The Compression

The shift I observe at Sliday is from routine production towards judgement: reading a founder’s repository, choosing what survives and deciding what needs hardening. We have not established a measured percentage for that shift. The two weeks below illustrate the difference; they are not time-sheet records.

Much of a traditional studio week went into skilled, repeatable work. A senior engineer could apply a pattern he had used on earlier projects without needing to reconsider the product.

A senior engineer might spend a 2022 Tuesday writing a database migration whose shape he had written eleven times already, adding a forgot-password flow to a product, authoring a test suite against a controller he had sketched the previous Friday, updating copy in a landing page because marketing had asked, reviewing a junior’s pull request that touched a pattern he had designed himself and was now debugging a third time. None of those tasks would have surprised him. None required a genuine call. Each required a competent pair of hands and a couple of hours, and the bill went to the client as engineering work at a senior rate. The judgement work — the architectural call about which service owned which write, the Thursday-afternoon conversation about what to do with a founder who had changed his mind for the fourth time, the half-hour spent telling a designer that the pattern she had drawn would not survive a real keyboard — was threaded through the week in small amounts. It decorated a day without dominating it.

That illustrative Tuesday looked like an ordinary studio day. The diary was full of meetings. The terminal was full of git diffs. The office (by 2022 a shared video call; we had given up the Auckland lease in 2020 and not looked back) rang with the ordinary sounds of a studio keeping several projects alive at once. Work began at nine and finished at six and the hours in between were tractable ones. You could, and most of the team did, carry a conversation with a partner over dinner without trailing off mid-sentence.

In the second illustrative week, the founder arrives with a repository. The engineer has an artefact to judge before he has a build to begin.

The same senior engineer who wrote that migration in 2022 now reads a founder’s repository on Monday and spends Tuesday making calls about it. Which three of the seventeen features survive to ship. Whether the auth model the founder hacked together on Saturday is worth hardening or worth throwing out. Whether the third-party integration the model suggested is a library the studio will support in production. Whether the founder is being defensive, or whether she is right about the dropdown. How much of the prototype’s stack can be kept and how much rewritten without costing the engagement what it is being paid for. Each of those is a call. None is rote. Each takes twenty minutes of steady thinking or an hour of talking, and between the calls he writes code, but the code is a compressed expression of a decision rather than a keyboard exercise. He removes code as well as adding it, and the decisions carry the week.

The routine work also takes less of the week. Scaffolding a service — a thing that used to cost half a day — is now an afternoon at most, because the model does the boilerplate and he checks it. Writing first-pass tests is a morning instead of two. The rote moved from his keyboard to an agent the studio pays for, and from the centre of his day to the margins. What is left at the centre is the judgement.

The Recalibration walks this shift at the level of specific people — a senior designer, a senior engineer, a QA lead, a content lead — and the detail there is the honest texture of what being inside the change feels like. The chapter you are reading is the altitude view. From up here the headline is cleaner: the rote has migrated, and the judgement owns more of the day. The studio hour costs the same. The studio hour contains more.

I want to sit with the word denser, because a studio owner reading this wants the commercial consequence and a senior engineer reading it wants the lived consequence, and the two are the same sentence said at different volumes.

Judgement hours are expensive hours, in the specific sense that they are the hours that cannot be further accelerated by buying more tools. A 2026 rote hour has a model inside it doing half the work. A 2026 judgement hour does not, and cannot. A judgement hour is a person sitting with a problem the model cannot hold the whole shape of, and the person is the one holding the shape. Two of those hours back to back cost more of a person than eight rote hours did, because the cost is cognitive rather than manual. At the end of a 2022 day of rote work a senior engineer was tired. At the end of a 2026 day of judgement work a senior engineer is a different kind of tired — the kind that does not recover over lunch, the kind a weekend barely touches.

We have watched this in the team. David has watched it in himself. The most common quiet complaint inside the studio through the second half of 2025 was density. Nobody said I have too many hours on my calendar. Several people said, in various words, the hours I have are harder than they used to be. They were describing the same phenomenon the spreadsheet was measuring.

The calendar that used to hold six engagements and sixty hours of rote a week now holds six engagements and thirty hours of judgement a week. It looks shorter than it did; the hours inside it are not.

The trap is the one a studio owner falls into first, and it is the reason I am writing this chapter in front of the hardening one. If you look at the calendar and not at the density, you will price the engagement against the hours on the wall clock rather than against the hours that are costing you a senior engineer’s morning. We did this, repeatedly, across the first three engagements of 2025. I will not name them; the failure was ours, not theirs.

Each of the three was quoted at something like the old hours, because the visible work — the code that shipped — was visibly less than a comparable 2022 engagement would have contained. Each then consumed more of the senior team’s judgement than any 2022 engagement had. We came in under the hours. We came out, in the studio’s internal accounting, on two of the three below the rate we should have been billing. The mismatch was a shape error rather than a pricing error in the narrow sense. We were selling judgement as if it were code, and quoting the code, and billing the code, while the judgement was the thing actually being consumed.

The pricing chapter later in the book argues the fix — shape the invoice around the scarce thing, not the visible thing — and I will not steal its argument here. The chapter you are reading is the one that names the mistake early enough for the pricing chapter to matter. A studio that does not notice the density has priced the wrong thing, and a senior engineer who does not notice the density will be exhausted by Thursday without knowing why.

There is a cheerful reading of this shift and an honest one, and a book worth writing owes both.

The cheerful reading is that senior engineers now spend the majority of their week on the interesting work. That is true. The boilerplate they grumbled about for fifteen years has mostly left the calendar. The review-the-junior’s-scaffolding hour is ten minutes. The first-draft-of-anything hour is gone altogether. What is left is work that, a decade ago, they would have told you was the reason they stayed in the profession.

The honest reading is that the interesting work, done every hour of the day, is harder to sustain over a year than the old blend was. The rote hours were not only filler; they were also recovery hours. A senior on a long Tuesday used to pull a migration out of muscle memory between two hard conversations and catch his breath inside the task. That rhythm is largely gone. We are — David and I, and the team — still learning how to plan a senior’s week now that the judgement is the centre of the week rather than the seasoning on it.

We have made three changes so far that have held. Shorter engagements, or engagements with explicit no-meeting mornings baked into them. A cap of two judgement-heavy engagements running at once for any one senior, instead of the three or four we used to let roll because the rote hours absorbed the overhead. A Friday that is explicitly not a deliverable day, which we call a rotation — the polite name for a recovery day for the brain. None of these is a finished policy. All of them are the consequence of the number in the first paragraph being true.

The number moved because the work moved, and the work moved because the tools did, and a studio that does not move its week in response is a studio running a four-hour day through a twelve-hour calendar. I will argue it more bluntly in Pricing the Cut, where the commercial defence is worked line by line. It is enough, here, to say that the hour we sell in 2026 is not the hour we sold in 2022. Treating it as the same hour is the commercial mistake of the cycle, and we have made it often enough that I can write about it from the inside.

The calendar got shorter and the Tuesday got heavier.