Let me name it.

The whole everyone sees

Here’s what the growth data won’t tell you at a glance: custom purchasing hasn’t exploded in the categories where fit is genuinely hard. It’s exploded where the mass-market option already works.

Orthotics, prosthetics, bespoke tailoring for unusual bodies — those categories grew at boring, steady rates. The blistering growth showed up in notebooks, coffee subscriptions, sneakers, meal kits. Products that come in 47 standard variants and fit most people who buy them.

That mismatch is the tell.

When someone pays a 40-to-60% premium to configure something already available off the shelf, they aren’t solving a fit problem. I’ve sat in enough consumer-research debriefs to know what the surveys actually surface — and it isn’t “the standard version didn’t fit.” It’s a quieter line, usually the second or third thing said: I just wanted something that felt like mine.

Read that as an emotional transaction, not a taste one, and the whole category rearranges itself. The feeling being purchased is the low, persistent dread that you might choose wrong — and that choosing wrong would mean something about you.

I spent two days trying to find a clean study that isolated “custom premium” from “brand premium” and came up empty. The categories are too tangled. So I’ll argue this from mechanism instead of citation, and tell you where the seams are.

Part 1: the visible piece — needing versus fearing

The counterargument deserves its full weight, because it’s correct as far as it goes.

Some people genuinely require custom products. A six-foot-four person buying off-the-rack trousers isn’t anxious — they’re tall. A left-handed surgeon needs instruments built for left-handed hands. These are real fit gaps, and folding them in with lifestyle customization would be lazy analysis.

So here’s the line that matters. Is the standard option functionally unavailable to you? Or is it available, and it just feels like settling?

If you could use the standard version but won’t — if the word “settle” showed up anywhere in your thinking — that gap isn’t anatomical. It’s psychological. Practical necessity and regret-aversion look identical from the outside. Same purchase, same premium, same receipt. But they come from completely different places — and the distinction matters because only one of them is solving a real problem. The other is paying to feel like you are.

The tall person isn’t the reader I’m talking to. You know if you’re the tall person.

Part 2: the mechanism underneath

Now the part I find genuinely strange, because it runs backwards from how it feels.

Customization doesn’t reduce regret by making the product better. It reduces regret by transferring authorship of the outcome to you. When you configured the dimensions, the color, the subscription cadence — any disappointment is structurally yours. And here’s the twist: there’s no one left to blame, which paradoxically makes the regret feel smaller. Diffuse. Absorbed.

Behavioral researchers have a cousin of this in the “IKEA effect” — the finding, out of work by Michael Norton at Harvard Business School around 2011, that people overvalue objects they had a hand in assembling. Build the shelf yourself and you’ll price it higher than the identical pre-built one. We fold our own effort into the object’s worth.

But underneath that’s the darker mechanism, and it’s the one nobody selling custom wants examined.

Customization doesn’t rescue you from the anxiety of choosing — it lets you perform choosing, more elaborately, and charges you for the stage.

Barry Schwartz, the psychologist behind The Paradox of Choice, spent years documenting that more options correlate with lower post-purchase satisfaction and higher paralysis. Customization sells itself as the cure for overload — you’re in control now. In practice it does the opposite. It multiplies the micro-choices you have to make and then defend to yourself later.

You didn’t escape the fear. You bought a more expensive way to enact it.

Part 3: the hidden cost

The premium is the visible cost. The invisible one is what the habit does to your judgment over time.

Each custom purchase quietly teaches you the same lesson: the standard option wasn’t safe to choose. Live inside that lesson long enough and it generalizes. What erodes isn’t your bank balance — it’s your confidence that you can make a good-enough call and live with it.

That erosion has been underway for a decade, and the compounding is quiet. You face a choice, feel the flicker of what if I pick wrong, and reach for the custom version because configuring feels like control. The dread quiets — briefly — so the brain files custom as the thing that made it stop. The next standard choice now feels slightly more dangerous than it did before. Your tolerance for good-enough shrinks, and the only thing that soothes it costs more each time.

That’s not discernment sharpening. That’s a tolerance dropping. The two feel nearly identical from the inside, which is exactly why the story you’ve built — I’ve high standards — holds up so well. It’s flattering. It’s also the wrong diagnosis.

Why the parts matter more than the whole

I’m not against custom things. I own a pair of shoes built to a mold of my own foot, and I’d buy them again.

The point was never the object.

It’s what the reaching-for reveals. That the choice felt unsafe. That safety had to be purchased. That somewhere in the last ten years, the quiet assumption that you could pick something ordinary and be fine — that just stopped feeling true.

Sit with which one you’re actually buying.

The fit.

Or the reassurance.

That’s the claim, and I’ll spend the rest of this arguing it. Because when a deployment failed at 2 a.m., I was the person watching the spinner, not the log. I thought I was missing facts. I was missing something stranger.

You’ve asked yourself this after watching someone senior fix in ninety seconds what cost you an afternoon: why do they know where to look?

Three things are happening simultaneously that nobody is connecting

The first: error-rich systems have gone mainstream. Cloud infrastructure, LLM APIs, deployment pipelines — in 2026 a solo founder now operates a stack that a 2015 startup needed a five-person team to run.

The second: the tools got democratized. Vercel, Supabase, the Claude and OpenAI APIs — anyone can wire together a system that fails in ways they can’t read.

The third, and the one nobody wants to name: interface design now deliberately buries diagnostic information beneath calming UI. The spinner. The soft “something went wrong.” The green checkmark that means request received, not task succeeded.

Put those together and you get a fast-growing population of capable-looking users who are, in practice, blind to the actual state of the system they’re operating.

When a deployment fails, a senior engineer’s eyes snap to the terminal. A newcomer stares at the screen, waiting for it to say what to do next. And it never will — because it wasn’t built to.

Here’s the reframe. The gap isn’t that experts know more facts. Experts and beginners are literally looking at different things. The difference is attentional, not informational — and that redefines what “getting good” actually requires.

Isn’t this just experience — you learn where to look?

Sort of. But the mechanism underneath is more structural than “you pick it up over time.”

Cognitive science calls it perceptual learning. Experts don’t just accumulate knowledge. They build a compressed internal model of the system that lets them skip surface presentation and query the underlying state directly.

A radiologist doesn’t see a chest X-ray the way a second-year medical student does. The student scans pixels. The radiologist sees deviation from an internalized baseline — the abnormality almost pops. Same image. Different perception.

A chess grandmaster doesn’t evaluate pieces one by one. They perceive board tension as a single shape. In Chase and Simon’s landmark chunking studies, masters reconstructed real game positions almost perfectly after five seconds — but on random positions, they were no better than novices. The skill wasn’t raw memory. It was pattern recognition built on structured exposure.

Technical expertise runs on the same wiring.

The error message is the system reporting its own state; the screen is the interface reporting what it was designed to show you.

And what it was designed to show you is almost always curated — latency-softened, reassurance-optimized, built for onboarding rather than diagnosis. A Stripe API call can return a 200 OK at the HTTP layer while the error object inside the body tells the real story. Beginners read the 200. Experts read the body.

So getting good at anything isn’t memorization. It’s learning which layer of the system is actually telling the truth.

That’s the tell. Expertise is a learned redirection of attention — a rewiring of what you even perceive as relevant. The facts are downstream of that.

But what about systems where the screen actually tells you what’s wrong?

This is the strongest objection, and I’m not going to wave it off.

Good interface design can carry signal. Stripe’s dashboard surfaces failed-charge reasons directly — card_declined, insufficient_funds — right in the UI, no log-diving required. A skilled designer can make the screen tell the truth.

But sit with what that concession actually reveals.

When good design works, it works because the designer did the expert’s attentional work in advance. They pre-filtered the noise and promoted the signal on your behalf. You’re still running on an expert model. You’ve just outsourced it to whoever built the interface.

Which means the distinction didn’t disappear. It relocated.

And here’s where it stops holding for the reader who wants a clean rule: most systems aren’t Stripe. Most systems ship on a deadline, and diagnosis is what gets cut first. Even in the well-designed ones, the edge cases — the failures that actually cost you — are almost never the states the UI was built to surface. The screen is optimized for the expected. Expertise is only worth anything in the unexpected.

So the honest limit: if you live entirely inside three or four mature, well-instrumented tools, this attentional gap barely touches you. The design absorbed it. This distinction bites hardest at the frontier — new APIs, glued-together stacks, anything under six months old — where nobody has done the filtering yet.

If I’m wrong about this, the cost is real: I’d be telling beginners to distrust interfaces that were, in fact, trying to help them — and teaching a suspicion that slows them down for no reason.

But I don’t think I’m wrong. I think the spinner has taught a generation to wait for permission from a layer that doesn’t have any.

If expertise is mostly learned attention — not stored facts — then what happens to it when the model gets good enough to read the error for you?

You know the behavior. A conversation ends, you walk away, and then you’re back inside the app four minutes later reading the same static messages you already read. Nothing has changed. You check anyway.

I used to explain this to clients as an attachment issue. I was wrong, and I changed my mind because of one specific finding.

Future A: you act

Here’s the fork. In one future, you keep checking. In the other, you accept the thread won’t update. Let’s take the checking future first, because its logic feels airtight from the inside.

The impulse runs like this: the ending felt unfinished, so you return to where the conversation lives, hoping resolution showed up while you were gone. Reasonable, on the surface. Anxiety and uncertainty drive repeated checking — that part is well documented — and after a breakup, reassurance-seeking becomes a primary motive. So the pull isn’t a character flaw. It’s a search.

But here’s the asymmetry nobody names.

The thread isn’t a conversation. It’s a record of a conversation that already closed. Returning to it thirty times isn’t searching for a new ending — it’s circling the locked door of an ending that can only be generated by a new exchange, never discovered inside the old one.

The rational origin of the impulse doesn’t make the behavior rational. It makes it understandable. Those are different things, and the difference matters more than it looks. The thread you’re mining is finite and closed. Every refresh runs the same query against the same unchanged record.

Three things I learned watching people do this in real time:

1. The checking is a query with no return value. Each refresh runs the same search against an unchanged database. You’re asking a closed file to answer a question it structurally can’t answer, because the answer lives in a future message that doesn’t exist yet.

2. The impulse strengthens each time you obey it. By the time you notice you’re doing it, the habit loop has already collected its reward — the tiny relief of at least I looked. That relief is what wires the loop tighter, not the information (there’s none).

3. Understanding the origin doesn’t lower the frequency. This is the part I got wrong for years. I thought naming the mechanism would loosen the pull. It doesn’t. Awareness is the prerequisite, not the fix.

Future B: you wait

Future B feels like the grown-up path. You don’t check. You let it breathe.

But waiting isn’t neutral. It’s a cost that compounds quietly, and this is where I changed my mind hardest.

The unresolved exchange doesn’t sit still while you wait. It recruits meaning. Your brain, denied a proper ending, builds one out of whatever ambient anxiety is lying around. Silence becomes subtext. The absence of a message becomes a message. That gap — between a truncated exchange and any real resolution — is precisely where rumination takes root.

Waiting doesn’t resolve the truncation. It hands the narrative to your worst-case interpreter while you stand politely off to the side.

So the asymmetry isn’t that one future is better. It’s that neither preserves the status quo. Checking produces false data actively. Waiting produces false data passively. One of you generates noise; the other lets noise generate itself.

A quick case in point. A founder I worked with in 2023 insisted he wasn’t bothered by a co-founder who’d gone quiet after a tense Slack thread. He hadn’t checked it in two days — proof, he said, of his composure. But he’d spent those two days building an entire storyline about betrayal from a single unanswered message. The silence wasn’t calm. It was a screen he’d projected the worst film onto.

The fork in front of you

Now the setup you needed all along.

In 2021, researchers analyzed 932 conversations and found something that reframes both impulses: conversations reliably ended at a time that matched neither participant’s preference. One person almost always wanted it to run longer; the other wanted out sooner. The ending was a social fiction agreed to by no one.

Sit with that for a second. When you walk away feeling unresolved, that feeling isn’t a distortion your nervous system invented. It’s accurate. The conversation did end too soon — for you, and statistically for the other person too, just in the opposite direction.

The truncation is structural. Baked into how humans exit exchanges. Not a reflection of anything broken in you.

(This applies to real-time talk more cleanly than to text, by the way — asynchronous threads on something like Google Chat never have a natural close at all, which is partly why they’re worse for this.)

The asymmetry between them, and the call

The biggest risk here isn’t heartbreak. It’s spending days treating manufactured meaning as evidence — and most people underestimate how likely that is. Given that anxiety and reassurance-seeking are the documented drivers, the odds you’ll fabricate subtext from silence aren’t small. They’re the default.

So the call isn’t check versus wait. Both leak false data.

The call is to generate the ending yourself. A new message. A closed loop. One real exchange.

That’s the only move that touches the truncation.

Everything else is circling.

We’re grading three ways your reluctance to negotiate got built, and how much each one costs you. The criteria are simple: where the discomfort comes from, who installed it, and whether you can still pry the number loose.

Here’s the truth: your hesitation before asking for more money isn’t a temperament. It’s a designed output.

Research linked to scholars at Harvard, Brown, and UCLA found that workers avoid negotiating compensation more because they believe employers aren’t open to it than because they doubt their own skill. Read that twice. The barrier isn’t confidence. It’s a belief about the other side — a belief that got installed in you before you ever sat across a table.

And that belief has a long manufacturing history.

Option A: the customary rate

Medieval guilds solved a commercial problem with a social trick. They set a “customary wage” and made the posted rate feel like natural law.

When a journeyman carpenter asked for more than the guild rate, he wasn’t being enterprising. He was being rude. The number was reframed as etiquette, and violating it became a moral failing rather than a reasonable commercial move.

Modern employers inherited that playbook whole. Salary bands, “market rate,” and the increasingly common line “our offers are non-negotiable” all perform the identical function — they convert a negotiable figure into something that feels revealed, fixed, true.

I keep waiting for someone to say the guild parallel is a stretch.

It isn’t. The mechanism is exactly the same: shift the discomfort of asking onto the worker, then let the worker’s own manners do the enforcement. The peer-reviewed literature on negotiation barriers keeps circling this — the people who set wages benefit enormously when asking feels like a breach.

So the discomfort you feel before typing “I was hoping for a bit more” is the intended product. Not a defect. The output.

Option B: the compounding cost

Fine — but does the awkwardness actually pay off, or is the gain marginal?

Not marginal. According to a UCLA Anderson Review summary, workers who skip negotiation face an average annual compensation gap of $27,000 relative to those who pushed back on the initial offer. Sit with that number for a second.

Then compound it. Every raise gets calculated as a percentage of your base, so a low starting figure isn’t a one-time loss — it’s a permanent multiplier working against you for a decade.

And the raise math is thin right now. Payscale’s 2026 Salary Budget Survey projects average raise budgets of just 3.5%, down from 3.6% actual increases in 2025, with large firms already compressing that toward roughly 3%. BambooHR pegs the average actual raise higher, at 4.8% — back up to 2023 levels — which tells you how much the range depends on who’s reporting.

Here’s the counterintuitive part. The “non-negotiable” framing tends to get most aggressive exactly when the stakes are highest. Korn Ferry research, reported by The Hill, found more companies telling candidates upfront that the offer won’t move — precisely at the offer stage, the one moment your use peaks.

Watch how the pressure escalates across a career:

1. At the offer, before you’re an employee, you’ve the most room — and that’s when “non-negotiable” gets said out loud.

2. Once inside, your raise is a percentage of whatever number you accepted, so the early gap replicates itself annually.

3. Then the budget logic flattens. Less than half of organizations plan to keep performance-based increases; 18% are considering flat “peanut butter” raises, 16% have planned them, 9% already use them.

4. By the time flat raises hit, the only lever that ever bent — the offer — is years behind you.

The cost of silence isn’t abstract. It’s arithmetic, and it compounds.

Option C: the transparency counterweight

The honest objection: pay-transparency laws and benchmarking tools have shifted real power to workers, so the medieval parallel is incomplete.

Partly true. When a range is posted by law, the guild’s core trick — hiding the number so the rate feels like nature — gets weaker. You can walk in already knowing the band.

But transparency has a named limit, and it’s worth checking. Posted ranges are often wide enough to be useless — a $95,000 to $150,000 band tells you almost nothing about where you’ll land. And a range does nothing against the “peanut butter” flattening happening after you’re hired, where everyone gets the same 3% regardless of what the market says you’re worth. Transparency fixed the information asymmetry. It didn’t touch the etiquette.

In a Glassdoor and Fishbowl survey of nearly 6,700 professionals, 54% didn’t negotiate their most recent salary and 46% did. Roughly a coin flip — for a coin worth $27,000 a year.

The surprise is what that coin flip is actually measuring. Not skill. Belief. People aren’t staying silent because they can’t argue; they’re silent because they’ve been taught the door is closed. It usually isn’t.

My own confession: I renegotiated one client rate this quarter and still haven’t touched the other two. I know the number. I just haven’t typed it.

So type it. Ask for the number.

That’s the figure I couldn’t stop thinking about last Monday morning, standing in my kitchen reading [a Yahoo Finance report](https://finance.yahoo.com/healthcare/articles/fidelity-says-retired-couple-may-120408702.html) about what a married couple retiring at 65 in 2026 might need for healthcare. Fidelity’s (FIS) estimate. I read it, absorbed it, and moved on to coffee. By the time I sat down at my desk, that number had already done its work on me — quietly, without permission, before any financial conversation had even started.

That’s the mechanism I want to name for you, because it’s operating in every major financial decision you’ll make this year, and you’re the one pulling the trigger.

The position, stated flat

Your preparation — the research you do before a negotiation — isn’t your shield. It’s the weapon being used against you, and you’re the one who loaded it.

1.

Anchoring is more automatic than persuasion.

Most people think of anchoring as a sales tactic. A car dealer throws out a high number; you negotiate down and feel smart. But anchoring isn’t a trick performed by clever people. It’s a cognitive default that fires whether you want it to or not. Decades of behavioral economics research confirm this: the first number you encounter in any decision context becomes the gravitational center of every judgment that follows.

Daniel Kahneman and Amos Tversky demonstrated this in the 1970s with experiments where participants spun a rigged wheel landing on either 10 or 65, then estimated the percentage of African countries in the United Nations. The wheel was obviously random. It still moved their estimates by double digits. The anchor doesn’t need to be credible. It just needs to arrive first.

2.

You’re anchoring yourself before anyone else gets the chance.

Here’s where the warning gets specific. Before you walk into a salary negotiation, you Google average compensation for your role. Before you discuss a mortgage rate, you check three comparison sites. Before you think about retirement, you read that Fidelity says a 65-year-old retiring in 2026 can expect to spend an average of $185,500 on healthcare and medical expenses throughout retirement — a figure that rose 7.5% from the prior year’s $172,500 estimate. You absorb these numbers as neutral data.

They aren’t neutral. They’re anchors you’ve installed in your own mind before a single counterparty has spoken.

Your “research” is a pre-negotiation performed against yourself.

The person sitting across from you — a hiring manager, a financial advisor, a benefits coordinator — doesn’t need to set the anchor. You arrived pre-anchored. They only need to nudge the number you already planted, and your brain registers a small concession as a win. You feel the victory. You’ve lost.

You don’t get anchored by the other side’s opening number. You get anchored by the number you rehearsed in the shower that morning.

And that’s the tell. The $371,000 couple estimate from Fidelity? It excludes long-term care, most dental care, and over-the-counter drugs. Long-term care is often the single largest retirement health expense. So the number that’s already lodged in your planning — the one shaping how much you save, how early you retire, what you think is “enough” — is structurally incomplete. You anchored to a floor and mistook it for a ceiling.

3.

Knowing about anchoring doesn’t fix anchoring.

This is the objection worth taking seriously. Nicholas Epley and Thomas Gilovich, psychologists at Cornell, ran a series of studies on what they called “anchor-and-adjust.” Their finding: even motivated, forewarned participants adjusted away from the anchor but stopped too early. Every single time. They moved — but not enough. Awareness softens the pull. It doesn’t dissolve it. You can know every word of this piece and still be measurably shifted by the first number you encounter.

The research closes the exit marked “just be smarter about it.”

So what do you do instead? The decision rule is narrower than you’d like. If you’re entering a negotiation where the other party benefits from your pre-research (salary talks, vendor pricing, insurance planning), don’t open with the number you found — generate at least two alternative anchors from deliberately different sources before you walk in.

If you’re planning around a published benchmark like the Fidelity retirement estimate, treat it as a floor and add a specific margin: add a margin of roughly 25–35% to account for excluded categories — a heuristic, not a formula, and not derived from actuarial data. But it breaks the frame.

The structural fix above applies at any age, but the stakes shift. If you’re 30, anchoring mostly costs you in salary negotiations — each anchored starting salary compounds across decades of raises, and a $5,000 shortfall in year one can mean $100,000 or more over a career. If you’re 45, you’re anchoring around retirement planning numbers that are (as we’ve seen) structurally incomplete, with less runway to correct.

If you’re 60, the anchoring risk is acute: the gap between Fidelity’s $185,500 estimate and the true cost including long-term care and dental isn’t an abstraction. It’s a shortfall you’ll feel in your seventies.

4.

There’s one objection I can’t fully answer.

If anchoring operates below conscious awareness — and the research strongly suggests it does — then no amount of structural workaround eliminates the bias entirely. You can generate alternative anchors, you can multiply by 1.3, you can refuse to open first. But the moment you encounter any number in any context related to the decision, the pull begins. The bias is baked into how human cognition processes numerical information. I don’t have a clean solution for that. Nobody does.

What I can say is this: the difference between a person who walks in with one anchor and a person who walks in with three isn’t awareness. It’s that the second person has made the automatic response compete with alternatives, and competition — even imperfect competition — produces better outcomes than a single number sitting unchallenged in your mind.

That small refusal to let one figure own the frame is the only move available. It isn’t enough. But the alternative is worse.

I keep thinking about the $371,000 — and what it doesn’t include. It arrived first. That was enough.

I spent a decade in consulting rooms watching this framework get applied to smart, capable people. It made guilt feel logical. You weren’t starting, therefore you were failing.

And that mental loop you kept running — replaying the project, imagining its architecture, drafting opening lines in the shower — got classified as the symptom. Not the treatment. Self-help books reinforced this relentlessly. The before-picture was always the same: a person circling a task obsessively, convinced their own thinking was the problem, doubling down on shame as motivational strategy.

It didn’t work particularly well. But it was the only model available for about twenty years.

The reframe I want to offer isn’t motivational. It’s diagnostic.

Avoidance and rehearsal produce opposite behavioral signatures — and there’s reasonably consistent evidence that the underlying cognitive processes diverge in measurable ways. This is the piece most writing on procrastination leaves out — not by accident, but because it complicates the clean narrative. Genuine avoidance triggers cognitive suppression: the brain actively dampens access to the avoided material, reducing intrusive thoughts over time. You stop thinking about the thing. That’s avoidance.

Rehearsal produces exactly the opposite. Intrusive ideation. The inability to stop thinking about the task, because the brain is running background processing on an unsolved problem without triggering the threat response that formal effort activates.

The tell is almost embarrassingly specific.

If you can’t stop thinking about it, you’re not avoiding it. You’re working it. The brain treats unsolved problems like open loops — it keeps the file active until resolution. What looked like paralysis was, structurally, preparation. The change in understanding didn’t require new behavior. It required a more accurate way to read what was already happening.

Here’s an example that’s been sitting in my notes for months. A product lead at a mid-size fintech — I won’t name her, but she gave me permission to describe the situation — spent eleven days not writing a strategy document her CEO had asked for. During those eleven days, she sketched three different structural approaches in her head, rejected two, and arrived at the third with enough specificity that she wrote the entire 4,200-word document in a single afternoon.

Her manager’s read: she procrastinated for eleven days. Her brain’s read: she rehearsed for eleven days and executed in four hours. The output was the best strategic memo her team produced that quarter.

Now — and this is where I’ve to be careful — the reframe isn’t a permission slip to sit indefinitely.

The distinction between rehearsal and anxious spinning is real, and from the inside, it can feel nearly identical. Both involve persistent, involuntary thought about an unfinished task. Both produce guilt. Both look the same to your manager and to your calendar.

If your thoughts about the task are evolving — new angles, clearer structure, different entry points — the background process is working; if you’re having the same thought on loop, that’s anxiety performing the appearance of preparation.

That’s the diagnostic that makes this reframe usable instead of just flattering. Ask not just “am I thinking about it?” but “is what I’m thinking getting anywhere?” The former question excuses everything. The latter is where the work is.

One named limitation, because I think it matters: this reframe fails completely for tasks with no creative or strategic complexity. Filing your taxes, scheduling a dentist appointment, responding to a three-line email — if you’re “rehearsing” those, you’re stalling. The mechanism I’m describing applies specifically to work that requires structural problem-solving, where the brain needs to model multiple configurations before committing. Not everything qualifies.

You might be wondering what any of this has to do with retirement savings, and — fair enough — the connection is less obvious than it looks.

But consider: Fidelity’s Q1 2026 analysis found that total 401(k) savings rates reached 14.4%, with the average employee contribution rate hitting 9.6%, both record highs. Those numbers are moving closer to Fidelity’s suggested combined savings rate of 15%. Meanwhile, the IRS set 2026 IRA contribution limits at $7,500, with a $1,100 catch-up for savers 50 and older, bringing the maximum to $8,600.

Those record savings rates didn’t materialize because millions of people suddenly stopped procrastinating. They materialized because automatic enrollment and escalation features removed the decision from the loop entirely — the behavioral work got done while the brain was still circling. Which is, structurally, the same resolution: not forcing action out of a rehearsal loop, but designing around the loop’s tendency to stall.

The mechanism is the same. The domain is just money instead of memos.

I’ll say what I’ve been dancing around. I haven’t finished a particular essay I’ve been “working on” for six weeks. But the thinking has gotten sharper each week. The structure has changed three times. I haven’t typed a word of it, and I’m no longer convinced that’s the problem.

The opportunity here isn’t to relabel all your delays as genius-in-progress. It’s narrower and more useful than that. The next time you catch yourself in that loop — thinking about the task constantly, not touching it, feeling guilty — run the diagnostic. Are your thoughts evolving or repeating? If evolving, you might be closer to execution than you think. If repeating, the loop isn’t preparation. It’s a warning.

Most of what we call self-discipline is just pattern recognition applied at the right moment. Knowing which loop you’re in changes what the next move should be — and more, it changes the story you’re telling yourself about why you haven’t moved yet.

That’s not a permission slip. It’s just a more accurate map.

Not metaphorically. Literally. They cancel dinners. They ghost on group chats. They RSVP yes and then vanish forty minutes before you’re supposed to meet. And we keep treating this like a character flaw — some gap in their manners or their care for other people. It isn’t. It’s the same mechanism that makes them exceptional at their work, running in a direction you don’t want it to run.

That makes no sense, right? Sit with that for a second.

The pattern has a name in organizational psychology: opportunity cost sensitivity — the speed at which someone recalculates whether their current activity is the best use of their time. Research on high-output knowledge workers consistently ties this trait to output quality. The people who produced the best work were the ones running that calculation constantly, in the background, like an app you can’t close.

But here’s the part nobody wants to say out loud. That same sensitivity appears to run inversely to follow-through on social commitments. The pattern holds often enough to be worth naming: the better the work, the flakier the person. Probably not by accident.

The mechanism underneath is simpler than it looks. These people aren’t managing one calendar. They’re managing two. There’s the visible one — the shared Google Calendar, the Outlook blocks other people can see. And then there’s the invisible one, where their actual cognitive bandwidth is already allocated. A lunch slot that looks free on paper is already mortgaged against an afternoon deadline the other person can’t see.

By the time you’re texting “still on for noon?”, they’ve already spent that energy somewhere else in their head.

The most productive people look flaky precisely because they’re less willing to respond to ordinary requests, meetings, and messages than average workers.

Consider how most knowledge workers actually spend their time. According to Microsoft WorkLab research, the average knowledge worker burns roughly 60% of the day on coordination work — emails, status updates, meetings about meetings — rather than skilled output. Only about 27% of a typical workday goes to the kind of focused effort that actually produces something. And workers get interrupted approximately every two minutes, adding up to around 275 interruptions per day.

The high-output person has looked at that math and refused to accept it.

They’ve built an invisible schedule that protects their 27%. They guard it like it’s oxygen, because for the kind of work they do, it basically is. Highly engaged workers under ideal conditions can sustain productive output for 75–80% of their active hours. But “ideal conditions” means ruthless filtering of what gets in. Your brunch invitation isn’t competing with laziness. It’s competing with the only window they’ve found to do the work they’re actually good at.

This is where it gets interesting. The pattern is structurally identical to something you’d never expect — improvisational musicians. Great jazz players are famously difficult bandmates. They show up late, they skip rehearsals, they change the setlist on stage. Why? Because the skill that makes them extraordinary — a hair-trigger responsiveness to the moment, the ability to hear a phrase and reroute everything in real time — is the exact opposite of reliability to a predetermined plan.

Responsiveness and reliability pull in opposite directions. You rarely get both at full volume.

Now here’s the strongest counterargument, and it deserves a fair hearing. If the invisible calendar is the real one, then the honest move is to block it visibly. Share it. Let people see that Tuesday lunch is spoken for even though nothing’s on the books. Persistent failure to do that — the argument goes — isn’t a structural attention problem. It’s avoidance. Poor communication dressed up as cognitive sophistication.

That’s a reasonable objection. And it’s wrong. Not because transparency wouldn’t help (it would), but because the invisible calendar isn’t fixed. It shifts. The high-output person doesn’t know at 9 a.m. That they’ll need Tuesday lunch for deep work. They discover it at 11:47 when a problem reveals itself — and the background calculator fires immediately, repricing every commitment on the board. You can’t block time you didn’t know you’d need.

The reallocation, in other words, happens faster than the communication can.

So what do you actually do with this?

If you’re the flaky one, the work isn’t to stop running the calculation. You probably can’t. But you can name it — to partners, to friends, to collaborators. Not as an excuse. As an explanation that lets people build around it. That small refusal to let the pattern stay invisible is worth more than another apology text.

If you’re waiting on the flaky one, here’s the reframe worth carrying into the week:

1. Stop reading cancellations as signals about your importance — they’re signals about bandwidth you can’t see.

2. Build redundancy into plans that depend on them — assume a 40% cancellation rate and design accordingly.

3. Watch what they do show up for, because those are the moments when the invisible calendar actually cleared, and that’s the tell about what they value most.

Studies consistently find interruptions ranking as the top self-reported barrier to productivity. Your most productive friend has simply decided — consciously or not — that social commitments fall into that category more often than they’d like to admit. The gap between what they intend and what they deliver isn’t hypocrisy. It’s arithmetic.

We keep framing flakiness as a moral failure. It’s a resource allocation pattern. One that costs real relationships real trust, yes — I’m not pretending otherwise. But treating it as selfishness when it’s actually a wiring problem means you’ll never design around it. You’ll just keep being disappointed by people who are, in every other context, doing remarkable things.

The person who always shows up and the person who always ships — they’re rarely the same person. Why are we still pretending they should be?

Leave a comment