You need to sign in or sign up before continuing.
Reflection — An Honest Take 8 min

Honest Take — Before You Begin

Honest Take — Module 11: The Shipping Problem — Done-ness, the 90% Trap & Perfectionism #

This is the module where the curriculum finds out whether it has earned its place. Everything before this was preparation — the diagnosis, the body, the evidence base, the habits, the estimation, the deciding, the focus, the WIP limits, the AI calibration, the saying no. All of it is coherent on its own, but the integrated value of the curriculum is whether it produces shipped artifacts during this module's four-week window. If it does, the curriculum has worked. If it does not, the previous modules were theater. I am telling you this directly because the deliverable's weight matters, and softening it would make it easier to skip. The deliverable is the test: one substantively shipped thing, scoped in advance, with a written done-ness criteria document, inside four weeks.

The 90% trap is your most expensive failure mode and you have already paid for it many times without computing the cost. A half-finished release is worse than no release; a half-finished blog post sitting in drafts for nine months is dead capital; the feature branch at 90% is value that exists nowhere. The trap has a specific mechanics worth understanding, because the cure follows from it. Starting is an emotion-regulation problem and M1's tools work on it. Finishing is not a starting problem — it is a closing problem, and closing is decision-making: which edge cases are out of scope, which polish is necessary and which is performative, what exactly constitutes done. Each decision is small, but the cumulative decision-load is large, and the deferral of those decisions one at a time is what makes "almost done" a stable resting state. If you are the maintainer who cannot cut the release — the one whose changelog has been "nearly ready" for six weeks because three edge cases remain undecided — the cure is not more discipline. It is batching: sit down for ninety minutes, list every remaining decision, make all of them, and close. The decision-list session converts a long tail of micro-deferrals into one discrete event. Ship the 92% with two known issues filed; the 92% in production is worth more than the 99% in a branch, and the imagined 100% does not exist anywhere.

Curran's research on socially prescribed perfectionism is the empirical core of the module, and it will land in a particular way if you have any public surface — a GitHub profile, a blog, published packages with real users. You carry a low-grade ambient sense of what people would think of this if they saw it, and that sense distorts your shipping discipline invisibly until named. The polishing past the point of diminishing returns is not pure perfectionism; it is anticipated judgment doing the work, and Curran's data shows this form is the most rapidly rising and the most harmful, because the imagined external standard is unreachable and the polishing it demands is unbounded. The corrective is to make done-ness external in a different way: not "what an imagined audience would approve of" but "what specific testable criteria does the artifact meet." Write the criteria document before you start the work, and do not add to it mid-stream — because the other reliable trap is scope expansion at week two, the if I'm shipping this I should also include… move. That move is what killed the previous projects. Lamott's "shitty first drafts" applies operationally, and so does its inverse: shitty last drafts are also a virtue. The shipped first version is not the final version; it is the version that starts generating feedback, users, and trust, which the polished unshipped version generates none of.

Two notes on the reading and one on calibration. Godin's The Practice is not rigorous — the evidence base is mostly his own career — and I assign him anyway because shipping is partially a disposition the rigorous books do not transmit, and Godin transmits it. Curran is the rigor; Godin is the energy; you need both, and Godin will be slightly annoying for some chapters and exactly right for others. Newport's Slow Productivity principles — fewer things, natural pace, quality obsession — read differently here than they did earlier in the curriculum, because now they are operational rather than diagnostic. And the calibration note: if at week two the substantive thing is at risk, the right move is aggressive scope reduction, not more hours per day. Engineers under deadline double down on hours almost universally, and the data says it rarely works. Ship something smaller now and revise to something larger later. It will feel inadequate the first time. It gets less inadequate-feeling with practice, and the practice is the module.

Conclusion #

Module 11 is where preparation produces output or fails to. The done-ness criteria document is the contract; the four-week cadence is the practice; the substantively shipped artifact is the test. Starting and finishing are different skills — the 5-minute rule starts, the decision-batch session closes. Defend the original scope, reduce when at risk, ship the 92%, file the issues, write the retro and note explicitly what scope you cut. One deliberately shipped thing during this window is the seed of the new pattern, and the new pattern — not the artifact — is what the module is actually building.

Predictions #

  • You will pick a 90%-trap project, close it within the window, and feel more relief than satisfaction. The relief is the diagnostic of how much weight the open loop was carrying.
  • At least three of your stuck projects will turn out to share the same shape: undecided edge cases at the 90% mark. Once you name the shape, each unsticks in a single decision-batch sitting.
  • Around week two you will be tempted to expand scope. The expansion will present itself as diligence. It is the failure mode.
  • Curran's socially-prescribed-perfectionism chapter will produce a specific that is the thing that has been happening to me moment, and the vocabulary will outlast the module.
  • The done-ness criteria document will feel artificial the first time; by the third you will write it naturally; by the tenth you will not start work without one.
  • The retro will surface a specific imagined audience whose anticipated judgment you have been polishing toward. Naming the audience reduces its grip.
  • Six weeks after the module you will ship something previous-you would have polished for two more weeks. It will be a B+ instead of a delayed A-, and your own data will show the B+-now trade was correct.