The routine you never wrote down

Try this now, without looking anything up: state your routine. All of it, in order, including how long each part is meant to take.

Most people can’t. They can produce a rough version — there were six things, or possibly seven, and one of them you were never sure you were doing properly. What they can’t produce is the specification they are supposedly following, because it doesn’t exist anywhere outside a memory that has been quietly revising it for weeks.

And the revisions all go one way. The item you liked least stopped appearing about a month ago. The one at the end got shorter, then got dropped on busy days, then just got dropped. You didn’t decide any of that, and you’d have argued against it if anyone had proposed it, and the routine still has the same name it had at the start.

Why an unwritten routine erodes downward

Every session is a copy of the last session, not of the original. If the routine only exists in recall, what you do today is reconstructed from what you remember doing yesterday — which was itself a reconstruction. Small omissions therefore compound rather than cancelling out, because each one becomes part of the source material for the next. This is a mechanism rather than a measured effect, but it’s easy to check on yourself, and the check is the point of this post.

The drift is directional, not random. If it were random, some sessions would come out longer. They don’t. What falls out is specifically what’s least pleasant, most awkward, hardest to remember the details of, or last in the order — so the routine converges on its most agreeable subset. That’s not a character finding; it’s what any unconstrained system does under repeated small edits.

Nothing about it registers as a failure, so no rule fires. A missed day is visible and there are procedures for it. A session where you did four of six things records as done, feels done, and gets a tick. Anything that counts occurrences rather than content is structurally blind to this, which means the more diligently you track, the more confident you get about a routine that’s shrinking.

The floor goes with it. A routine designed with a small bad-day version has two levels, and the gap between them is what makes bad days survivable. When the everyday version erodes toward the floor, that gap closes. You’re now doing the minimum every day, calling it the routine, and you have nothing left underneath it for a day that goes badly.

And you can’t reason about a routine you can’t state. Every useful decision downstream — whether to add something, whether the length is the problem, what actually failed last time — requires knowing what the current version is. Adding to a routine on top of an unspecified base is adding to something you can’t see the size of.

Write the spec. Four lines, one place

The intervention is small and clerical, and it is emphatically not a document you maintain.

What goes on it:

The items, in order. Named as briefly as you like, as long as you’d recognise them. This is the list, and its function is to make an omission explicit rather than invisible.

The floor version. Which of those items survive on a bad day, stated in advance. If this line is missing, the card becomes a source of guilt on hard days instead of a structure, and you’ll stop reading it.

The cue. The event the routine follows. Writing it down is what lets you notice later that the event has stopped happening.

The rule for a missed day. One sentence. In-the-moment judgement is precisely what fails.

Where it lives matters more than what it looks like: one place, and it has to be visible at the moment you’re doing the routine, not filed somewhere you’d have to decide to go. A card, a note on the wall, a single pinned note. If you have to remember to consult it, it’s part of the memory problem rather than the fix for it.

THE TEST — is your routine still the routine

  Write from memory what you think it is. Then compare
  with the last version you actually decided on.

  · Can't state it at all
        → the spec is gone. Rewrite it today.
  · Items you haven't done in weeks
        → erosion. Decide: keep or formally drop.
  · You do it in a different order now
        → harmless. Rewrite the card to match.
  · Everyday version = the floor
        → you have no floor. Rebuild the gap.
  · Matches → the routine is what you think it is.

The second row is the common one, and the useful move there is to make a decision either way. Dropping an item on purpose is fine and often correct. Dropping it by attrition is what leaves you with a routine you can’t describe.

The spec is not tracking, and it isn’t a project

Two objections, both fair.

“It’s six things — I don’t need a list.” The failure isn’t forgetting what the items are. It’s that the order, the durations, and the reduced versions are unrecorded, so each is free to move a little each week without anything registering. Nobody notices a routine getting very slightly shorter.

“Writing it down turns it into admin.” The spec and the record are different objects. Tracking accumulates — it grows a row every day and it costs you something daily. A spec is static: written once, consulted, edited when you deliberately change something and not otherwise. If yours is being edited weekly, that’s a signal about the routine rather than the card.

One boundary. If the routine came from a clinician, the spec is theirs and not yours to revise: write down what you were actually given, and if the honest version is that it’s too long to do, that’s a conversation with them rather than a private trim. Writing it down is what makes that conversation possible, because “I’m doing four of the six” is a useful thing to be able to say, and the accurate number is the whole value of reporting it.

What this costs

A written spec removes deniability. You will find out that the routine you’ve been describing to yourself for two months is not the one you’ve been doing, and that’s uncomfortable enough that some people respond by ignoring the card. If that’s the likely outcome, write the floor line first and the list second — a card that mostly tells you you’ve failed is one you’ll stop looking at.

It also reduces flexibility slightly. An unwritten routine can be informally trimmed on a hard day with no friction at all, which is genuinely useful on hard days. Writing the floor down is what preserves that, and it’s why the floor line is not optional.

What this doesn’t fix

It says nothing about whether the routine is the right routine. What should be in it isn’t a question this site can answer — the spec only makes the version you have visible.

And it doesn’t prevent drift. You’ll still shorten things; you’ll just be able to see it, which converts erosion into a series of small decisions. That’s the entire claim.