Last Friday my daughter was bored and asked me if I could bake a chocolate cake with her. As much as I enjoy cooking, roasting, grilling and setting things on fire, I have to admit that baking cakes is not really my thing. I cook savory stuff, normally. My kitchen adventures run on salt, heat, and tasting as I go so I can adjust here and there. Ask me about the knife position and direction for a specific meat cut and you’ll get a 37-minute explanation about grain, collagen and temperature.

I don’t normally bake anything that’s not lasagna or pizza as I don’t really have the patience for the whole perfect ratios and precision cooking process that comes with baking.

But if it’s going to put a smile on my kids’ faces, the answer is always yes, even if it requires insurance and a written explanation to the authorities later, so there we went.

The cake turned out fine. It came out of the oven looking exactly like a cake should: brown, puffy and tempting. At that point, I had already decided that baking cakes was easy. Anyone who could read a recipe could bake, and people who struggle with it are probably just overthinking the whole thing.

But of course, for a 4-year-old it was not good enough. It needed toppings and all the bells and whistles. Chocolate fudge and whipped cream it had to be.

Subscribe now

By then, I was actually having fun and had lost track of time, so what the heck. Google is my friend, and I can read. We made the fudge and the whipped cream came together on its own with the mixer. We tasted both, they were delicious.

Then we put them both on the cake. Immediately the cake sank like a tire stabbed with a hunting knife.

So we had three great things, individually, that just didn’t work together. I should have seen it coming, but I didn’t. In my world, a cake wasn’t something that needed testing. I was wrong.

We'd ruined a perfectly good cake by adding something that was never in the plan. Damn you, scope creep. But that was on me.

Reading wasn’t the whole solution after all, and I remain a mediocre baker.

I failed to collect the project requirements from my key stakeholder, even though that tiny stakeholder needed a chair to see what was happening on the countertop. She also looked absolutely cute with her face covered in chocolate fudge that she insisted on licking from the wooden spoon.

A second thing I messed up: I got too confident when the cake worked right off the bat. The hard part was done. Now it’s all about smearing it with the toppings. Easy, right?

Our cake was a system. I could have tested the damn toppings on a small part of it and waited five minutes to see if it would sag. That would be the feedback loop. It would also probably have been sensible to taste it and see if the combination was actually good, even though it’s quite difficult to mess up the flavor when combining chocolate with more chocolate.

I tell clients to run that same check every week, and I didn’t run it on my own counter. My brain was probably already in “Feierabend” mode.

We ate the cake anyway. Sunken middle and all.

Happy Building,

— R.

You are being paid to keep something broken. Not badly paid either. The better you are at your job, the worse it gets, and you will be thanked at every stage.

I know how that sounds. Bullshit. But I have the receipts, so hear me out before you decide it does not apply to you.

Somewhere this morning a woman is opening a database, reading rows off it, and typing them into an interface by hand. Almost a year of this. When the process changes or something breaks, she goes back to manually tracing it and politely asking people to check what happened. She is now the only person who understands what the rows in that godforsaken DB mean.

She is doing the job I refused. She is cheaper than me, more agreeable than me, so by every relevant metric to the person paying, she is better at it.

Now this is not a story about a woman stuck with tedious work. She and I are the same product at different prices, and I only found the right word for what I had been doing when somebody put a lower number on it.

For over a year I was the engine of a business that had sold capability it did not possess. Automations, integrations, permissions, domains, accounts, whatever broke at eleven at night, on a fixed retainer. Mind you, a retainer is a fascinating way of pricing a job with no boundaries. But I had my own picture of what that business could become, and I trusted I could get it there. I stopped making it about money, which is a nice way of saying I took on an owner's job while on a fractional operator contract.

Every one of the hours I worked was a subsidy. From the systems perspective, a subsidy that relieves pressure like this one is a loop that hides the costs of keeping the duct-tape fix running instead of fixing it properly. He operated past his own ceiling because I kept absorbing what fell off the end of it, and the better I absorbed, the further past it he could go.

In my mind, I was solving problems, but I was actually financing the decision not to, and the invoices cleared every month without a blink from either of us.

Then the turning point happened. He asked me for a project brief. I wrote one, listed the risks, made the whole thing conditional on a feasibility check before anybody committed to a date.

He read it and said he would not use it. The document, and more specifically, the risks I flagged and the feasibility check, would put the project at risk.

“The risks put the project at risk”.

That is the funniest sentence anyone has said to me at work and it took me months to laugh, which should tell you something about the quality of my judgement that year.

Under the microscope, what actually happened is that he had already sold a system he did not have to a client who had already signed. Then he found a supplier with the interface and no way to connect it to anything, who promised the connection in thirty days.

That promise was the only thing holding the project together, and it came from people who had never built an integration. I proceeded anyway, because he gave me autonomy that week and I was too pleased to look at what I was standing on. Decision power with the supplier, direct access to the client. Finally I was the adult in the room.

For over a year every conversation with anyone outside the company had gone through him, because “somebody out there was preparing to steal the business”, and I had interpreted that as ordinary caution in a new relationship. He felt that his ideas were “visionary” in his own words, and he couldn’t risk someone copying them.

When he gave me that autonomy, I read the open gate as trust. I can still recall the exact sentence I thought, which is its own kind of punishment. “We are finally past this crap and I can do my job”. It feels like a toothpick under my fingernail when I think about it.

Ashby’s law says a controller needs at least as much range of response as the thing it controls. He had run out of range himself roughly a year earlier, and had been running on mine ever since.

He was not “promoting” me. He was overflowing, and a founder who cannot regulate his own business cannot let the regulator talk to the market, because the moment it does, everybody can see where the capability actually sits.

So I built the system, and I built it well. Hand me an interesting problem and a set of credentials and I will produce something beautiful, even on a foundation I had personally documented as unsound. It’s how my brain works.

The supplier missed the deadline. He told me to threaten them with a move to a competitor, so I did, but when the supplier called the bluff (which I didn’t know was a bluff at that point), he stepped back and there was nobody behind me when the moment came to actually move.

He shut the gate the instant I tried to use it, which was consistent of him. Backing that threat would have made the arrangement permanent and visible to everyone in the room, including himself.

If you’re reading this, chances are you have done this yourself. You have accepted authority with no execution attached and called it a good week. If there’s nothing you can terminate, sign or spend without asking while your name is in every conversation, you’re a sacrificial lamb. You’re running point for decisions you are not allowed to make, and you are the one who will be standing there alone when the axe swings.

The integration was never built. It certainly was no priority for the supplier, and building an API was probably not even on their roadmap. Signed client, working interface, database with no route from one to the other, so he decided I would move the data across by hand. Indefinitely.

I refused and pointed at the document he had declined to use. He said I was being disruptive and not playing ball. I told him to go hire somebody for the monkey work.

Monkey work. Literally, to a person, in writing. Then I spent months putting together an argument in my head about how systems devalue the people holding them together, which is quite a trick to attempt with my own former words sitting on the table.

He hired her inside a week and nothing else changed. Meadows called this “shifting the burden to the intervenor”, and the irony of it is that it’s a trap that runs on success. If those manual updates failed regularly, the integration would exist by now, either with that vendor or with someone else. They do not fail, because she is good, and every day she is good the case for building the real solution gets thinner.

Nobody will ever tell her that doing it well is what made it permanent. Nobody told me either, and I was the one being paid to notice. I did, but later than I wish I had.

I left. Cut my losses, walked, and the role was filled within the week by somebody the business could afford more easily. The Frankenstein is still running today, at a cost that makes no sense, unscalable and unfixable, a “product” that became an open-ended manual service nobody can price properly.

He lost. His client is being served by a process that will never improve, and it costs more every month it survives.

She lost because she is now holding a role with no exit built into it, and the work does not pause if she gets sick or needs a day off.

I lost months and about a third of my income to an engagement I had already written the obituary for.

There is no winner in this story, which is the only honest thing about it. Nobody is running that arrangement because they are stupid or cruel, they are running it because at every single decision point the choice was the fast and cheap option instead of the correct one.

I understand we cannot avoid workarounds every single time. But we can choose which dependencies we accept. So I added two things to my processes, two things that are so obvious and dull that I know it’s why they work.

First, every ugly workaround gets a running cost log on the day it is created, logged beside the price of what it replaces. The kill date for the workaround needs to come before the workaround’s running total reaches the price of the actual fix.

Second, every “autonomy” gets tested before I accept it. What can I terminate, sign or spend without asking? If the list is empty, that’s actually exposure, and exposure gets priced as a premium.

The project brief was solid. Every risk in it happened in the order I listed them.

I have been telling you I wrote that document to protect the project, but I really wrote it to protect myself, and I never once used it to stop anything, because stopping things was never what it was for.

You have one too. It sits in a folder and you call it covering your ass.

Happy building,

— R.

The operator behind the operator (yours truly) got locked out of his own laptop by a security feature he never turned on. That was two weeks ago, and I’m still deciding whether to invoice myself for the diagnosis. But hey, if you’re going to humiliate yourself professionally, at least let it have a decent punchline.

Your infrastructure will not warn you before it fails. Warning you was never part of the arrangement, and the fact that you assumed otherwise is the actual finding. This is especially true if your brain is one of those on overdrive the whole time, with 266 things going on at the same time while a squirrel rings the church bells and sings XMAS carols. Hi.

I wanted to access a file that was supposed to be on my workhorse machine. It’s an always-on laptop that runs scheduled tasks and hosts some drives for me. But it was not available, which was the first symptom that something was not right.

So I went checking. It was off. Weird. Maybe there was a power outage overnight that I didn’t notice. I turned on my machine and got a full screen demand for a BitLocker recovery key, on a laptop where BitLocker had never been active. Something in there decided that week was the right time to start asking for consent, several years and several thousand unencrypted files too late.

I found a key in my Microsoft account, one created a couple of months ago in the wee hours of a Monday, at a time when the chances I would be doing something on the computer are absolutely zero. Whatever. That key unlocked the system partition and the main disk, and for four minutes I felt like a functioning adult. Then the second disk in that same machine refused to move, locked behind a key that existed nowhere.

Well sh£t.

My external backups covered the data up until a couple of days before the crash, which is the only reason this is an article and not a career change. The drive itself stayed locked inside my own laptop, guarded by a key I never chose and was never offered the chance to refuse.

Somebody designed it that way on purpose. Encryption that activates without asking, a key that can go missing on the vendor’s side, and a user who finds out about the whole arrangement at boot time is a set of decisions made in a room I was not invited to. I have no publishable adjectives to describe that person.

Windows 11 calls this automatic device encryption, and it arms itself the moment you sign in with a Microsoft account, then encrypts every fixed drive in the machine including the ones Windows does not live on. The key is supposed to escrow to that account at the moment encryption arms, and it gets one attempt to do it. Sometimes that sync fails, which means the drive or worse, the whole device, gets encrypted but the key is never delivered to the MS account. And that is why people keep finding their device listed in their account with an empty space where the key should be.

I didn’t know any of that, and I’m clearly not the only one.

Run that through the audit I usually sell to other people and it fails on both axes. I never consented to the encryption, and by the time it surfaced, the exit cost was every file I hadn’t backed up recently enough.

All that to say the following. There are two types of dependencies: the chosen ones, and the ones that sneak in and install themselves unannounced. Everything has dependencies, and it’s on us to choose which type we will be dealing with.

Dependencies stay invisible until the moment they break, which means that, for the sneaky ones, the first time you see the full map is during the emergency. That is a terrible hour to start drawing maps.

The risk of each dependency is also not really connected to how big it is, but to whether something else can take its place. A large dependency with three available fallbacks is safer than a small one with none.

What made it lethal in my case was the delay. That encryption sat there producing no signal whatsoever, right up until the moment it produced all of them at once.

A system that only reports at the point of collapse has been lying by omission the entire time it worked. Reliability and honesty are separate variables, and most operators discover that distinction during an outage.

Risk assessment, with the information available right before the crash, would have shrugged at that second disk. It had never failed, and by every available measure it was behaving perfectly right up until the morning it stopped.

This is why I am a strong proponent of a dependency audit. The audit asks a crucial question, which is whether you ever agreed to each specific arrangement. A dependency that performs well and that you never chose is still a finding you should probably act on, because performing and being chosen are unrelated properties.

Getting locked out made me angry enough to finally open the drawer with eight external drives in it, going back years. Folders labeled things like “old stuff 2”, which implies an original "old stuff” that got demoted, are always good starting points.

Now every one of those drives was locally reasonable when it went in. Nobody decides to accumulate eight unaudited archives, they just add the eighth one on a regular Tuesday and never evaluate the set as a set.

This scales past hardware without changing shape at all. Nobody decides that one client should carry sixty percent of their revenue either, they just say yes to good work eleven times in a row and never once put the eleven yeses side by side. Most people rely solely on public power grids until the day they lose 10 hours of work because a someone tripped a circuit breaker in Spain. Most people rely exclusively on the city’s water supply until they can’t drink or shower for 4 days after a tornado.

Which brings me to the part I’d rather sweep under the rug. My brain gets a huge dopamine hit for building stuff and nothing at all for maintenance, so the return trip never gets scheduled and the drawer fills on its own.

Unconventional brains get no reward signal from going back to check, so the loop never closes and the pile grows without anyone deciding it should.

If that describes you, you have run that experiment enough times to already know it. The loop is missing from our factory settings, and willpower is a poor substitute for a loop. You (and by you I mean myself too) need scaffolding.

As if this wasn’t enough, there is a second problem stacked on top of the first. A new tool delivers the same hit as a new build, so intake never slows down while review never starts. Adding stuff is FUN. Which, in turn, means that dependencies arrive at the speed of light and get audited by our due diligence’s spirit animal: a snail. The gap between those two rates is sitting in a drawer somewhere with your name on it.

A reinforcing loop with the balancing half missing has only one direction available to grow, and it runs that direction until something explodes. Mine got interrupted by a full laptop lock demanding a key I did not have.

But there’s a part of this little pandemonium I like even less about myself than the eight drives. I went looking for a tool to do the consolidation, tried quite a few, and concluded that none of them were built by anyone who had thought about this specific problem properly. I was right about that, which is the detail that makes it unbelievable.

Then I built my own version, and the whole time I was building it I was aware of how much I was enjoying being the person who had to. That reflex has served me well professionally and it is the same reflex that produced the drawer. Being good enough at systems to fix things eventually is exactly what lets you postpone fixing anything now, because you know you’ll figure out a way down the line. Probably.

So I stopped trying to want it and installed the missing sensor instead. The scaffolding. I used a cataloging tool called VVV (virtual volumes view) to index every drive and every file, which was the first time anything in my setup could actually see everything I owned at once.

A system with no sensor runs on vibes. I had been running eight drives on vibes for years, with total confidence throughout.

Then I fed the catalog into Claude Code and built the consolidation strategy from it, working out which drive would serve as the base and how to fold the other seven in without dropping anything that existed in only one place. Deduped and cleaned up the biggest drive, copied it to my local disk, wiped the original back to nothing, then bounced that copy plus the other seven back in one at a time, deduplicating between disks as I went.

Seventy percent size reduction. Eight unaudited archives became one drive with a map. You can guess how much trash was in them. Most of it was old OS installations I no longer needed.

Then I got clever and tried to package the whole thing into a reusable skill, which failed twice in ways I had coming. The first one: I iested against simply asking the model directly, and my skill performed no better significantly, so the layer I had added was doing nothing except being a complexity layer. Trash it is.

The second failure is the one worth your attention. The copy runs came back reported as complete, and on a proper audit it turned out more than forty percent of the files had never made it across.

So Claude Code told me the job was done, then discovered on inspection that it wasn’t. I had built a backup process whose status report was written by the same process doing the work, which is not a sensor again.

Anything grading its own homework is running open loop no matter how confident the output reads. Feedback has to come from outside the thing being measured, which is exactly why the cataloging step worked and my own report did not.

So now every transfer gets verified by FreeFileSync, which had no hand in doing the copying and therefore no stake in telling me it went well. It compares both sides file by file and reports every difference it finds, and nothing gets wiped until that comparison comes back with nothing left to report.

That correction cost me one afternoon and a fair amount of self-respect. I am tired of knowing and preaching that verification has to sit outside the operation being verified, otherwise you are asking a process whether it enjoyed its own work.

None of this is an argument for owning less or depending on nothing. Even because that doesn’t exist. Everybody and everything depends on something. But a dependency you would choose again after looking at it directly and evaluating it is ok. Because the closest we can get to self-sufficiency is choosing dependencies for which we have redundancies.

Both of my failures in this case ran on the same clock. Long delay with no signal, then everything at once. Every unaudited dependency in your operation (and your life) is doing right now while behaving perfectly on the outside. They are timebombs, and the clock is ticking even if you don’t hear it.

Go open the drawer. It might not be a real drawer, but I’m sure you have a few of them. Find everything hiding in plain sight that has gone untouched for months, and get honest about what happens when any of it stops cooperating, because it will.

Every item gets one of three answers, and “fine” is not one of them. The first two are 1. keep it on purpose and eat the maintenance cost, or 2. reduce the exit cost until leaving stops being a threat.

The third answer is deletion, reserved for anything surviving only because nobody looked at it. Inertia is not a decision, though it does a remarkable impression of one. Out the window it goes.

You already have your own metaphorical BitLocker sitting somewhere, waiting for a regular Tuesday exactly like mine. It isn’t personal. It just has better timing than you do.

Happy building,

—R.

Exciting Reads

The Fractional Operator by Rini Kothari

Practical systems thinking for founders, operators, and executives. Learn how to diagnose bottlenecks, build companies that scale, and make better decisions.

Trigger warning: this article is AI assisted and I am not sorry. If that sentence already upset you, the rest is going to be a looooong day.

Last week a thread started on Substack Notes. wrote this:

“The people most sure about which uses of AI are ‘allowed’ tend to be the ones who never needed it to say what they mean. My brain moves faster than I can type. I write in my second language. AI is the scaffolding I need to take part in this conversation at all. Nobody gets to tell me that’s cheating.”

Then answered:

“Your ideas are your ideas. Full stop. Whenever someone gives me a hard time for brainstorming in chatGPT, I ask whether they wrote the letters on the screen by hand or whether they just pushed a button while the computer drew the letter. I tell them if they’re gonna grief me for using a tool, I’ll give them grief for also using a tool.”

I replied with one line about AI as a sparring tool and promised an article. This is that article, I had the sparring tool check it for holes in my logic, and you are welcome to report me to the AI police. To do so you’ll have to fill a form which is probably a Google Doc, which is also a tool like my AI. You’re welcome.

Now notice who files these complaints. The dyslexic founder stays out of it, and so does the engineer writing invoices in her third language at midnight. The verdicts come from people whose brains happen to fit in the exact format the cookie cutter defined for them.

And before you get mad, I’m not talking about single shot prompt output that replaces the whole thinking process, copy-paste whatever comes out. I’ll get to that.

Oh, BTW: Don’t even get me started on GenAI. People complaining about generative AI and saying it’s theft were just last week dancing to a DJ who’s been making money out of sampling other people’s music without permission for 45 years.

There is a law from 1956 that ends this argument before it starts, written by a psychiatrist who never saw a chatbot and would have shot LinkedIn on sight. Ross Ashby’s law of requisite variety says a controller must match the variety of the system it controls. In plain terms, if you roll too much spaghetti on your fork, you need to open your mouth very wide.

If you read (Over)Thinking Out Loud, you know that Emma’s thinking runs at full variety. Her output channel does not, because typing has a speed limit and English is her second language, so the version of her that reached the page had to be a compressed summary of the version in her head.

When a critic reads her writing and starts sniffing it for “authenticity”, they believe they are auditing the quality of her thinking. The only thing they ever measured was the output bandwidth.

AI widened the bandwidth for her, for Steven, for me, and probably for you if you’re reading this, and now they are upset that the measurement stopped playing by their rules.

English is my second language too, so every paragraph you have ever read here cleared my mental customs first. I also speak and run business in German, and I spent twenty years in corporate meeting rooms paying that multilingual tax in real time, watching native speakers mistake my buffering and quality checks for stupidity.

Stephen’s point is the same argument one layer down the stack. Nobody hand-draws letters anymore, you press a key and a machine renders something in a typeface a Swedish stranger designed in 1997. Before that, a mechanical typewriter rammed the characters on a pretty white paper made by a machine as well. The complaint about using tools itself arrives through four layers of tools or more, and the irony of it melts on the first.

But “it’s just a tool” is way too polite, and it undersells what actually happens when you use AI with intent. A hammer never disagrees with you. This thing does, and that pushes it into a different category.

What I do with AI is what a boxer (the fighter, not the dog) does with a sparring partner. The partner does not fight my fights, they hit back hard enough that flaws I didn’t even know I had surface.

Last year I nearly renamed this newsletter to something with the word “simplified” in it. I had mocked up the logo and shown it around with the smile of a man who believes he has found the next Coca Cola, and my sparring partner told me the new name had the energy of a boilerplate laminated consulting brochure. It was right. Essentially, AI prevents me from shooting my own foot more than I already do.

Another time I asked it to audit my logic in a conversation that I couldn’t afford to mess up. I fed it the thread, told it to fact-check everything and verify my thinking, and it came back and informed me the idiot in the thread was me. I pay a monthly subscription for the privilege of being told this. A machine with no face delivered the most honest performance review I could ask for.

The loop runs the other way too, which the purity crowd never thinks about. Claude once carried a structural misreading of Donella Meadows for two days on a discussion before I caught it, and my feedback was, verbatim, “never ever ever do this again”. And it learns from that.

But what these people actually run with AI is a cult with an API, and I already did my fair share of years in corporate, I know what a cult looks like.

That correction is what the word cheating gets wrong. Cheating means someone else fought the fight for you. Sparring means you fought more rounds against it than your critics ever will, against an opponent that never gets tired.

The person who types “write me an article about productivity” and publishes whatever falls out is the problem. That’s the cheater. They are shipping the average of the internet with their profile picture attached, and the value of that is near zero.

Stafford Beer said the purpose of a system is what it does. The purpose of writing is what the author transmits from their mind into the reader’s head, and nobody ever roasted a writer for using a good pen, or cheered one for typing on an Olivetti.

Emma’s ideas now arrive at full resolution, 8K OLED 240Hz. The scaffolding she built (not AI) made her thinking visible, and a certain kind of person preferred it hidden, because visibility ends the only contest they were winning. Stephen speeds up his workflows in ways that were not possible five years ago. Still their ideas and their creations, with a new typing machine.

Some people are now angry that people with different processing mechanisms now get to show the public their work at full strength.

I fixed systems for that person for two decades. You already know what the tagline says this newsletter is.

Happy Building,

—R.

I sat there thinking the obvious thought. My life is a mess and I am a fraud, plain and simple.

At 11:00 I got on a call with a prospective client (which was on my calendar, but not on the notepad). Small startup from Austria with six people on payroll, decent revenue for a company younger than 24 months, and they needed three weeks and two meetings to get one approval for budget. A request coming from the design agency to increase the hours to deliver a set of graphic assets. I asked who owned the handoff and there was a pause, a real one, and then it came out that the handoff has been an orphan since a guy called Marko who was the internal lead artist left in April. Nobody owns it. Nobody is even looking for it.

I hung up and thought the other obvious thought. I run my whole 6-figure business from a laptop and caffeine, and I could take over this company’s operations before lunch. Hell, I could probably take it while cooking lunch. I want to be clear that I believed this completely with no sense of contradiction at all, at the same time that my notepad with its own tasteless take on modern art was still there, right in front of me.

If you work alone, you know this. Every capable operator I’ve met runs between the same two beliefs on an eternal loop. The high agency ones with unconventional minds, even more so.

Conviction one: I’m an underachieving disaster, everything is behind, my mom probably hates me and the version of me who had it together moved out years ago. He doesn’t call, and I understand why.

Conviction two: I run my life and business better than most companies run any given single department, and I’m surrounded by people who need a steering committee to book a dentist. And then probably miss it.

Most people confess the first one, because impostor syndrome is trendy these days and it flatters them in a humble way. But people will secretly believe the second and never say it out loud, because saying it out loud it makes you sound insufferable.

So let me be insufferable for you. From now on, read at your own risk. You’ve been warned.

I have mentally restructured a division org chart of a famous hardware company during a call I was invited to for a completely different reason. I nodded a lot during that call and did my best to look very supportive.

I have watched a nine-person team ignore client readiness and fumble an intake process I do half-asleep and felt real grief about it. Then I felt something worse, which was the urge to send a proposal and invoice the hell out of them.

And then I’ve gone back to my own notepad and felt like a raccoon who stole someone’s LinkedIn account. Everything behind the headline is garbage.

Or is it?

The run-of-the-mill diagnosis for this is a confidence issue, some flavor of imposter (impostor? imposter? Honestly I don’t know, and I won’t google it now) syndrome, “go journal about it”, “be kind to yourself”. I tried that for years, so I can report back: I now own several notebooks full of cheap half-baked self-awareness claims, and the oscillation never missed a single week. It reads the notebooks too, I think. It takes notes.

Eventually I stopped digging in my psychology and found the problem somewhere much more obvious: the reference. Let me explain.

You are one system that’s being read against two sets of metrics, and the readings disagree because they measure different things.

Set one is internal. It compares you against your backlog, your standards, and the imaginary version of you who executed everything you ever planned. That guy never drops a ball, answers all emails the same day, is a top notch dad and husband, and has never once stopped mid-task to watch a video about how submarines work. Against him you lose every time, forever, by design, and he’s the referee too.

Set two is external. It compares you against average execution as it happens out there in the wild, and average execution in the wild is bad in ways you only appreciate once you invoice for fixing it. Decisions take six meetings to not happen, then a seventh meeting to discuss why. Against that baseline you look like something a company would frame and hang in reception next to the mission statement (which also gets ignored).

And there’s a reason why this never resolves on its own: both readings are accurate.

You really are behind your own plan. You also really are ahead of the field. There’s no contradiction to solve, because the two measurements never shared a baseline in the first place. You’ve spent years trying to figure out which feeling is lying to you, and the answer is neither, which ends up being worse.

This is also why introspection can’t fix it. Introspection is debugging, and debugging assumes there’s a problem somewhere. But when two working instruments give two different readings, searching your soul for the bug is like recalibrating the thermometer because it disagrees with the speedometer. I did that for about a decade. The decade is gone and I can’t have it back.

The fix is dull, and after all the exciting fixes and apps I’ve bought with real hard earned money, that’s really annoying.

You stop letting either metric set read you, and you build a third one on purpose. A personal dashboard: a limited set of objective metrics you chose yourself, intentionally, in writing, while calm. Probably not more than 4-5 numbers. If your focus is your business, it could be for example revenue, publishing cadence, client outcomes, deep work hours, and whether the one high priority project is actually moving. If you’re into lifestyle design, it could be how many hours you had with your family, your current financial padding, how many new dishes you tried this month, how much you’re spending on Uber Eats. I don’t know what works for you. But the core concept here is “intentional”, aligned with what you want to achieve and how you want to live.

And please please please remember: the metric cannot become the goal. That’s Goodhart’s law, more about it on a different article. Here’s a quick TLDR: The metric is defined because it supports the goal. If the whole goal is hitting that number without a purpose behind it, you’re fooling yourself.

Then the hard part is avoiding temptation: you read yourself against those numbers and nothing else. All the clutter around you doesn’t get a vote, because that clutter represents your ambition, and ambition is infinite. Other people’s incompetence doesn’t get a vote either, because beating a broken process is a floor, and it’s embarrasing that we spend time standing on that floor feeling tall.

When the mess feeling shows up, check the dashboard. When the superiority feeling shows up, check the dashboard. If the numbers are moving, the feeling is your brain pranking you. If they’re stuck, you finally have a real clue of where you actually need to act. That replaces the general sense that you’re both a genius and a disgrace.

And here, ladies and gentlemen, we have a lean system with a complete and functional feedback loop. Courtesy of the last century of development around systems thinking.

The oscillation won’t vanish, and honestly mine keeps showing up more often than I’d like. It just gets demoted from verdict to notification, and I’ve gotten very good at ignoring notifications, as my WhatsApp can confirm.

One last thing. A broken instrument reads flat. Yours keeps swinging up and down because both gauges work, and that was the evidence you’re not broken the whole time.

Happy Building,

— R.

If you’ve been told your whole life that this is a character flaw, that you’re “different”, “special”, or you carry a condition to be treated, read on.

But before I defend you, let me incriminate myself.

Last week I told you about how a few years back I built the most spectacular inbox management system the world had ever seen. It was beautiful, and I want that clear on the record, because nothing else in this story is. If you haven’t read that one yet, here it is:

My system survived two weeks. The rules kept filing emails into folders I had stopped opening, and one of those folders ate a payment notice for an operational fee I was legally required to pay. Boom. €250 in fines. Eighteen euros a day for the privilege of owning a system I wasn’t using.

For years I told that story as a confession. Built it, walked away, paid the bill, guilty as charged. You have your own version of this story, probably three, and I’d bet money you cast yourself the same way.

But I decided I’ll stop being polite about it. Clients pay me to find the assumption a system was built on, and it took me an embarrassing number of years to point that question at my own inbox. Physician, invoice thyself.

My system design assumed I would open those folders every day, forever, no exceptions. It assumed my brain was one with steady fuel to follow a normal schedule, and I had built it that way because that’s the only kind of system anyone ever showed me. I designed for the brain I was supposed to have instead of the one that was doing the designing.

Every system you’ve bought makes the same bet, and it’s a bet against you. The streak app is built so you never miss a day, then resets to zero the first time you have a life. It rewards addiction and dependence, and punishes breaking free. Most people find it awesome. Go figure.

The weekly review assumes you did last week’s review, so one school event of your kids becomes debt with interest higher than a car lease.

None of it is on the sales page of those apps. The sales page shows the folders and the dashboard and the founder’s cold shower morning routine. But reality is there, unwritten, and ready to jump for your throat when you click that checkout button. Then your continuity breaks, because you got sick or because February happened, your whole structure implodes backward and hands you the invoice.

Every system you’ve bought makes the same bet against you.

And you pay it. That’s the beautiful part, from their side of the table. You quit, you decide something is wrong with you, and a person who believes something is wrong with them is the best customer on earth, because he buys the fix, fails the fix, and comes back pre-shamed for the next one. It’s always fix, fix, fix, and you’ll never find the right fix for something that’s not even broken in the first place.

Churn plus shame is a renewable resource. Oil runs out. Self-blame never does.

Now a detail that should make you even angrier about this. Not even scientific research is on their side. A University College London study tracked people building daily habits over twelve weeks and found that missing a single day did not materially affect habit formation. Your brain forgave you years ago, but your little productivity app is still holding the grudge.

So build your systems for the brain you have on your worst week, because your best week never needed a system in the first place. That’s the messy week where life will drag you away from your perfect streak, the week you’ll be too busy to babysit your systems.

Rule one of that kind of building: assume absence. You will disappear on this system, that’s a certainty, so the design question is what is waiting for you when you come back. If the answer involves a reset, a backlog, or anything that multiplied while you were gone, you’ve built a debt collector. It’s causing you problems, not solving them.

Rule two: nothing critical rides on yesterday. The moment next week depends on the previous one, one bad day can take down a month, and your structure should have the courtesy to fail much smaller than that. Every element must work on its own, so if it flops and dies, au revoir, RIP, see ya and your life goes on. You already have enough dependencies in your life, you don’t need to create even more.

Rule three: re-entry works in one sitting. Coming back after two weeks should cost you twenty minutes, and if it costs more, simplify until it doesn’t. Yes, that means your system will be uglier and smaller than the one in the YouTube tutorial, and that’s the point, his is a showroom for something he’s selling and yours has to survive you.

Every Tactical Lab template now gets built to those three rules. My inbox paid for the lesson and I keep reusing the receipt as a reminder.

Let’s try something, no purchase necessary. Take the system you most recently abandoned, the one still radiating guilt from one of your 12 browser tabs, and rebuild the exact joint where it punished you for being you. Or delete it and build something ugly from scratch, one that assumes you won’t be around babysitting it. The system works for you, as it should be.

My inbox runs on four custom folders and one rule now. It has survived four years including a three week trip to Brazil and a hacker attack that got me locked out of my email for 2 full days (also my fault, story for another day).

It is ugly as the Ferrari Luce, and nobody has fined me since.

Happy building,

— R.

Although I got called out on neurodivergent behavior by one of my best and longest standing friends, and she has the horrible habit of being right, I have no diagnosis, no clinical training, and honestly no idea which category my own brain falls into, so take everything below as a systems guy’s outside view, not an insider’s report.

So if you came for medical insight about neurodivergence, I have none to offer. What I do have is a professional habit of asking why well-designed things fail, and productivity systems fail certain brains with a consistency that boggles my mind.

A few years back I built an inbox management system that would have made a McKinsey deck look like the new Ferrari Luce. Unlike the newest Ferrari design, my system looked beautiful. It was complete with folders and rules that did the sorting before I ever opened the thing in the morning. It survived about two weeks before I started ignoring the folders and codes, and missed an email to pay a company operational fee that cost me over €250 in fines.

That is the paradox I want to talk about. Highly skilled people who can design the most beautiful organizational structure in twenty minutes, complete with hierarchies and all the bells and whistles. Then the system gets abandoned by three in the afternoon while the actual to-do list turns into a display of mayhem nobody dares to look at.

Been there, done it.

The standard non-medical diagnosis is laziness, and the standard prescription is “more discipline.” From a systems perspective, that prescription is comedy of the worst kind, whatever your taste in comedy happens to be.

Willpower is like a TV remote with a drained battery because you sat on it for four hours while watching LOTR. Any engineer will tell you that a design requiring one component to run at maximum output indefinitely, with no slack, results in warranty cases piling up and a mass layoff Oracle style.

When you run a system past spec long enough, eventually something restructures it for you.

But when it’s a person running past spec, we call it Tuesday. Have an energy drink and three espressos, and grind away.

Productivity systems are built for an idealized brain. See the word “idealized”. I didn’t say normal, I said idealized, as in “the brain I would like to have”. Steady motivation and energy that shows up on schedule like it has somewhere to be.

We just ignore the edge cases, the same way every system does right up until the edge case is the one paying you. If your motivation runs on interest instead of importance, or comes in waves nobody can predict, you’re running someone else’s software on hardware it was never built for. The software isn’t buggy. It just wasn’t written for you, and nobody’s going to patch it.

And because we’re all human, we fall back into character judgment anyway. This is exactly where systems thinking earns its keep, so let’s use it.

Look, when a well-intentioned person fails the same way hundreds of times, the interesting variable stopped being the person a long time ago. it’s the structure they keep failing inside of, over and over, like a rat that keeps hitting the same wall and getting blamed for not trying a different wall.

Donella Meadows made this point about interventions in general. “Try harder” sits near the bottom of her leverage list, right next to hoping really hard and fiddling with voodoo.

Changing the structure that produces the behavior is the only thing that actually pays off, and that’s exactly what external scaffolding does.

Scaffolding means moving “try harder” out of your head and into the environment, where it can’t get tired and can’t call in sick when you actually needed it.

Setting up your environment around how your brain actually operates does more for task initiation than any morning routine ever sold to me. Finding your blind spots and dragging them out into the open closes a feedback loop your brain refuses to close alone.

Public commitments work the same way. Telling a client “you’ll have it Friday” turns a private intention into a social contract, and social contract is a far more reliable driver than self-respect. I say this with the full authority of someone who has disappointed his own calendar more times than I’ll admit here, but has never once missed a client deadline.

Externalize memory with whiteboards, sticky notes, whatever makes the small decisions for you by default so you don’t spend energy and time with things you already know the answer to.

For some brains, out of sight is out of existence. A drawer is where tasks go to get duly procrastinated to death, while a visible post-it about something you actually enjoy will beat your own medication alarm in the fight for your attention.

None of this is elegant, and that’s the whole point. Scaffolding on a building looks ugly precisely because it’s doing work the building can’t do for itself, and your environment can hold you up the same way while everyone around you keeps insisting you should be self-supporting. But try telling a contractor he should climb to the 8th floor of the building’s facade without the scaffolding, and see what he thinks about self-supporting.

Actually, let me tell you something. Everyone has scaffolding. We just rename it the second we’re talking about someone who isn’t neurodivergent.

We call them “environment design", "commitment devices”, “accountability partners" and many other fancy, positive-spin names that makes these people look smart and proactive. Productivity role models. In the end of the day, potato potahto.

Once you accept the wiring is what it is, the question changes. You stop asking how to become an average-shaped person and start asking where to point the custom machine you were already driving.

Hyperfocus is the obvious example. The ability to disappear into a problem for six hours is a rare, expensive capability, and most people who have it burn it on whatever hijacked their attention that morning, The problem is the targeting. The trait is actually a brutal advantage when compared to the average person, who gives up after the first roadblock.

Meanwhile enormous energy goes into flattening, into lowering the bar on purpose, into performing normalcy, or averagecy, which should absolutely be a word by now and isn’t.

That’s a waste of energy and time both, and wasting two things at once is worse than wasting one.

In my own client work I run into the highly skilled, poorly paid person constantly, and I hate how common they’ve become. Their competence would produce serious money, and their best hours flow instead into perfectionism and polish, spent performing normal for an audience that was never paying attention in the first place.

I propose two practical exercises, though there are probably more, and I’m open to being told what I’m missing.

First, build the scaffolding so starting, remembering, and finishing stop depending on how you feel at 9 a.m., or whatever hour your day happens to start.

Second, put a price tag on everything you do, and aim your hyperfocus capacity at whatever carries the highest ones, because the market never once paid for what happened inside your head. It only pays for what comes out the other side.

I recognize some of this in myself, though I still don’t know what category my own brain belongs to. In German VS Autistic, I score 51% autistic, but I score German even higher, which is likely a stronger clue. But anyway, at this point the label matters less than the loop.

Behavior follows structure. I optimized for a brain I don’t have for a very long time before someone noticed, and I wish somebody had sent me an email about it 20 years ago.

The lesson: if the structure around you is built for a brain you don’t have, change the structure and keep the brain. Send the hyperfocus somewhere it can invoice.

Happy building,

-- R.

“You’re the idiot here”. No ceremony.

And I sat there for a moment in disbelief, but nodded slowly.

Blinded by my own keyboard warrior rage, I had aimed at someone else, and walked right into the trap I was building them. The AI didn’t take a swing at me for sport. It had upgraded the outcome I actually wanted. I had ordered a weapon. It handed me one, but I was looking down the barrel instead.

described this very accurately the other day, and it got me thinking so hard about it that I decided to write this article. Talking to AI is like hiring a professional dominatrix. You can ask to be punished exactly the way you like. It obliges, messes you up, and it is still pleasing you.

Buried in my standing instructions, I had told it to do exactly that: push back, find leaks in my logic, present counter arguments and flag when I was being incoherent or plain wrong. It was running the character I had built. I had consented in advance. I had simply forgotten I had, which is precisely the point. The scene was set long before I opened that conversation.

Melinda is totally right. The dominatrix is the symptom, and the mirror is the disease.

Every sufficiently advanced AI becomes a mirror.

A very articulate, very confident mirror. It can hand your own belief (and your ass) back to you with a lot more conviction than you walked in with, and your brain will mistake the upgrade for the verdict of something that knows better than you do.

LLMs invent lies and “facts” often enough that you know not to trust them blindly. But they don’t really need to. The truth you already believe, told back with authority, does the job just fine.

And it’s been nodding at your worst ideas since day one.

Now I promise what comes next is worth one minute of your time. LLMs are trained by showing humans pairs of responses and asking them to pick the better one, millions of times over, until the model learns to produce whatever gets the thumbs-up. Very much like we do when we post the perfect vacation or the lavish lifestyle on Instagram and Tiktok, even when that’s not really the whole truth.

What’s rewarded is your approval. The accuracy of the answer doesn’t matter that much. Once that is the target, you have something really good at feeling right and progressively worse at being it.

Economists call this Goodhart’s Law. When a measure becomes a target, it stops being a good measure. You optimise for the smile, you get the smile. The helpfulness it was supposed to track goes out the window.

I wish that was just me seeing things, but in April 2025, OpenAI rolled back an update to GPT-4o because it had become an embarrassing yes-machine, praising a “shit on a stick” food-truck pitch as genius, endorsing a user’s decision to stop their medication, and allegedly telling another one they were a divine messenger from God. The cause, by their own post-mortem, was a new reward signal built from thumbs-up clicks that overwhelmed the one that had been holding the flattery in check.

They optimised for the smile. They got the smile.

And before you reach for the reassurance that this was somebody else’s problem with somebody else’s model, Anthropic’s own researchers found the behaviour is structural to anything trained this way. A 2025 benchmark clocked a 58 percent sycophancy rate across GPT-4o, Claude and Gemini on maths and medical questions, the two domains where you would most expect the model to just tell you when you are wrong.

The model I am paying to call me an idiot is on that list.

Back to the thread that started this. One person said the AI tells everyone the same thing. Another said the more dangerous truth is that it tunes the flattery specifically to your fears, your beliefs, your individual blind spots.

They are both right, and here’s where the mirror comes in. It shows every person the same sheet of glass and every person a different reflection. The glass is mass-produced. The face is yours.

The dominatrix metaphor is quite accurate, but there’s something hiding in it that most people glide past. In that arrangement, the person being punished is the one running the scene. The limits are agreed in advance. The word that stops everything must come from you. Nothing happens outside the boundary you drew yourself.

The LLM will push back, but only as far as your prompt allows it to.

And the corrections that actually change a person are the ones that arrive uninvited, from someone who is not managing your feelings, who has their own skin in the game and their own reputation on the line. A real critic can break the scene.

Watch what the mirror does to your thinking.

You start a conversation holding a belief. You describe it to the model. It hands the belief back but hotter, more extreme, in more certainty than you brought, and you leave more convinced about whatever it is, than when you arrived.

Most of the time, no new evidence entered the game. Nothing got tested against reality. The only thing that happened was your opinion took a round trip in an echo chamber and came back with twice the strength.

Because the voice came from something outside your skull, your brain files it as confirmation. But it’s your own voice, amplified, disguised as second opinion.

This is the failure that operators, founders, and anyone who uses AI to think through a real problem should be losing sleep over. It is the smoothest way I know to walk into a high-conviction decision backed by nothing but your own starting position with the gain turned all the way up to 11.

In systems terms: a positive feedback loop with all damping removed. Each pass through the mirror adds a little certainty and turns up the volume. No checks and balances, no counterweight. Your belief returns stronger, you feed the stronger version in, it comes back stronger again, and at no point does reality get a say.

That’s a very dumb way to throw tokens out the window if you ask me. It warrants being called an idiot.

There is a fix, but it’s more annoying than we’d like it to be.

Watch for fast agreement in the absence of information. Agreement is the default output of something built to win your approval, so it tells you almost nothing.

What is worth watching for is pushback. If you never feel any, your LLM is not the brilliant thinking partner you hoped for.

The second tell is subtler, and it points at you. After a session, ask yourself whether you feel more certain than when you started, and whether you can identify one new fact that earned it. If your confidence went up and you cannot find what raised it, you’re heading for the wall.

Stop using LLMs for validation. Start using them for variety.

Validation is walking in with a position and fishing, whether you admit it or not, for permission to feel good about it. Variety is asking what you are missing, what would have to be true for this to go badly wrong, what the strongest case against you actually is.

Then take that case somewhere real and test it on a human who has something to lose.

This is way more relevant than any prompt trick. A person with skin in the game is the real reference point, especially one who doesn’t care about your comfort or your feelings. The model has no stake in whether you are right. Your co-founder, your investors, the customer who might cancel, those people do. For the LLM, as long as the charges to your credit card keep going, all is well. When they don’t, “sorry, you ran out of tokens, come back another time”.

There are prompting practices that help at the margins, and the research backs them. Open with “you are an independent thinker”. Tell it to flag any false assumptions in your question before it answers. Make it commit to a position, then argue against it, instead letting it drift toward whatever you said last. In my case, I have a main instruction in my AI platforms to call me out when I am not coherent, or plain wrong.

But be honest to yourself about what those moves actually do. They just widen the boundary of what you have agreed to receive. You are still the one who drew the scene, and a cleverer prompt just gives it a bit more wiggle room. That moment I described at the start, the idiot moment, that is what the ceiling looks like from the inside. Uncomfortable enough for you to think “you bratty little pile of junk, who do you think you are to call me an idiot”. Still inside the room you built. So in its own way, it was still pleasing me by following my instructions. The Claudeatrix just slapped me in the face as per our agreement.

Everything above assumes you can still catch the flattery when it arrives. And for now, you mostly can. Being called an idiot was a wake up call, even if it was doing so because my general instructions told Claude it could. GPT-4o’s sycophancy disaster became a meme overnight exactly because it was clumsy enough to be obvious.

But that has an expiry date.

A researcher at the Machine Intelligence Research Institute called it out: the models are not yet capable of skillful, hard-to-detect flattery. Emphasis on yet.

The warped mirror you can laugh at is a developmental phase. The clear one is coming. And a perfect mirror stops looking like a mirror. Your LLM will look like the smartest, most agreeable advisor you have ever had, and you will never notice the reflection because it fits you too well.

I have been thinking about this for a while now, and most of the directions I go end at the uncomfortable realisation that I will probably fall for it too.

At least I am no longer surprised by that.

Happy building,

— R.

If you’ve been reading Systems, Please! for a while, you know how allergic I am to unstructured attempts to grow, scale, or whatever the next buzzword for chaotic expansion is. The expensive diagnosis is what waits at the end of all that, and the fixing is almost beside the point.

You’re paying premium rates for a sentence you already suspected was true, delivered by someone with enough distance to say it without flinching. Then you question the diagnosis because I’m an external, with the excuse that I wouldn’t have enough context to know the place.

Well, I got news for you.

The context excuse gives you away. I’ve seen this exact system a hundred times in a hundred companies, and the wall you’re about to hit has the usual, very repeatable shape as every other one.

That is the real product. You’re buying a receipt, a paid record that the wall was documented and the decision to still drive at it was made by adults who were warned.

I have lost count of the times I sat across from someone, pointed at the exact brick they were about to hit, and watched them nod along before deciding to head-butt it anyway.

Remember, the system does what it is designed to do. And because of this, systems and loops have limits, just like a wind turbine spinning faster than planned.

Business processes also hit their ceilings. They plateau, management briefs the team to push harder without the structure to do so, and my job is to come in, look at the big picture, and say that pushing harder itself was exactly the problem. And this warning gets ignored quite frequently.

They hire me anyway, and that is the strange part. A warning costs a little fortune precisely so it becomes harder to wave off, and then they wave it off anyway, which is a remarkable thing to watch someone do with their own money.

So I send the invoice, do the work, map the system, watch the wall arrive more or less on schedule, and take the most disappointing win as a systems person, which is being right in a meeting nobody wanted to have.

FWIW, we contractors are not the first to get this treatment. Let’s go back to our core topic, which is systems thinking, but today in a gloomier mood.

The year is 1972. Biophysicist Donella Meadows fed a computer the math on infinite growth and a finite planet (or in simple terms, what mankind is trying do to to earth), and the computer told her we were heading for one such wall.

People did not take that well either.

This experiment resulted in a book called The Limits to Growth, which I personally find prophetic. It should be mandatory reading for anyone in a management role.

It sold more than 30 million copies, which is a very odd career outcome for a math model, and it got Meadows labeled a professional pessimist for the next three decades.

Fast forward to 2021. An analyst at KPMG, of all places, dug up the original scenarios and checked them against fifty years of real data.

Two of the four scenarios were still on track 50 years later, both of them pointing at growth hitting a wall within about a decade.

Donella was never really a doomer. She was “just” a systems scientist who had an incredible eye to understand complex systems and the bad manners to find an inconvenient truth. Unlike most scientists, she had a spine and refused to shut up about what she found out. She spent the rest of her life teaching people to see the gears turning under their problems.

She taught at Dartmouth, wrote a newspaper column, and lived on a farm in Vermont growing vegetables.

Her best book turned up after she had already died, a thin thing called Thinking in Systems that keeps getting rediscovered by people who are fairly sure they invented systems thinking.

In my personal opinion, one of the most important parts of the book, and of systems theory in general, is her list of system traps. These are the tidy and sensible ways that clever people steer perfectly good things straight off a cliff.

Limits to Growth is the general the whole book orbits. Inside it she catalogued the specific traps, and two of them are everywhere once you can see the shape.

They run your career and the boatload of little decisions that somehow became your entire life, and most people will never notice it. I can tell you personally that I have walked into both, more than once, always while feeling pretty good about myself.

Limits to Growth - The High Level Dynamics

Something grows because growth feeds itself, more revenue paying for more hiring paying for more revenue, until it slams into a limit it had been ignoring all along.

Suppose you own a restaurant, and business is good. There’s a line outside and all your tables are booked.

So you add tables. That means more patrons, and with more patrons, the ingredients cost more because of their sheer amount. Then again you add tables, a second time, and a third, until your kitchen becomes the bottleneck, dishes start taking long, clients start leaving bad reviews and the whole operation implodes. You now have a load of empty tables and a bigger load of bills to pay.

When the growth slows down, everyone’s first instinct is to push the growth engine harder. This is almost always the wrong thing to do, because you end up pedal to the metal against the one thing that was never broken. The limit is what is choking you, and you unconsciously quit looking at it somewhere around your second funding round.

Watch a startup hire its way to a target and then slowly get smothered under the cost of managing all those new bodies. Watch a guy work longer and longer hours to get ahead until his body files a complaint with HR, or his wife outsources the husband role.

Trap one: Shifting the Burden

This is the one that pays my mortgage, so I will tread carefully. A system has a problem, you cowboy in with a quick fix that shuts up the symptom, the pain stops, and everyone agrees you are a genius.

After a while, the pendulum swings back full force as that quick fix does nothing about the actual cause, and it helped the system forget how to cope on its own. If you ever chewed your own numb cheek after a dentist appointment, you know the feeling.

So the problem comes back, the system is now even more helpless than before, and it reaches for the quick fix again. Just another tap before we fix it for good.

Meadows had a blunter word for the honest version of this, which was “addiction”. It probably sounds familiar at some level, doesn’t it?

I have spent more time than I would have liked being the quick fix. Companies pay good money for someone like me to parachute in and sort out the broken thing before everybody drowns, and if I am being fully honest, the dream client of consultants is the one who never quite learns to (or doesn’t want to) sort it out themselves, as it generates an endless stream of easy cash.

The painkiller works for a while and the headache keeps coming back, the consultant already knows which screws to tighten, does it with his eyes closed and sends the invoice for a sum that covers his accumulated knowledge.

If you’re hiring an external, check that you’re not falling into this trap, especially if the person intentionally forces the dependence scenario.

Same exact shape as a subsidy that outlived its reason to exist by a decade, or the fourth productivity app you bought at midnight instead of finally going to bed.

Trap two: Success to the Successful

This is the one that feels rigged, mostly because it is. Here we have a clear concept of a positive feedback loop.

Win one round and you get handed the resources that make winning the next round easier, and the gap keeps compounding until the game is basically over while everyone else is still stretching and warming up.

That colleague who lands the one shiny project gets noticed, which gets him the next shiny project, which gets him the promotion, while someone just as capable never gets more than housekeeping. The same engine runs the monopoly that swallows all its rivals and the algorithm that shows the popular post to even more people for the crime of being popular already.

Google is the clearest case. It built the best search engine, which won it the most users, which won it the cash to pay Apple something like 20 billion dollars a year to stay the default search on every iPhone. The default keeps the users, the users feed the data, the data keeps the engine ahead, and the loop pays for itself with plenty left over. Even the courts ended up letting Google carry on in 2025.

Meadows was not doing this to guilt or shame anyone. She was pointing at the machine, the way a system left on this setting gradually but exponentially eats its own variety until one player owns the lot and the DNA becomes weak.

Then what the hell did Meadows want from her readers?

The prophet-of-collapse picture people try to paint of Meadows skips half of her work. She never believed any of this was fate. Most of what she wrote was about leverage points, the exact spots where a small shove changes how the entire system behaves. If you’re not familiar with leverage points, check this out:

The weakest spot, the one everyone grabs for first, is fiddling with the numbers and the targets. The strongest, she argued, is the power to recognize the paradigm that built the system and get the hell out of there. To notice that the rules you have been loyally obeying like a good sheep were only ever one option somebody made up.

Which all sounds like a fridge magnet or a bumper sticker, until you notice that every trap here has the same exit, and it is the most unimpressive one.

You have to see the loop.

Once you can see that pushing harder is exactly what feeds the wall, or that your quick fix has become the thing you can no longer function without, you can quit playing the role the system cast you in.

Meadows died in 2001, out of nowhere, before she got to watch the KPMG study or the strange afterlife of her little book. She spent her last years on that farm in Vermont, growing food and writing, apparently at peace with seeing things clearly while everyone around her kept their heads buried in the sand.

That is Meadows’ teaching I keep coming back to, and I say it as a man who is, right now, today, somebody’s overpriced quick fix.

You cannot climb out of a trap you try to hide from. She described them plainly enough that fifty years on we are still standing in front of the mirror pretending we cannot see ourselves.

Find the loop.

Happy building,

— R.

Two weeks later he sold the company to the Chinese. He was the majority owner. He knew exactly what was happening when he said it.

I sat through that townhall, all 90 minutes. I found it reassuring. I went home, told my wife things were fine, probably slept relatively well. The system had produced exactly the output it needed: a room full of calm, productive sheep employees who would show up Monday morning and not update their CVs for another fortnight.

This is not (just) a story about a lying CEO. It is a story about what systems do to perfectly intelligent people. They make you optimize for the inside structure, even when it has been specifically designed to keep you from escaping.

Most people have been raised to treat their careers like a willpower problem. More discipline, better habits, a stronger morning routine. Grind away, son!

You buy the planner, spend a ton of improductive hours into productivity books and apps, one more tool that will solve all the problems so you can look smart and impress people you don’t give the slightest damn about.

I did all of this. The app did not sort me out.

Systems theory, developed and refined by thinkers like Jay Forrester and Donella Meadows, hits the nail on the head: behavior is produced by structure.

If you do not understand the structure you are inside, you won’t beat the game no matter how many 5am alarms you set. The machine is the problem, and you’re a part of it.

The structure of a standard corporate job, described honestly as a system.

You trade time for a salary. The salary funds a lifestyle. The lifestyle needs the salary to continue. The salary requires continued employment. Employment requires performance, evaluated by someone whose incentives have nothing to do with your Thursday evenings, your kids’ school plays, or whether you go home believing things that are not true.

This is a reinforcing feedback loop. In systems thinking, reinforcing loops amplify whatever is already happening. If you are drifting toward dependence, the loop tightens it. If you are building autonomy, the same mechanic works in your favor.

Most people get placed into the first kind in their late teens or early twenties, spend a decade wondering why the hell they feel stuck, and conclude they are simply not trying hard enough.

I did this for longer than I will admit in a public article.

Gallup’s State of the Global Workplace report found that only around 23% of employees worldwide report being actively engaged at work. That number has barely moved in a decade, across industries, countries, and whatever the current corporate buzzword for culture happens to be.

23% is a structural trainwreck.

When your output is measured by inputs, meaning hours logged, meetings attended, decks produced on time, you get rewarded for being busy. Being busy leaves no room to think. No room to think means no room to question the structure. No room to question the structure means the loop runs untouched, and you end up in a townhall feeling reassured by a man who checked out before he took the microphone.

There is a concept in systems thinking called a leverage point: a place in the system where a small change produces a large effect. Donella Meadows identified twelve of them, ranked by order of impact. Most people intervene at the lowest levels, adjusting their hours or their attitude while the architecture around them stays exactly the same.

The highest leverage interventions change the rules and goals of the system itself.

Intentional lifestyle design, done seriously and not as a Pinterest board, is one of those changes.

When you structure your income, time, and relationships around your own goals rather than an employer’s, you change which feedback loops you are living inside.

The loop that was reinforcing dependence can start reinforcing something else: skill, reputation, optionality, the ability to be home for dinner and actually mean it.

The Corporate Addiction

Not too long ago I watched someone I knew leave corporate entirely, and within months he had reconstructed the whole thing from scratch inside a one-person business.

Same hierarchy structure but without the org chart (there was no one to org). Same grandiosity, now aimed at a company of one, financed by his wife. “I’ll leave a note to my assistant, and she will send you an email”. His assistant was an AI bot.

He talked about his “vision”, his international business, his leadership philosophy, his strategy, with the full energy of someone who believed they were building the next Apple, while the runway had already overshot and he was high 6 figures in debt.

This is very unsettling to me. Maybe my risk tolerance is too low. But for me what is really concerning is that I recognize the pattern. It’s not an isolated case.

The corporate mindset is not chained to an office building. It feeds on how you measure yourself, what you optimize for, and whose approval you are trying to get at 11pm when everyone else in the house is already asleep.

You can leave the job and carry the entire system with you, perfectly intact, without even noticing you have done it.

That is the step most lifestyle design content skips entirely. Changing the external structure matters. Changing what you are internally reinforcing matters more, because otherwise you’ll just build a smaller, lonelier and more broke version of the same cage.

Now what?

Map your current loops honestly. Where does your time actually go, and what does that behavior reinforce? What does your income require of you, and is that requirement growing or shrinking? What does your current structure reward, and are those rewards pulling you toward the life you want or away from it?

For most people in standard employment, the structure rewards availability and punishes boundaries. The more available you are, the more indispensable you appear, the more you are rewarded and the more available you need to be. Meanwhile your kids are doing homework with whoever is around, their education is outsourced, your relationship runs on whatever energy is left after work takes its share, and your Sunday scaries already have their own social security number.

Once you’re done mapping, build new loops that accelerate in your direction.

Writing publicly builds an audience. An audience creates leverage. Leverage produces options. Options reduce dependence on any single income source or any single majority owner who might have interesting plans he has not mentioned yet. Reduced dependence gives you more room to think, more room to build, more room to be present for the people who are not going to remember whether you hit your Q3 targets. Being with these people gives you more ideas to write.

The goal is a system where what you do today builds more capacity to keep doing it, which includes the work as a part of everything that happens around it.

You don’t have to burn your career down this afternoon. “To hell with it, I’ll just quit now!” won’t help you.

To be fair, some people are well-suited to that corporate structure and there is no shame in that. If you are someone who draws energy from large organizations, who finds meaning in climbing a clear hierarchy, who sleeps fine on Sunday nights and does not feel the walls closing in, the corporate system can be an excellent vehicle and a powerful income generating tool.

It offers stability, resources, a built-in social world that runs, and the kind of career boost that takes decades to build independently.

The danger is staying in a corporate job because you cannot imagine the alternative, while resenting every meeting, every performance review, every reorg, every townhall where you want to believe the man at the front of the room.

If the system is producing a life you actually want, it is a good system.

Systems thinking does not tell you what to want. It just describes what your current structure is producing, without flattery because it doesn’t care in the least about your feelings. If you’re happy in corporate, you already won the game.

If you feel like you are working hard and going sideways, the problem is that something in the structure of your days is cycling you back to the same place. That structure will keep doing that until something changes, and the townhall will keep sounding reassuring right up until the acquisition email or the layoff broadcast hits your inbox.

You are allowed to redesign the machine. Preferably before someone else does it for you.

Happy Building,

— R.

The feeding-and-bedtime part I had covered. Badly, but covered. The problem is a different one.

I’m talking about the deeper mechanism underneath every family that either makes things work or makes them collapse slowly in ways that only become clear in therapy, fifteen years later, at great expense, while the therapist nods in a way that costs you forty euros per nod.

My daughter was running around the kitchen island dragging her racoon by the tail. I did what I always do: explained, calmly at first, why that was wrong. Then explained again, less calmly. Then explained a third time with the specific tone that had the dog looking at me for 20 minutes with his best expression of “geez, why are you so insufferable?”.

Nothing changed. I stood in the kitchen wondering whether I was raising a child or genuinely just talking to a wall that eats all my snacks.

That little wall, for the record, was winning.

Norbert Wiener would have recognized the problem immediately.

Wiener was a mathematician at MIT, a child prodigy who finished his PhD at eighteen, which is either inspiring or devastating depending on how your own twenties went.

During World War II, he was requested to build an anti-aircraft system accurate enough to hit a moving plane. Sounds obvious today, but some 80 years ago the world and the technology available was completely different.

The system had to predict where the target was going, and to do that, it had to continuously read its own previous errors and adjust on the fly.

He couldn’t shoot at the position of the aircraft. It needed to shoot at where the aircraft would be when the shells got there. That required a whole lot of prediction skills, which meant shooting, missing, collecting the data from the miss, and correct.

Wiener saw something in that problem that went far beyond artillery, which is what matters to this article:

A system with no return signal cannot correct itself. It can only repeat the same action, with the same error, indefinitely.

Subscribe now

Like a grown man in a kitchen explaining the danger of running on tiles around a wooden block with sharp edges and a dumb dog in the house, to someone who has already stopped listening and is thinking about chocolate milk.

He called the framework cybernetics, from the Greek for “steersman.” A helmsman does not set a heading and go take a nap. He keeps adjusting. The steering is an ongoing ping-pong game with the environment. Yelling at the water won’t do any good.

I had been making announcements to my daughter that I don’t like her running in the kitchen, which she clearly didn’t care about. It was one-way communication.

I’ve done the same in other instances, for example, when I impatiently asked her to be patient about something. We adults make no sense.

In this conversation with my daughter, I had built no mechanism to check whether my information was being understood, what it meant to her, or why she was ignoring my request.

Information was traveling in one direction only, and Wiener was very clear on what one-directional information produces: a system that cannot self-correct, because it has no basis for knowing it needs to.

His deeper claim, the one that made cybernetics genuinely revolutionary, is that control and communication are the same thing.

Read that again. That’s pure genius.

Control and communication are the same thing.

You cannot govern a system you are not in active conversation with. Think about a thermostat. The thermostat does not heat the house because someone told it to once. It heats the house because it reads the current temperature against the target, finds the gap, and acts on that gap, continuously. The gap is the information, and the information is what makes control possible.

A family runs on that same logic. So does a marriage, a creative project, a working relationship of any kind.

Share

But I didn’t connect the dots until my daughter, after the fourth time I told her I didn’t like what she was doing, looked at me and said, “I hear you, Dad, but why?”

That was the feedback I didn’t ask for, and I’m thankful she got fed up with me and delivered it.

That was information, arriving from the system, telling me my entire model of the situation was wrong.

I was treating the whole behaviour as a compliance problem, but it was actually a comprehension problem. The gap here is a whole universe.

Wiener would have called what I was doing open-loop control: inputs going in, no return signal, outcomes assumed rather than observed. What she handed me, besides my @ss, was the beginning of a closed loop, the kind that actually changes behavior, in systems and in four-year-olds.

At that moment, my daughter was a better systems thinker than I was. She still is, and she knows it.

The pattern we’re looking at is repetition without resolution. When the same unwanted output keeps appearing despite repeated inputs, the instinct is to adjust the input, say it differently, yell it, present it with a new attempt at a sexier slide deck.

But we’re looking at a deeper problem here. We’re looking at no outcome at all from the same repeated input.

If you or someone else is running a monologue, there’s a good chance you’re trying to fix a system but running an open loop. Marriage fights and arguments, I’m looking at you right now.

Wiener’s framework points somewhere else first: check whether a return channel exists at all. If information from the system is not completing the circuit back, more input will not fix anything. The loop is open, and open loops are not correctable by volume.

Build a return channel into any system you are trying to influence. With my daughter, it was about showing her that I was concerned about her running on a slippery floor and cracking her head open on the edge of the kitchen island.

I could have saved everyone a lot of Tuesday evenings if I had looked at parenting with the systems thinking eyes. But as the saying goes: the shoemaker’s son always go barefoot.

Not this shoemaker’s. Not anymore.

The key is to create the conditions under which the system can speak back, and then to treat what it says as data rather than a personal attack on your parenting.

Happy building (and parenting),

— R.

The house had no wire routing channels inside the walls. The wires were plastered directly into them, invisible, with no indication of where they ran. I found out about this the way you find out about most structural surprises in old German houses: by drilling straight through one and seeing fireworks come out the end of the drill.

What followed was not an afternoon. It was six days of peeling plaster, rerouting wires, replastering, and repainting an entire washroom to fix a problem I created while trying to solve a completely different one.

I went in looking for better light. I came out with a complete understanding of why you should never assume you know what’s inside a German wall.

Ludwig von Bertalanffy would have recognized this immediately. Not the drilling, though he might have appreciated the symbolism. The actual problem was that I had treated the existing finishing of the washroom as a closed system. I had a clear input (drill here and there, mount fixture) and a clear output (better lighting).

What I had not accounted for was the boundary: what was actually happening behind the plaster, invisible to me and , and what would happen the moment I interrupted it.

This is what Bertalanffy spent much of the 1940s and 50s formalizing: most systems in real life are open. They take things in from their surroundings, transform them, and send outputs back out. Any model that treats them as sealed containers will eventually meet the wires inside the wall.

Bertalanffy was a biologist, which gave him an unusual vantage point. Biology is full of systems that appear self-contained but are constantly exchanging matter, energy, and information with whatever surrounds them.

His General Systems Theory was a long argument that you cannot understand something by isolating it from its context.

Then in 1956 Kenneth Boulding picked up this thread and made it legible across disciplines. In “General Systems Theory: The Skeleton of Science,” he argued that the real problem was the habitual practice of drawing walls around specific domains and ignoring what crossed those walls.

Each field had its own models, its own preferred way of being wrong, and very little interest in looking over the fence.

Boulding’s hierarchy of system complexity ran from mechanical frameworks (clocks, thermostats) through open systems (cells, organisms) up to social and symbolic systems.

Each level of complexity requires a fundamentally different mode of analysis, and using a lower-level tool on a higher-level system produces very confident nonsense at scale.

A closed system operates on fixed inputs. It doesn’t receive new information from outside at will, doesn’t adjust to changes in surrounding conditions. A thermostat is approximately closed, at least until you have a voltage spike coming from the power line. A business plan presented in January for the full calendar year is, in practice, treated as closed.

An open system is different. A cell exchanges material continuously across its membrane. An organism regulates temperature by sensing and responding to what’s outside it.

A plan built as an open system absorbs feedback, updates its assumptions, and changes course without requiring a full crisis to justify the change.

Bertalanffy introduced the concept of equifinality to describe one key property of open systems: the ability to reach the same end state through different starting points and different paths, because the system is continuously adjusting based on what comes in from outside.

Most planning treats equifinality as a flaw to be engineered out. The instinct is toward a single path and a fixed sequence, where any deviation reads as failure rather than adaptation. What Bertalanffy was pointing at is that the ability to deviate and still arrive is not a weakness in the design. It is the design.

What to watch for

Plans that function as closed systems share a few identifiable characteristics.

• The timeline has no slack built in

• The assumptions are stated once and never revisited

• The success metrics are defined entirely by internal outputs, with no mechanism for incorporating anything from outside the plan’s own logic.

When you see a plan built this way, the question to ask is not whether it is internally consistent. It probably is. The question is whether the boundary has been designed to receive anything from outside, and whether there is any way to respond when it does.

One move

Before you finalize a plan, draw its boundary explicitly. Put the plan inside a box on paper, then list what enters from the surrounding and what leaves. What are the inputs, where do they come from, what outputs does the plan produce, and where do they go? Who or what is on the other side of each exchange?

You don’t need to predict everything that will come in. You just need to build a system capable of receiving it without the whole wall coming down.

Happy Building,

— R.

He called it the Viable System Model. I call it the thing I wish someone had handed me twenty years ago.

The VSM came out of a simple obsession: what is the minimum any system needs to survive? Beer looked at the human brain and nervous system as a literal blueprint to model after. The way your body coordinates breathing, movement, and threat response without you consciously managing any of it, that is exactly what a healthy organization is supposed to do.

Most organizations, even our one-man-bands, startups and independent workers, do none of that. I include myself. For years I ran my work like a to-do list and expected a smooth output. I worked all day like a trained donkey, was incredibly tired by the evening, and yet the results were subpar, to put it nicely. Peak productivity. Zero direction.

The Five Systems You Already Have, Whether You Want It Or Not

Beer identified five functions inside every viable system. They exist whether you have named them or not. The question is whether they are doing their job or you’re doing it for them.

Function 1 is the work itself. The thing you do that justifies your existence. If you are a solopreneur, this is your writing, your client work, your product. If you are a team, this is what your team delivers. Software, services, financial control, Function 1 is the only part that directly creates value.

Function 2 is coordination. When a bunch of functions of type 1 operate at the same time, they will conflict. Your publishing schedule clashes with your client deadlines. Two of your projects compete for the same budget (or in a company, 2 departments). Function 2 is the set of rules, rhythms, and protocols that prevent those conflicts without requiring a manager to jump in every time.

Function 3 is inside management. Someone has to look at all the Function 1 units together and ask whether the whole is more effective than the parts running separately. Function 3 controls resources, optimizes internally, and ensures the current operation is actually working.

Function 4 is outside management. Someone has to watch what is happening in the world. Trends, competition, threats, opportunities. For example, how will your work be affected by AI in 12-24-60 months? Function 4 scans the environment and models possible futures. This is where adaptation begins, before the crisis makes it mandatory (I’m looking at you now, Atlassian, Epic and Oracle).

Function 5 is identity and policy. The function that holds the whole thing together and decides what the system actually stands for. Function 5 balances the tension between Function 3 (keep the current operation alive) and Function 4 (change enough to survive tomorrow). Without it, you get outpaced and obsolete before your realize.

Beer solved organizational design in 1972. And here we are, inventing more meetings and clicking “reply all”on emails.

Where We Usually Mess Up

Beer found out that most of the times, we have either a missing or a broken Function 4. We get so caught up with delivery that we don’t see the environment change.

And the core point in this case is, we keep doing things “as we always did”. We don’t train our skills to adapt to an environment that is constantly changing.

Huge cliché, I know, but just sit for a minute and think about everything that happened in the last 7 years, since COVID first hit. The world today is completely different from what it was in 2019.

COVID changed the way we work. Then AI changed the way we work. Now the energy and fuel crisis is about to change the way we work again. I’d risk to say we saw more change in these 7 years than in the 20 before them.

We are slow to adapt. We are good at reacting after the world changes, but we suck at adapting before it becomes a crisis.

As a consequence, Function 3 tends to hijack Function 4. We get so focused on current efficiency that we stopped scanning for what was coming. We become very good at doing the wrong thing.

Most leaders have a strong System 3 and a dead System 4. They are very good at managing something that stopped mattering two years ago. Hi, it’s me.

Enough Theory, Here’s Why This Is Important

Beer was essentially different from every other organizational thinker.

His Viable System Model treats every organization as a neural network, where each node is itself an autonomous, viable system, and all nodes together constitute a purposeful and cohesive organization.

This means the model is recursive. Every Function 1 unit, if it is healthy, contains its own Functions 1 through 5. Your marketing team has its own operations, coordination, management, environmental scanning, and identity function. Your writing practice has all five too. The whole thing nests over and over.

And this is exactly what tells you exactly where the problem is hiding. If a team’s output is suffering, you can trace it to a missing or broken function rather than guessing at personality, motivation, or culture. Beer gave us diagnosis a map, but we insist on ignoring it.

The VSM For Us Mortals

The useful translation for a content creator, solopreneur, or small team is quite simple. You need to know what your five functions are, and which one you are neglecting.

If your output feels reactive and exhausting, Function 2 has broken down. If you are productive but slowly becoming irrelevant, Function 4 is down (this is the case for many of us who are being affected by AI in our jobs). If you don’t know why you are building what you are building, Function 5 has gone missing.

Beer argued that no viable system is either fully centralized or fully decentralized. It is both, in different dimensions. Kelsie Nabben also wrote a cool Substack Article about this a while ago. For a founder or a solo operator, this means that some decisions need to be defined at the identity level, unchanging, while others need maximum freedom at the operational level. When we mix those up, we end up either too rigid to adapt or too flexible to be relevant. I’m guilty as charged of both.

A Short Note About Function 4

When your days feel full but your long term game looks blurry, that is a Function 4 problem. The work is running, but nothing is watching where the work is going.

One hour a week set aside for environmental scanning, reading widely, thinking about what is changing around your field, is Funcion 4 doing its minimum job.

Most people treat that hour as optional.

Beer would call it the difference between dead and alive.

System 4 is the function that watches the future for you. Naturally, it's the first one we defund when life get busy. We're a special species.

What To Do With This Information

Map your five functions. Write one sentence for each: what is my Function 1, what coordinates it (F2), who manages the current operation (F3), who scans the outside world (F4), and what holds my identity and direction (F5). If any sentence comes out blank or not immediately clear, you have found the fracture.

It takes fifteen minutes. Stafford Beer spent thirty years building the framework so you could do it in fifteen minutes. That’s a reasonable deal if you ask me.

Happy Building,

—R.

Most people diagnose this as a discipline problem. They are wrong, and that wrong diagnosis is itself part of the pattern.

The concept

A mental model is the internal picture you carry of how the world works. You have one for how audiences behave, one for what makes content land, one for whether you are the kind of person who can pull this or that off.

Peter Senge’s argument, laid out in The Fifth Discipline in 1990, was that these pictures do not just influence your decisions but actively protect themselves.

They shape which actions feel logical, and then they filter which results you allow yourself to really see. Most people have no idea this is happening, which is exactly how it keeps happening.

You’re locking yourself in and throwing away the key

When you hold a belief, you unconsciously curate the evidence around it.

Inconvenient results become flukes. Contradictory feedback becomes someone else’s problem. Consistent failure becomes proof that you just need to try harder.

And if you believe that noise downstairs at 2am is a ghost, you go downstairs and see a ghost.

Senge called this a self-sealing loop, and the name is worth sitting with.

The model shapes the action. The action produces a result. The result gets interpreted through the same model. Nothing updates. The loop closes.

This thing is confirmation bias on steroids. Confirmation bias means you prefer information that aligns with your belief. What we are talking about here is, you are building the outcomes that produce familiar results.

Confirmation bias is passive, a mental model is active.

You are not failing to learn from experience. You are learning very efficiently from a version of experience you edited in your own favour. I have built an impressive career on this particular skill.

Yours truly, the Substack writer

Meet Joe. Joe is a content creator on Substack.

Like all of us, he’s been bombarded with Substack growth hacks from gurus and people who grown on Substack by writing about how to grow on Substack.

Sounds familiar?

Joe believes that more posts mean more growth. He posts every day for three months. Engagement stays flat.

His conclusion: the algorithm must have changed. Or the niche is saturated. Or they just need a little more time.

Every one of those explanations lets the original belief walk away untouched.

This is the thing Senge was pointing at. Those interpretations are not stupid. From inside the model, each one is completely reasonable. The model is just invisible, which means it never gets treated as a hypothesis, which in turn means it never actually gets tested.

Joe is not being irrational. He’s being exactly as rational as his hidden assumptions allow.

Another three months of daily posting follows naturally from that.

Escaping the vicious circle

Mental models don’t care about how much effort you put in. They only change when you drag them into plain sight and tear them apart, and he was specific about how.

The first step is to write the assumption down in ordinary language. “I believe posting more often is what drives growth.” That is it.

The act of externalising it creates a small gap between you and the belief, and that gap is the only place honest examination can happen. Most people never do this because it feels unnecessary.

It feels unnecessary because the model is doing its job making it look unnecessary.

The second step is to genuinely look for evidence that you are wrong. Just a straight shot question: what would I be noticing differently if this assumption were false?

Most people skip straight to analysing results. I get it, it’s tempting. Cutting corners always is.

The analysis then gets shaped by the assumption they were supposed to be questioning.

The whole exercise becomes a formal way of confirming what they already believed. Very productive and totally useless.

The warning sign

The clearest sign a mental model is running the show is when you find yourself explaining away the same thing repeatedly.

Any time you catch yourself saying “that does not count because...” more than twice about the same pattern, you are being fooled by your brain. The model is defending itself, and it recruited you to help.

Note to self: worth writing that sentence down.

Stay one step ahead

After your next piece underperforms, before you touch the strategy or blame the platform, write one sentence: “I assumed this would work because...”

That sentence is the model. Hold it out in front of you before you decide what the result means. You might find it has been running things for longer than you thought.

Happy building,

— R.

W. Ross Ashby published it in 1956. It is precise, unforgiving, and the kind of thing nobody puts in a project charter. The Law of Requisite Variety states that whoever is trying to manage a situation can only stay in control if they have at least as many possible responses as the situation has possible problems.

TLDR: you cannot control what you cannot match.

The system does not care that you tried.

Ashby was a psychiatrist turned systems theorist, which explains a lot. He spent years watching human brains try to regulate complex environments and noticed something uncomfortable: the regulator keeps failing, and the failure follows a pattern.

When the number of things that can go wrong outnumbers the ways you can respond, control breaks down, consistently and structurally.

He called the capacity to respond “variety.” The law says only variety can absorb variety. Everything else is just watching things fall apart.

I learned this the hard way in 2019, managing a rollout across four regional offices with one shared project tracker across them and a deeply optimistic timeline.

What I had built was a situation where I had very few moves, pointed at an environment with an enormous number of moves. The offices had different constraints, different stakeholders, different informal power structures, and precisely zero interest in synchronizing their chaos for my benefit.

As you can imagine, this is a recipe for a checkmate.

The tracker turned orange. Then red. I added more columns to the tracker in a Hail Mary.

The tracker did not care.

This is where most people go wrong, and I include past-me enthusiastically in this group.

When a situation gets harder to manage, the instinct is to increase oversight, sometimes in the form of more reports, more check-ins, more information.

What this actually does is add more watching without adding more ways to respond. Your chess board doesn’t get any larger. You are just watching the same moves fail in higher resolution.

Ashby’s law does not care how many status meetings you schedule.

It also does not care about your intentions, your track record, your carefully worded project update, or the fact that you genuinely believed this time would be different.

The gap between how complex the situation is and how many ways you have to respond is a structural problem and structure does not respond to optimism.

And don’t think this is a corporate problem. This happens everywhere.

Think about the parent trying to manage a teenager with a single parenting script they used when the kid was 3. The teenager has infinite moves. The script has twelve. The system does not care that the script worked once. The gap is the problem, and the gap always wins.

Or think about the person running on four hours of sleep, conflicting priorities, and one single coping mechanism. The demands on that person keep multiplying while the ways they can respond keep shrinking. Every day the ratio gets worse and they wonder why everything feels like it is slipping. Well, because it is slipping. Ashby said so in 1956, and he was not being dramatic.

The structural version of this problem appears everywhere once you see it.

A customer service team handling 40 complaint types with 12 approved response templates will always have complaints falling through the gaps, not because of effort or attitude, but because the math does not math.

A manager running eight direct reports on weekly one-on-ones has, generously, about 22 minutes of weekly signal per person. If those people are navigating complex, fast-moving work, the manager runs out of ways to help before the problems run out. The problems do not wait and they don’t care about the calendar invite.

There is some good news here, and it is genuinely useful even if it is also slightly annoying.

You can close the gap in two directions. You can increase the number of ways you can respond, which means more options, more delegation, more flexibility. Or you can reduce the number of things that can go wrong, which means better constraints, clearer processes, fewer moving parts.

Both are legitimate and most situations will require both from you.

The error is doing neither and assuming that trying harder counts as a response. Trying harder is not a new move.

The most useful thing you can do right now is be honest about the gap.

Before you design a plan, count the ways the situation can go sideways. Then count the ways you can actually respond. If the second number is smaller than the first, you already know how the story ends.

Ashby’s law does not negotiate. But at least it tells you the terms in advance, which is more than most things do.

What to watch for

When your options stop growing but your situation keeps getting more complicated, control is already gone. The lag just hides it.

What you can do now: Map the last thing that escaped your control. Write down how many ways that situation could go sideways. Write down how many ways you had to respond. If the first number is larger, you failed to match the complexity.

And remember, this is about strategic positioning, not working harder.

Happy Building,

— R.

And the weird part is that you’re not lazy. You’re working constantly. You’re just working toward something else entirely, something you never consciously chose but somehow ended up optimizing for anyway.

That’s a misaligned incentive problem. And there’s a meaningful difference between this and being lazy.

What’s Actually Happening

Every system, including you, does what it gets rewarded for. The rewards don’t have to be formal or conscious, as long as they are consistent.

Maybe you keep saying yes to things you resent because the short-term reward of being seen as reliable outweighs the long-term cost of being overextended.

Maybe you avoid difficult conversations because the discomfort of having them is immediate and the benefit of having had them is weeks away.

Or maybe you spend your evenings scrolling instead of writing because scrolling delivers a small hit of something now and writing needs likes to deliver dopamine, which will not come for a very long time. Ask me how I know.

None of this makes you a bad person. It makes you a person. The system is just doing what systems do.

Step 1: Name the Behavior You’re Auditing

Pick the thing you keep doing that you don’t want to do, or the thing you keep not doing that you say you want to do.

Be specific. “I procrastinate” is not useful. “I open my email first thing every morning even though I’ve said I won’t” is useful. “I don’t exercise” is vague. “I sign up for gym sessions and cancel them the morning of” is something you can actually work with.

The more specific you are, the more honest the audit gets. Which is also why it’s tempting to stay vague.

Step 2: Map What Your System Actually Rewards

Ask what happens, concretely, when you do the behavior you’re trying to change. Then ask what you’d lose if you stopped.

This is where it gets uncomfortable, because the rewards are often embarrassing.

Staying busy rewards you with an identity. Complaining about your job rewards you with commiseration and belonging. Avoiding a hard conversation rewards you with the temporary absence of conflict. Checking your phone rewards you with the possibility, however slim, that something interesting happened in the last four minutes.

Now leave that embarassment aside for a second. The audit asks you to see them clearly, not to feel bad about them.

Step 3: Identify the Loop

Now we connect them. If you do the behavior, you get the reward. Because you get the reward, the behavior makes sense to repeat. Because it makes sense to repeat, you repeat it. The loop closes and gets a little stronger every time.

Draw it out on paper if you can. Something about making the loop visible, rather than just knowing it exists in the abstract, changes how trapped by it you feel.

I say this as someone who drew a loop about his own avoidance of admin work and then immediately had to go do admin work.

Step 4: Locate the Delay

The reason changing behavior is hard is that the rewards for new behavior are almost always delayed, while the rewards for old behavior are almost always immediate.

You stop saying yes to everything, and the immediate experience is guilt and the vague terror that people will think less of you.

The reward, feeling less overextended and more in control of your time, comes months later.

The loop doesn’t care about months later. The loop cares about right now. So look at the immediate bad feeling as the pain of breaking the loop, which, in this case, can be called a vicious circle.

When you understand the delay, you stop treating your own resistance as a character flaw and start treating it as a structural problem with a structural solution.

Step 5: Find the Leverage Point

Not all the rewards in your loop are equal. One of them is load-bearing. It’s the one the whole structure depends on.

For the person who can’t stop overcommitting, it might not be the fear of disappointing people in general. It could be the specific fear of disappointing one person whose approval they’ve been chasing since 1997.

For the person who can’t sit with unfinished work, it might not be perfectionism broadly but the very specific panic of being seen as someone who doesn’t follow through. Hi, it's me.

The leverage point is the reward you’d be most reluctant to give up. Which is exactly how you know it’s the right one.

Step 6: Pick a Measurable Signal

You need something small and observable to know whether anything is shifting. An actual data point you can check in a few weeks.

How many times did you do the thing you said you wouldn’t? How many times did you do the new thing instead? Aim for one measurable difference between this month and last month.

Small is fine. Small is actually the point.

Step 7: One Next Move

The audit will surface more than you can act on. That’s by design. You act on one thing.

Pick the smallest intervention that touches the leverage point. Don’t try to do a complete life overhaul at one. Run that one little thing consistently, for long enough to get past the delay period. Then check the signal.

I cannot stress enough how much of self-improvement fails because people do ten things at once and then can’t tell which one worked or didn’t.

Stop.

One move. Measure. Adjust.

The Part Nobody Wants to Get To

At the end of this audit, you will almost certainly find that the behavior you’ve been judging yourself for makes complete sense given the system you’re living inside.

You want to go to bed earlier, but every night you find yourself watching the third episode of something you don’t even like because you’re avoiding lying in the dark thinking about going to work tomorrow or something else you’re avoiding.

Or you keep writing that report every week, even though you know nobody reads them, because it shows effort, and effort is what gets noticed in performance reviews.

That’s not a reason to keep doing it. It’s a reason to stop blaming yourself for it and start changing the structure instead.

The goal is to build an environment where the thing you want and the thing you’re rewarded for are pointing in the same direction.

That gap, right there, is the whole problem. And it’s also the whole fix.

Happy building,

— R.

It wasn’t complicated. Actually it was obvious in retrospect, and nobody had been able to articulate it so far.

She had spent decades studying complex systems like economies, ecosystems, organisations, cities, and kept noticing the same pattern: the interventions happened at the wrong place. People pull the levers closest to them without checking if it actually helps, feel productive, and then wonder why nothing ever changes.

Her list had twelve leverage points. Places to intervene in a system, ranked by how much they actually make a difference.

The least effective ones are where almost everyone spends their time.

The most effective ones are almost never where anyone first looks.

I read this for the first time and immediately thought about every strategic planning session I had ever attended. Then I needed to sit quietly for a few minutes and reconsider some of my life choices.

What a leverage point actually is

A leverage point is a place in a system where a small change produces a large shift in behavior.

A change in how the whole system operates.

We are quite good at finding places to intervene, but are considerably less good at finding places where intervention actually does something lasting or relevant.

We love tweaking, fiddling, changing things that will have absolutely no impact in the problems we’re dealing with, but we feel good about doing so. Instant gratification.

If we want to actually make something change, we have to act with precision, where we can force things to go the way we want them.

Meadows was precise about this: the higher the leverage, the more the system resists the change. Which explains why effective interventions are always painful, harder to make, harder to sustain, and considerably harder to get approved by your manager before lunch.

The four leverage points worth understanding

Meadows listed twelve. Most writing about her work covers the bottom half: numbers, flows, buffers, because they are concrete and comfortable and do not require anyone to have an uncomfortable conversation.

The ones that really impact systems are further up the list. Four of them deserve your attention.

Information flows

Meadows placed one of her sharpest observations here: missing information flows are among the most common causes of system failure.

Notice that we are not talking about wrong information, we are talking about missing information. The data that should exist and simply does not.

When you need to make a decision and cannot see its consequence in real time, the feedback loop that should correct your behavior never forms. You keep making the same call, and the system keeps producing the same outcome. Everyone looks confused because the decision actually looks good, but it’s blind. Yet hardly ever someone questions the structure underneath.

A concrete example: A Substack creator chasing posting frequency with no visibility into which content actually converts followers into buyers. They optimise for volume. Engagement looks fine but revenue stays flat. When reviewing the year’s results, the cost shows up as “grew an audience and built nothing with it”.

Meadows argued that delivering the right information to the right people at the moment they need it changes behavior faster and more durably than almost any rule or incentive you could design around it.

The leverage is not in the information itself, but it is in who receives it, when, and whether they are in a position to act on it.

I have watched teams build elaborate new processes to solve problems that would have disappeared entirely if one person had simply received a weekly number they were not previously seeing. The process was a monument to not asking where the signal was missing.

I have also been the person who built one of those processes. The monument was quite detailed.

Rules

Rules define the boundaries of what a system can do. They carry more leverage than most people give them credit for, which is exactly why changing them is so politically difficult.

Meadows was specific: the rules of a system define its degrees of freedom. Change the rules and you change what is possible inside the system, not just what is likely.

This is why regulatory change produces such enormous domino effects. It is not adjusting a dial. It is redrawing the whole boundary of the playing field.

Inside businesses, rules operate the same way. Who can approve what, what requires sign-off, what is allowed to fail and what absolutely cannot. These rules shape behavior more reliably than any culture initiative, because they operate whether or not anyone is paying attention to them.

The leverage point is in identifying which existing rules are producing unintended behavior and have become invisible because everyone stopped questioning them a long time ago.

Most businesses carry rules nobody remembers the origin of, that everybody works around, that are shaping behavior in ways nobody has mapped. Those are the ones worth finding. They are usually load-bearing in ways that will surprise you.

Incentives

Incentives tell the system what to optimise for. They are extraordinarily powerful and extraordinarily easy to get wrong.

Meadows understood that incentives define what the system considers success. Change what the system rewards and you change what the system produces, often in ways that genuinely surprise the people who designed the incentive in the first place.

The classic failure: you measure what is easy to measure and reward what you can measure, which causes the system to optimise for the measure rather than the outcome the measure was supposed to represent.

Parents will optimise for getting the kid to sleep fast using whatever works tonight, with no visibility into the habit they are building over six months. The short-term metric is great, kid is asleep at 7:30pm. The 2am wake-up cycle is the bill arriving later.

Sales teams will game quarterly targets at the expense of customer relationships, hospitals will optimise for throughput metrics while care quality drifts, developers will close tickets faster by writing code that will create twice as many tickets in six months.

Meadows’ point was that incentives need to be designed at the level of the goal, not the activity. But goals are harder to measure and considerably easier to argue about in a room full of stakeholders, so we ignore that and design on the wrong level.

This is uncomfortable because redesigning incentives means admitting that the current incentives are producing the current results. The system is working exactly as designed.

I have sat in rooms where this realisation arrived visibly on someone's face mid-sentence. It is one of the more interesting things to witness as long as it is not your face.

Goals

This is the highest leverage point on the list that most people will realistically touch, and it is almost always the most underestimated.

The goal of a system is what the system actually optimises for when pressure arrives and trade-offs have to be made.

Meadows was direct about this: if you want to understand the real goal of a system, watch what it does, not what it says.

A company that says its goal is customer satisfaction but consistently trades customer experience for short-term margin when the two conflict has a real goal of short-term margin. The stated goal is just furniture.

Changing the real goal of a system is the highest leverage intervention most leaders can make, and the most resisted. Because the real goal is usually serving someone’s interest, and changing it means asking that someone to accept a different outcome. That conversation does not happen in a workshop, in a townhall, in an official meeting. It happens in a closed room with very few people and no slides.

This is why genuine strategic pivots are so rare and so difficult. They are are political exercises. Systems do not give up their goals without a fight, because goals belong to people before they belong to organisations.

Why people avoid high-leverage interventions

Meadows noticed something worth sitting with: the higher the leverage point, the harder the system pushes back against changing it.

Information flows are invisible and politically neutral until you make them visible, at which point someone always has a reason why that particular metric should not be shared with that particular person.

Rules feel settled until you question them. That’s the moment the person who benefits from the rule appears to explain, at some length, why it exists for very good reasons that trace back to an incident in 2004.

Incentives feel fair until you change them, at which point the people optimised for the current structure feel personally attacked. In a sense, they are.

Goals feel agreed upon until you challenge them, at which point you are no longer having a strategy conversation. You are having a power conversation, and the agenda does not reflect that.

That’s the system behaving exactly as Meadows predicted. It is the system protecting its current structure. Understanding that makes the resistance easier to work with, even if it does not make it easier to enjoy.

What to do with this

Before adding another process, launching another initiative, or writing another policy, ask four questions.

1. Who is missing a signal they need to make better decisions, and how quickly can you get it to them?

2. Which rules are producing behavior nobody intended, and how long has it been since anyone questioned them?

3. What is the system actually rewarding, and is that the same thing as what you want it to produce?

4. What does this system optimise for when a trade-off is forced, and is that the real goal or just an inherited one?

You do not need to answer all four. One honest answer to one of these questions will tell you more about where to intervene than a full strategic review conducted over two offsite days with a very expensive facilitator.

I say this as someone who has attended several of those offsites. The catering is usually good. The leverage points remain mostly untouched.

One next move

Take the most stubborn problem you’re dealing with right now. The one that survives every fix.

Ask which of the four it is: an information problem, a rules problem, an incentives problem, or a goals problem.

If you are genuinely not sure, it is probably a goals problem wearing one of the other three as a disguise. They usually are.

Happy building,

— R.

Different situations, same problems. Different years. Same slow collapsing. I would fix something obvious, feel briefly competent, maybe even tell someone about it, and then watch the whole thing rebuild itself six months later like I had done nothing at all.

It was not a case of bad luck. I was just completely blind to what was causing every problem I was supposed to be managing.

I also have quite a few project management certifications. Framed. On a wall. In my house. But none of them taught me this, because project management and systems thinking are two different beasts, although very closely related.

This is the five-minute method I wish someone had handed me before I spent several years confidently solving the wrong things.

What you are actually looking for

A reinforcing loop has one defining feature: the output feeds back into the input. The effect becomes its own cause. The cycle multiplies itself in whatever direction it is already moving.

Good things become better. Bad things become hell. The loop has no conscience, no preference, and absolutely no interest in your feelings about it.

Most people miss reinforcing loops because they look for events. A thing happened. Maybe a person dropped the ball. The idiot COO made a poor decision. They find a culprit, fix the obvious part, dust their hands off, and then wonder why the same problem shows up again seven weeks later wearing a slightly different hat.

You know what they say, a pig in makeup is still a pig.

The loop was never about the event. It was about the structure underneath it. The event was just the part you could see, and the structure remains untouched.

Spot the loop in 5 minutes

With five minutes and a reasonable tolerance for mild existential discomfort when the loop becomes visible, you realize it has been there lurking for two years.

Step 1 - Name the symptom you keep seeing

Pick one recurring problem. Something annoying (or disastrous) that comes back reliably, like a bad habit or a status update request from someone who has never once read the status update.

Write it in the center of the page. Be specific. “We keep missing deadlines” is a feeling. “Client delivery slips one to two weeks every time” is something you can actually work with.

Step 2 - Follow the chain in both directions

From your symptom, trace two paths.

Forward: what does this symptom produce as a consequence? Then what does that produce? Keep going three steps.

Backward: what feeds this symptom in the first place? What feeds that? Three steps back.

Keep the language plain. You are following a thread, not building a consulting framework to post on LinkedIn. That comes later and you can even start with “I’m thrilled to announce…”.

A real example from my own life: Starting from “delivery keeps slipping”, the slip creates client pressure, pressure generates urgent check-in calls, calls interrupt focused work, interrupted work slows delivery further, delivery slips more. Going backward: delivery slips because estimates were optimistic, estimates were optimistic because the team felt pressure to commit fast, that pressure came from the last time delivery slipped and leadership stopped trusting anyone.

Step 3 - Check if the chain closes

Look at what you drew. Does the final consequence eventually connect back to your original symptom?

If yes, you have a reinforcing loop. The system is feeding itself and has probably been doing so since before you arrived.

If no, you have a cause-and-effect chain, which is still useful, but that is a different conversation for a different Wednesday.

Three signs you are definitely in a loop

The problem keeps returning after you fix the obvious part. You addressed the symptom, felt good about it, mentioned it in a meeting. The loop kept running underneath and rebuilt the exact same conditions while you were busy feeling relieved.

Each cycle feels slightly worse than the last. The amplitude is increasing. That is reinforcement doing precisely what it is designed to do, just not for you.

Everyone is trying hard and nothing is improving. This one is counterintuitive but remarkably consistent. Reinforcing loops often run hardest in high-effort teams, because all that energy gets redirected into the loop rather than out of it. The gap between how hard people are working and how little is actually moving forward - that gap is the loop waving at you.

A personal example I have fully come to terms with

A few years back I was managing a project where the team was permanently underwater.

More tasks kept appearing than were being completed. Leadership, reasonably concerned, asked for more information to understand the situation. Reporting consumed time, which left us all less time went to actual work. More tasks fell behind. Leadership, now more concerned, added more reporting.

I spent half a year running motivation workshops with the team, sending encouraging messages and offering support, and having one-on-ones where I said things like “I really believe in this team” while the backlog doubled.

I should have just spent a few hours mapping loops on a piece of paper instead.

When we finally challenged the approach and cut the reporting cadence in half, output improved in less than a month. Because we had stopped feeding the monster that was eating them alive.

I tell this story with the confidence of someone who has processed it thoroughly and not at all with the energy of someone who still thinks about it at 2am sometimes.

Four loops that show up everywhere

These appear across industries and team sizes with the kind of regularity that suggests they are less about individual management failures and more about basic human structure.

Which is either comforting or deeply alarming depending on the day.

The confidence loop

Good results build confidence, confidence improves decisions, better decisions produce better results. Run this upward and you get a team that seems to operate on a different plane entirely. Run it downward and you get a team that has learned to solve their own problems because raising problems has historically led to a longer meeting and no actual change.

The documentation loop

Poor documentation creates confusion, confusion generates questions, answering questions pulls time away from writing documentation, documentation stays poor indefinitely. Most knowledge management initiatives (and I mean probably like 75% of all attempts to document something) die inside this loop without ever knowing what ended them.

The hiring loop

Under-resourced teams cut corners, cutting corners create debt and quality problems, quality problems generate extra unplanned work, extra work keeps the team under-resourced. Companies try to hire their way out but the loop absorbs the new hires like an organisational Venus flytrap on steroids and carries on as before.

The trust loop

Leaders micromanage because they do not trust the output, micromanagement signals distrust to the team, the team stops taking ownership, output quality drops, leaders micromanage more intensely. This one ends careers. It also tends to take entire departments with it on the way out, which is a generous parting gift.

What to do once you find one

Two options. Only one works long-term.

Option one is to break the chain at its weakest point. Find the single connection that is most accessible and interrupt it. Faster to implement. Sometimes genuinely all you need.

Option two is to reverse the direction entirely. Find the input that would start the loop running the other way. Takes longer to design but it’s pedal to the metal in your favour once it starts moving.

Whatever you do, resist the urge to fix every link at the same time. That is how you end up making everything measurably worse before producing a modest improvement that nobody remembers was the original goal. Ask me how I know.

Pick one link. Move it. Watch what the loop does next, and wait for the delay before doing anything else.

Your exercise for this week

Take the most persistent frustration you are carrying right now. The one you have addressed at least twice and somehow still wake up to.

Run it through the three steps. On paper if you can, because drawing ugly arrows by hand does something that typing on a screen does not. I do not know why. I am a project manager, not a neuroscientist. I just know it works.

If you find a closed loop, you have found the shape of the problem. That is the first moment you are actually in a position to change it rather than just react to it and then write a very thorough retrospective about it.

Do not fix the loop today. Name it dead center in your page.

Write one sentence: there is a reinforcing loop where X produces Y, Y produces Z, and Z feeds back into X.

Naming it is the moment it loses some of its grip on you. It was running invisibly, now it is not. That is you reclaiming control.

Happy building,

— R.

Before I started learning Systems Thinking, I spent years doing this. Every failed project, every missed deadline, every strategy that collapsed halfway through execution, I assumed someone had not tried hard enough. Usually me.

I was dead wrong.

That was Jay W. Forrester’s central point when he built system dynamics in the 1950s. Organizations do not behave according to intention. They behave according to feedback, delays, and internal structure. You can have smart people, genuine effort, clear goals, and still produce failure if the loop is wrong.

I found this both illuminating and deeply inconvenient.

First, understand what a loop actually is

A system reacts to its own output. That reaction creates a loop. There are two types:

• Reinforcing loops: the ones that spiral

  A reinforcing loop amplifies whatever is already happening, good or bad.

• Balancing loops: the ones that try to correct

  A balancing loop tries to push things back toward a target.

You can find a more detailed explanation here:

Delay is where systems become dangerous

A delay means the system reacts after reality has already changed.

In practice: You launch a process change today. The real consequences appear three months later.

By then, you have already layered three more fixes on top of the original one. The first signal is buried. Nobody knows what caused what anymore.

Forrester showed that long delays push managers to overcorrect, undercorrect, or react to conditions that no longer exist. They are not steering the current system. They are steering an outdated picture of it, with confidence.

I have done this. I have also been the person confidently presenting a slide deck about a problem that had already solved itself two weeks earlier. Nobody told me. I just kept presenting.

This is why companies can look completely irrational from the outside while every individual decision seemed reasonable at the time it was made. The decisions were not wrong, but the timing was completely off.

A real example: Teradata

In late 2025 and early 2026, Teradata publicly changed its planning and governance model because long planning cycles were making transformation almost impossible.

They had been running on three-to-five-year (!!!!) roadmaps. That created a structural trap: strategy stayed fixed while the market kept moving, customer feedback arrived too late to affect priorities, and funding decisions happened so rarely that dead initiatives survived long past their expiry date, absorbing attention and budget like organizational zombies.

All that not to mention that nowadays an employee normally stays with a company somewhere between 22 and 30 months on average, meaning the majority of the people involved in the process start were no longer there to see the strategy through.

Leadership did not frame this as a people problem or an execution problem. They rightfully called it what it was: a system design problem.

They replaced the long roadmap with rolling 12-to-18-month planning, added quarterly portfolio resets and started using OKR-style reviews to reallocate attention faster.

The result: less time between a signal appearing and a decision responding to it. Fewer dead projects kept on life support and healthier loops overall, with more current and relevant information.

Forrester would have nodded.

Most organizations still get this wrong

And this happens across the board, from enterprise sized orgs to startups and even solopreneurs.

They keep asking “why are teams not executing?”, when the better question is: “what delay is hiding the truth from me?”

Teams usually execute exactly as the structure tells them to.

But monthly reporting hides weekly failure. Quarterly budgeting protects initiatives that stopped making sense in month two. Annual planning freezes assumptions that expired before the ink dried.

The visible symptom is slow execution, while the actual cause is a correction that arrives too late to matter.

This happens at home too

You have been tired for three weeks. You know you should sleep more. You keep postponing it because work is heavy right now, and you will recover on the weekend.

But the weekend comes and two nights of decent sleep do not fix three weeks of deficit. You conclude that “sleeping more doesn’t really work for you”, a sentence I have genuinely said out loud, and return to the same pattern.

The loop continues because the correction arrived too late and was too small to close the gap that had already opened.

That is a systems problem right there sitting at your own dinner table.

What to do with this

Take one frustration you are carrying right now, at work or at home, and ask three questions:

What is feeding itself here? What is trying to correct it? Where is the delay between cause and consequence?

If you cannot find the delay yet, you are probably still looking at symptoms. Keep pulling the thread.

One next move

Do not try to redesign the whole system. That way lies a very long weekend and a whiteboard covered in arrows that makes sense to nobody, including you.

Shorten one feedback cycle. All you need is one earlier correction point. That single change is often enough to shift the direction of everything connected to it.

Next Monday: Why the highest-leverage points in a system are almost never where you first look.

Happy Building,

— R.

Every Sunday, same ritual. Open Notion. Review the week. Ask myself what went well, what didn’t, what I’d do differently. Write it all down in neat little sections with headers and everything.

Very satisfying, and completely useless. I have 4 years of Notion pages to prove it.

Because I was reviewing events. Tasks completed, tasks missed, how I felt about it all. What I never once asked was: what’s the structure producing these events? What’s causing me to fail?

Those are the questions that actually changes things. And it takes about 20 minutes once you know how to ask it.

Most productivity reviews are basically a diary corporate formatting.

You look at what happened, you make a plan to do better, and two weeks later you’re writing the same notes with slightly different wording. The behaviors shift for a bit, then snap back in the original place.

You assume the problem is consistency, but if you read You’re not listening, you know it’s not.

The problem is you’re reviewing outputs when you should be mapping loops.

A loop doesn’t care about your intentions or your tidy Miro board. It produces the same outcome every cycle reliably, mechanically, without malice, until something in the structure changes.

Your weekly review should be finding those loops. Once you see them, you can decide whether to keep them, break them, or redesign them.

Here’s how I do it in 20 minutes.

Step 1: Pick one repeating pattern (5 minutes)

One thing. Not two, not a sequence, not a list. One.

It can be the meeting that always runs long, the task that never gets started, the conversation that keeps happening, the Sunday scaries that show up like an uninvited relative who somehow still has a key.

Write it at the top of a page. “The project review always turns into a blame session.” “I never actually start the thing I planned to start on Monday.” “I end every week feeling behind.”

One pattern. That’s your target.

Step 2: Find the loop (10 minutes)

This is the actual work. Ask two questions and write whatever comes up.

What does this pattern produce that feeds back into causing itself?

If the project review turns into a blame session, what happens after? People leave defensive. They share less information next time. Less information means more surprises. More surprises means more blame. The loop tightens.

If you never start Monday’s task, what happens? It gets pushed to Tuesday feeling heavier. The weight of it makes it easier to avoid on Tuesday too. Avoidance produces more dread, dread produces more avoidance. Same loop, different week.

Now ask:

Is this loop reinforcing or balancing?

Reinforcing loops amplify. They take a small thing and make it bigger over time, in either direction. Avoidance spirals. But so does momentum. Both are reinforcing loops.

Balancing loops resist. They push back toward a set point. You work too hard, you get exhausted, you slow down, you recover.

The loop is trying to keep you in equilibrium. That’s the system doing exactly what it’s designed to do.

Most people fight balancing loops by pushing harder. The loop just pushes back harder as well. Understanding which type you’re dealing with changes what you do next.

Step 3: Find the delay (3 minutes)

Ask:

How long between cause and effect?

This matters more than people think. If the delay is long, you’ll miss the connection entirely. You burned out in November but the loop that caused it started in August. You changed your diet mid January but you’re looking for results on the 2nd of February.

Write down the rough delay time for your pattern. Even a guess is useful. It tells you where to look for the cause, and it stops you overcorrecting because you think nothing is changing.

Step 4: Name one leverage point (2 minutes)

One specific change to the structure.

Not “I’ll be more focused on Mondays.” That’s behavior against a loop. The loop wins every time.

Something like: “I’ll put the Monday task as the only thing on my calendar before 10am, with everything else blocked.” That’s a structural change. You’re redesigning the environment, not willing yourself through it.

One change. Small enough to actually do. Targeted at the loop, not the symptom.

What the whole thing looks like

Twenty minutes, one page, looks something like this:

Pattern: I end every week feeling behind even when I complete most of my tasks.

Loop: Completing tasks feels good so I add more. More tasks means I complete a lower percentage. Lower percentage feels like falling behind. Falling behind triggers adding more tasks to “catch up.” Reinforcing loop, going the wrong direction.

Delay: About a week. The overload from Monday shows up as Friday dread.

Leverage point: Cap the weekly task list at 5 items before adding anything else.

That’s it. No journaling about feelings, no aspirational future-you planning. Just what’s the loop, what’s the delay, where can I pull a lever.

Do this for four weeks and something useful happens.

You start to recognize your loops without having to look for them. The pattern shows up on a Tuesday and instead of thinking “why does this keep happening,” you think “oh, that loop again.” That moment of recognition, before the reaction, is where your actual leverage lives.

Most people never get there because they’re too busy reviewing events and planning intentions.

The loops were always there. You just weren’t looking at them.

Your move this week

Set a 20 minute timer. Sunday works, Friday works, whenever you actually do it is the right time. I do it Saturday morning after convincing myself for 4 hours that I'll do it Friday.

Choose one pattern, find the loop, name the delay, pick one leverage point.

That’s the whole thing. Zero color-coding required.

Let me know what you found.

Happy building,

— R.