r
reecha.ing

Arguments

Arguments. The long way round, on purpose.

Long-form pieces that make a case and give the strongest objection its day in court. Each one starts from a question I keep circling and stays until it earns a conclusion. Not takes. Not threads. The whole argument, counter included.

Discernment in the age of AI

What to build when AI can build anything

When AI can build anything, the constraint stops being whether you can build the thing and becomes whether it is worth building. Capacity was the old bottleneck; it is gone. Choosing is the work now, and the scarce skill is judgment about what to build, not proof that you can build it.

Read the argument

Taste was never the moat

Taste and discernment are different skills. Taste is a verdict on the thing in front of you: is it good. Discernment is a bet on what that thing does once it is loose in a system of other people, incentives, and time, and unlike taste it actually gets graded. When anyone can build anything, taste is abundant and discernment is the scarce, teachable skill.

Read the argument

The infinite menu

You decide what to build with AI by re-imposing a constraint the tools removed. When everything is buildable, the menu has no bad options to cut, and good ones cause the freeze. Declare the one resource you are short of, rank the options against each other, commit to one, and name what you are giving up.

Read the argument

The build was never the cost

You decide what not to build by pricing the whole life of the thing, not the build. Now that AI makes building cheap, "we can build it" is true of everything and decides nothing. Build only what you would still take on after counting maintenance; whatever you cannot cheaply reverse must clear a real bar.

Read the argument

The wrong question, beautifully answered

AI gives good answers to the wrong questions because it optimises the answer, not the question. Nothing in "make this good" can ask whether the thing should be built at all. Worse, a fluent answer feels correct, so its polish stops anyone reopening the question. The scarce skill is deciding what to ask.

Read the argument

Years don't turn into judgment

Judgment is built, not born, but experience alone won't build it. Practice will: a call you write down before you act, an outcome reality settles, and a correction you're forced to make. Most people have years of experience and almost no practice.

Read the argument

Range as a strategy

The polymath penalty

Being good at many things reads as a failure to commit, right up until the tools get cheap. When building is free, range stops being a liability and becomes architecture: seeing across domains is worth more than depth in any one of them. It was a penalty in a world that charged a premium for depth. That world is closing.

Read the argument

The pair test

The pair test tells a valuable skill combination from mere scatter. Your combination is an edge only if two conditions hold: the two skills are each common on their own but rarely held together, and they compound on a live problem, each making the other worth more. Everything else is a longer resume.

Read the argument

Multipotentialite or dilettante

A multipotentialite and a dilettante look identical from outside: many fields, no single lane. They split at the point where a field stops being fun and turns into work. The multipotentialite stays past that and finishes something. The dilettante quits there every time and calls the quitting range. Count finishes, not interests.

Read the argument

The generalist's moment

Generalist vs specialist in the age of AI: AI commoditises depth, not breadth. A specialist's expertise is now a cheap API call anyone can rent. The advantage moves to the integrating generalist who combines, judges, and directs cheap depth across domains, work no single skill contains. Not the dabbler, who just makes beginnings faster.

Read the argument

The world cannot file you

How to brand yourself when you do many things: do not present the list, choose the file. A market files you under the one skill it can price, so lead with one legible category it already understands, and position the rest of your range as what makes you the strongest example of that category rather than as competing labels. The label is the door; the range is the room only you have.

Read the argument

The narrowing tax

Niche down or stay broad? For most people, niche down: focus, a market that can see you, and depth all compound. But niching has a hidden cost, the narrowing tax, the optionality and the rare seam between your skills that collapsing to one thing quietly sells. Niche unless narrowing would delete an edge you can name in one true sentence.

Read the argument

The interior of hard work

What being stuck actually feels like

Being stuck is almost never about not knowing what to do. Most of the time you know exactly what to do. Being stuck is the distance between the knowing and the doing, and no better instruction closes it, because a missing instruction was never what stopped you.

Read the argument

Knowing exactly what to do and still not doing it

Knowing what to do and being able to do it are two different capacities. You can hold the exact plan and still not move, not because you are lazy or missing information, but because knowing is a description and doing runs on a trigger, and a description does not become a trigger on its own. You close the gap by building the trigger for one specific step.

Read the argument

Motivation was never the problem

Motivation is the feeling of wanting to do a thing, and it arrives after you start more reliably than before. So waiting for it keeps you still. The people who do the work are not more motivated. They act on a fixed cue and a tiny first move, and let the wanting turn up late.

Read the argument

The friction was the point

Enjoying the difficulty means finding the difficulty itself to be the point, not gritting through the hard thing on the way to an easier one. It is the opposite of a life spent arranging things so nothing is ever hard, which turns out to be its own quiet failure. When tools can remove all friction, choosing some on purpose becomes a decision.

Read the argument

The architect who will not move in

Building a system you never use is usually avoidance, not planning. There are two ways to hold a system you built: the architect keeps designing and redesigning it and never moves in, and the resident actually lives in it. Designing is safe because it touches nothing yet. Using it is exposed. The fix is not a better system. It is to move into one.

Read the argument

The silence after you ship the wrong thing

The silence after you ship a thing you cared about feels like a verdict on you, and it is almost never that. It is one low-resolution fact about the thing. The work reached no one, or reached people and did not move them, or deserved the quiet. The skill is telling which. You cannot feel it. You decode it.

Read the argument

Building one-person ventures in the open

What one person can build now the tools got free

The tools got free, so one person can now build what used to take a team. Teams became optional. The bottleneck moved from capacity to judgment: not whether you can build it, but whether you should, and whether you know what to build.

Read the argument

Teams became optional

Teams became optional. Hiring used to be the only way to add capacity, so headcount became the proxy for progress. Free tools broke that: you can now build what took a team without hiring anyone. Hiring is no longer the default path to scale. It is one option, worth taking only when it buys a capability the tools cannot.

Read the argument

What an AI operating system for business operations is

An AI operating system for business operations is the whole operational layer of a business run as one production-grade, zero-trust system, built by an architect in about a week instead of bought as a fragile prototype. It runs the operational work itself, and it is used only as the whole platform, never as separable tools.

Read the argument

The last thing they'll let it touch

When software is cheap to build, the software stops being the product. The product is trust: a business's willingness to hand a system the operational core it cannot undo, like money going out or the customer record. Code compiles in a week. Trust does not. It is earned by evidence over time, which is why it is the moat.

Read the argument

Running a markets scanner as a discipline

Systematic trading follows written rules; discretionary trading follows a trader's judgment. Markets are a low-validity environment: noisy, always changing, feedback slow and confounded. That is exactly where a gut fails to calibrate. A scanner run as a discipline, rules first, calls logged before the outcome, reviewed cold, sized for survival, substitutes the feedback the market withholds.

Read the argument

The seams are the part worth showing

Building in public is worth it when the seams are real and you are already visibly competent. The seams are the mechanism, the caught failures, and the unfinished parts, and only they hold a gradeable incident that becomes a line in a track record a rival cannot copy. It is a trap as a reach tactic, a highlight reel, or exposure of your unshipped edge.

Read the argument

You are the only reviewer, and you made the call

The real bottleneck on a one-person company is not time, money, or headcount. It is judgment. Once free tools make execution cheap, the value of any move comes from whether the bet was right, and a solo founder has no team to average out a bad one before it ships. Rebuild that grading yourself.

Read the argument

How power works

How power actually works

Power is not one thing you have or lack. It comes in seven separate kinds, each with its own rules: position, relationships, information, story, money, force, and legitimacy. Most people are fluent in one and cannot read the other six, which is why they keep losing to people who, on paper, hold less.

Read the argument

The seven domains of power

There are seven kinds of power, and they are not variations of one thing. Position (the org chart), relationships (who owes whom), information (who knows first), story (whose version wins), money (who pays), force (who can hurt you), and legitimacy (who is believed to be in the right). Each is won differently, faked differently, pays out in different rooms, and trades for the others only at a loss.

Read the argument

The six P's

The six P's are the six domains a deliberately designed life runs across: Physical, Psychological, Pecuniary, Professional, Philomathic, People. It is not a to-do list or a balance chart. It is a way to check that the parts hold each other up instead of one domain quietly eating the rest, each domain scored on its own scoreboard, because a surplus in one cannot pay a deficit in another.

Read the argument

Beauty is a standard

In serious work, beauty is not decoration added at the end. It is a standard, because the clarity that makes an idea sound is what lets the work be made clean, and a thing that cannot be made clean is usually a thing not yet clear. Finish signals care and works as a fast, public proxy for the thinking underneath.

Read the argument

Philosophy is a motor skill

To apply philosophy to daily life, stop trying to recall it and start drilling it until it runs on its own. The skill is not knowing the idea or being able to quote it. It is having it at fingertip feel, the way a touch-typist finds the keys without looking. You get there by using the idea, badly at first, in situations it was never written for, until it stops being something you retrieve and becomes how you already see. If you still have to name the source, you are still looking at the keys.

Read the argument

They orbit a handful of questions I cannot put down:

  • What is worth building, now that almost anything can be built.
  • Whether being many things is the liability everyone warned you about, or the actual edge.
  • What hard work feels like from the inside, past the part the productivity advice will admit to.
  • What one person can build now, and what still cannot be bought or rushed.
  • How power actually works, for the people who were never handed the manual.