machine god, with the mantle of being the sort of person who cares about the dangers and is smart enough to navigate them on behalf of humanity; that every policy change that falls out of that happens to be great for business is the most beautiful coincidence in the world.

Ben’s favorite charge to make against Anthropic is arrogance: they are claiming their safety-related concerns give them the right to make hugely consequential decisions for everyone else about how their software can be used.

(I rather liked TheZvi’s take, that we’re observing in Ben “a moderately strong version of the negative form of Anthropic Derangement Syndrome (ADS), which like Trump Derangement Syndrome (TDS) exists in both directions and to varying degrees.)”

Now, as it happens I’m biased; I’m of the camp that thinks superintelligence is a big fucking deal and more-than-nuclear-weapons-dangerous so of course I’m sympathetic to their safety concerns. But, sure, this is a minority opinion in the greater tech ecosystem and not really what I’m interested in discussing here.

Really I just wanted to point out that there is no way for Anthropic to abdicate making huge decisions on behalf of everyone else.

Allow me to illustrate with an example:

Anthropic recently got into a fight with the US government due to them telling the US government that their ToS is more restrictive than “any lawful use” (which given the current administration could be simply rounded off to “any use.”) In particular they were unwilling to sign off on Terms of Service changes that would allow the Department of War to use Anthropic’s models to perform mass surveillance of American citizens, and to use Anthropic’s models to power completely-automated killing machines. Anthropic refused and thus was branded a supply-chain risk.

Put down, for the moment, whether these are reasonable demands for the US government to make of a private company. Anthropic had, at that point, found itself at a crossroads: it can either allow the US government to use Claude in this way, or it can forbid the US government from doing so. Either option is hugely consequential and there is no secret third option where Anthropic can simply not make a decision for everyone, since whatever it does, it will have been the one to decide whether US citizens get mass-surveilled or not. There is no possible decision Anthropic could have made which would have led them to not be responsible for that decision.

This is contra Ben:

At the same time, what is the standard by which it should be decided what is allowed and not allowed if not laws, which are passed by an elected Congress? Anthropic’s position is that Amodei — who I am using as a stand-in for Anthropic’s management and its board — ought to decide what its models are used for, despite the fact that Amodei is not elected and not accountable to the public.

And, on the second point, who decides when and in what way American military capabilities are used? That is the responsibility of the Department of War, which ultimately answers to the President, who also is elected. Once again, however, Anthropic’s position is that an unaccountable Amodei can unilaterally restrict what its models are used for.

Did Anthropic engage in arrogant overreach by refusing to amend their terms of service at the instruction of the Trump administration? I declare the whole question as malformed: you can frame the humble-inaction option as being “quietly agree to give the US government heretofore unknown powers of surveillance” or “refuse to do that” based solely on grammar and word choice.

But that’s not the thing that bothers me most about this specific case.

We All Know Courts Are A Thing, Right?

Check the quote above: Ben’s claim is that the American people should make these decisions, rather than a small group of Silicon Valley nerds. This sounds so proceduralist! So reasonable!

But I think that’s stupid. People always forget that we’re not just a democracy full-stop, we’re a democracy backed by a constitution. That constitution lays down the things that our duly elected officials are and are not allowed to compel private actors to do, and— more importantly for my point— it lays down what conflict resolution looks like, which is distinct from Ben’s implicit “do whatever the executive branch tells you to do because they were elected and you weren’t.”

If the government wants you to do something and you don’t do it, the appropriate response is for them to bring you to court. American democracy does not equate to “the legislative and executive branches can do whatever they want.” The judicial branch is a separate branch for a reason and I’d argue that the appropriate, proceduralist thing for Anthropic to do here is to dig in their heels as principles and reason dictate, then let the Trump administration bring them to court. Which is the traditional form of mediated conflict resolution in our country.

Would the Trump administration win? Probably not, no. I know of no specific legal theory by which this case could even make it to court in the first place. But, like, that’s the point of laws: they constrain the power of our officials who were duly elected by voters. If no legal theory exists by which the Trump admin can sue Anthropic for not amending it’s ToS, then this whole affair is simply Anthropic asserting its legal rights in the standard American tradition.

Also giving in wouldn’t even let Anthropic dodge responsibility for the outcome

Disregard all that for now.

If you decide that the American People should decide what to do about AI (or more specifically that Trump’s cabinet should decide, who were kinda-sorta-but-not-really-elected along with Trump), it is very likely that this meta-decision will result in very poorly-informed concrete decisions down the line. Some of these decisions may be existentially weighty. If you’re comfortable with that, fair enough; and if you aren’t, then this meta-decision (“let the American People decide”) has brought you to a very bad place. And it was your meta-decision to Let The American People Decide, which means it’s your fault.

Is that unfair? I don’t know, maybe? It’s the same principle where if you tell the drunkard where his car keys are, then you are causally responsible for the car accident that follows. It cannot be otherwise. (Keeping the keys, of course, will also mean that whatever follows is your fault: the tragedy of the Trolley Problem is that there is no escape from accountability if you’ve found yourself at the lever. Pull or don’t pull, there’s blood on your hands either way.)

Ben’s Not The Only One

I think Pause AI people (though I am fond of them) can be guilty of a similar thing: I’ve heard a couple of them talk about how Anthropic is performing actions that could lead to the deaths of them and everyone they love (true!), and thus the Anthropic research people should all quit. Which ignores the fact that there are also very clear possible timelines where Anthropic unilaterally pausing also leads to everyone dying. There is no possible decision that can be made at the level of a single frontier lab that, guaranteed, doesn’t lead to everyone dying. Allow me to illustrate:

At some point you can’t dodge having to make a judgement call. You can’t be like “oh I did the modest thing and thus I am not responsible for whatever happens.” You need to weigh the probabilities and play the hand you’ve been dealt. You can’t just not kill everyone.

My Object-Level Views On AI X-Risk

I think the Pause AI game plan and the Anthropic game plan are both likely to fail. You can see this illustrated on Twitter, where for the most part anyone making the pro-Anthropic or pro-AI-pause safety arguments is doing it primarily by emphasizing how unworkable the alternative is.

For Pause AI: attempting to rig together an international agreement that effectively bans frontier model development is absurdly expensive, far more than global warming remediation which we also haven’t been able to force the rest of the world to do. Even if you can achieve this agreement in principle, you’re now tasked with running the kind of police state that can supervise every small and large research lab in the world to ensure they’re not doing anything dangerous; one slip-up and you’re still dead.

This is probably the best game plan we have! I do not dispute this. “Best” does not mean “likely to work.”

The Anthropic game plan, meanwhile, is also brutally difficult. You’re trying to bootstrap superintelligence and that means you have exactly one try to get recursive self-improvement going in a way that preserves humanity’s values, because the superintelligence you get at the end of this process won’t allow you to fix whatever mistakes you’ve made. Anthropic is doing a lot of this work to try and figure out what robust alignment means and how to measure it; whether you judge this as realistic depends on how big you think the basin of “human-compatible values” is. Perhaps it is very large and we’ll be fine with current efforts; perhaps it is vanishingly tiny and near-impossible to hit in practice. Differences between people online seem to mostly be around how small they suspect the basin is in-practice.

(Part of why I dislike the constant low-grade infighting between the Pause AI people and the Anthropic people is that the Anthropic game plan will be the only game in town if the international agreement falls through, which seems quite likely given the current administration’s stance toward the international community. A defense-in-depth approach to existential risk seems better than putting all your eggs in one basket, I think.)

So I think the Anthropic people should keep trying to align Claude as best they can, that God-King Claude Magnum Opus VVII may save us all from our terrible decisions; and the Pause AI people should keep pushing for their international ban which will prevent the Anthropic people bringing about God-King Claude Magnum Opus VVII.

In Conclusion

I think people ought to argue ‘till they’re blue in the face about what the object-level best plan for dealing with AI is. According to Metaculus, we don’t have long— a scant few years before what x-risk experts refer to as “the Fun Zone.”

If you think Anthropic’s plan is foolhardy and doomed, by all means say so! What I think people should stop doing is scolding Anthropic or anyone else for implying any particular group “has the right” to decide things for humanity. I’ll solve it for you: nobody has that right.

Nevertheless, a decision will get made. Better make it a good one.

Several years ago I was at a festival— details don’t really matter— and we were doing karaoke. A girl got up on stage, haltingly sang a few bars, started to realize she was losing the thread of the song, was getting visibly more and more stressed out, then apologized and bolted offstage while the last half of the song played in the background. I think she then hid in her tent for the next several hours.

I was (very slightly) tapped into this girl’s social circle/Twitter feed, which is how I know that she spent much of the rest of the weekend spiraling about this otherwise-forgettable incident. Unfortunate; she seemed quite nice.

Also relatable, because I used to do this quite a lot.

Embarrassment Is Not Just One Thing

An embarrassing incident spins up an enormous number of semi-voluntary compensatory activities in my brain, and I suspect in yours. We have:

1. The initial knowledge that I have done a thing that will result in others thinking less of me

2. The fact-finding mission: “okay, that was pretty bad. How bad was it? And who all plausibly thinks ill of me due to this?” (This involves lots of rehearsal of the incident in question from various points of view.)

3. Planned attempts at damage control: “I’ll just avoid anyone who saw The Incident for the rest of the weekend so that they won’t see me and remember how contemptible that was. Easy-peasy.”

4. Maybe even attempts at repair: “Maybe if I just explain the situation to whoever was there I can repair the reputational damage I have clearly just suffered.”

The first of these lasts for perhaps a few seconds. Each of the others share the horrible phenomenology of the original embarrassment— the heat to the face, the punched-gut nausea, the self-loathing— but can last for hours or days. Which means the felt duration of embarrassment was never just one big flare of emotion decaying slowly (as emotions do): rather, it’s dozens or hundreds of near-identical flares, each one very brief, each one brought to life by the obeyed impulse to analyze and repair this terrible thing that happened. Which is handy, because it implies that the length of your embarrassment spiral is basically a property of this behavioral loop.

I’d like to take a step back for the moment and notice that every step except the first one is literally optional, and each is caused by the implicit belief that you need to do anything whatsoever about the reputational breach that occurred.

Reputation Is A Marathon Not A Sprint

Reputation is mostly not based off of one or two events unless they were, like, extraordinarily terrible and involved someone else getting harmed in some significant way.

You exist around someone for long enough and you’ll get a good idea of how they act; any individual incidents inevitably get averaged out by the vast ocean of their past behaviors. As a result, everyone’s reputation coalesces almost totally outside of their direct control, based on the long tail of small things people hear them say and do.

So the attempts at repairing embarrassment are futile— you’re spending an enormous amount of time and effort trying to fix a single water droplet in the lake of experiences people have had with you— and this is also the cause of your embarrassment-related suffering.

In conclusion, I love this meme:

See Also

I gave a talk on my overall framework for social anxiety to a crowd of nerds at Lighthaven and it was fun! I spoke to a bunch of anxious nerds (affectionate) after the talk, and found a lot of people who diverge from my own specific type of anxious nerd, which I’d like to discuss here.

Here’s a very rough possible typology: people are either good at instrumental system-2 socializing (the Mind-Control Project: conscious effort at selecting social actions that will result in observers liking you) or bad at it. And their underlying system-1 social reasoning is either skillful or unskillful when the Mind-Control Project goes offline.

So here’s a taxonomy based on that:

Standard-Issue Socially Anxious Behaviors, But Functional Underneath

Me, pretty much. When I am attempting reactive impression management (that is— attempting to cause people in my immediate environment to have a high opinion of me) it degrades my social performance a lot and I become stressed out and I spent a long time vaguely aware of this fact but wasn’t sure what to do about it. When I’m not trying to do this I have good social instincts and have a very good time. This is who I wrote Abandon Hope thinking about.

I suspect this is the norm, but everyone suspects they’re the norm, so that suspicion tells us nothing.

High-Functioning Socially Anxious Behaviors

I have a bit in my Abandon Hope essay about people I refer to as Politicians— people who are highly skilled at impression management and do it without obvious strain or discomfort. I made the (in retrospect obvious) error of assuming that the lack of obvious discomfort implies an actual lack of discomfort, given that being skilled at reactive impression management almost necessarily involves hiding any discomfort with this that might exist. I talked to one of these people at Festival Week! Crippling social anxiety hidden within a person who is— as far as I can tell— extremely gregarious and well-liked.

I have no real way of knowing how common this is. Ping me if that’s you! I’d be curious to hear about your experiences.

Low Baseline Social Capability, Plus Anxiety

[epistemic status: EVEN MORE SPECULATIVE]

This is the “they’re genuinely a bit much and are frequently told this and are stressed out about it” quadrant. You don’t have to have Aspergers to be here, but if you do this might be where you end up.

This is a hard one. My Abandon Hope framework started from the assumption that there is, for the modal reader, a basically-functional set of social schemas and behaviors that is masked by the anxious computation. I do still think this is the most-common situation, but that brings us to the question of “what about everyone else?” What about people where they do receive a lot of explicit negative social feedback?

I’ve already received a couple blog comments to this effect— that some people have a really rough time with social appropriateness and are socially anxious— and while I still feel pretty good about the diagnostic and phenomenological end, the prescriptive end (“so what do I do about it”) is extremely half-baked for that population because the Mind-Control Project for all its faults may be substituting for real and complicated social skills deficits. I don’t know this is true, but I also don’t know it isn’t true (hence the Social Wrecking Ball disclaimer in the Contraindications section of Abandon Hope).

I suspect there are a subset of socially-anxious behaviors that are still probably bad regardless of whether someone is in this category. Namely:

1. Pre-event worrying (experienced in the moment as effortful planning). Trying to plan out how to make everyone pleased with you at the end of the night before they’ve even arrived is futile (aside from obvious logistics stuff), and serves mostly to get yourself worked up in ways that are unhelpful. If your ability to model what other people around you think is very bad, then any planning activity based exclusively around that modeling will fuck you up. Either “don’t plan” or “verify with a friend your plan isn’t terrible” both seem like clear improvements.

2. Post-event rumination… I’m not sure about. If your in-situation ability to read people is dicey then I’m not sure your post-situation ability to reconstruct based on half-remembered conversations will be any better?

My tentative conclusion is that what’s actually needed for this sub-population is actually just practical skills building, which isn’t really my domain but these skills seem likely to be in alignment with what I describe in Process Goals In Social Interaction (see below). Process goals that are internally checkable without mind-reading almost have to be how one functions socially without anxiety, and this is true especially if your mind-reading faculty is worse than the typical person’s. But this is still, ultimately, speculation about a population I don’t really know that much about.

It doesn’t help that social skills are deeply subculture-specific: Professional-Managerial Class norms are different from rationalist norms are different from an evangelical church are different from your progressive activist group. Still, there are commonalities: people, I suspect, like talking about themselves regardless of subculture.

This is where I have the least information and am the most uncertain. Does anyone else have thoughts on this?

See Also