Project Hong Kong

Motivation

I have started a lot of self-improvement projects and finished almost none of them, which is ordinary and not very interesting. In October 2023 I sat down to work out why, and noticed something I had been walking past for years.

When I missed the deadlines I had set myself, I was not upset about it.

That is the whole diagnosis, and it took me years to see it because the absence of a feeling is hard to notice. A real commitment produces friction when you break it. You feel something: you reschedule, you make it up, you at least argue with yourself. Serene indifference to a missed deadline means the deadline was never load-bearing, and if the deadline was never load-bearing then the plan it belonged to was not a plan.

What I had been doing, in my own words at the time, was daydreaming about an imaginary version of myself who could execute the thing I had described. Writing the plan was pleasant because it put me in that person's company for an hour. The plans had about as much use as the paper they were written on.

The failure mechanism

The mechanism is worth stating carefully because it is not laziness and it does not feel like self-deception while it is happening.

Talking about something as though it exists is, after enough repetition, nearly indistinguishable from the thing existing. Describe a system often enough and your sense of it becomes as vivid as your sense of things that are actually running. The description is doing all the work that the implementation was supposed to do, and it is doing it much faster and much more pleasantly.

So the whole design problem becomes: how do you stay tethered to reality when your own mind produces such convincing evidence that you are already there? I called the answers reality-checks, and I got four of them down.

Objective measurement

The first is objective measurement written down on persistent media.

The reasoning is short: it is easy to believe you have accomplished something without having done it, so you need a record that was made at the time, by a process that does not care how you feel about it now. Persistent is the operative word. A number you remember is a number you have already edited.

The second follows from the first and is easy to skip. Tracking on its own produces a pile. There has to be something that reads the pile and tells you what it says, or you have built an archive rather than an instrument. I wrote this down as a system for evaluation: something that turns the log into a statement about progress, so the recording changes a decision instead of just existing.

There is a dependency here that I noticed and did not like. My tracking lived inside my daily journal, which meant tracking only worked on days I journalled. So the instrument was measuring my journalling habit as much as the thing it was pointed at. Either I commit to journalling as a first-class practice, or I decouple the two and give tracking its own mechanism. Naming the dependency is most of the work; either resolution is fine and drifting between them is not.

Accountability and lenience

The third defence is accountability. Every commitment I fail to complete gets surfaced to me, and then either rescheduled or compensated with something else. Nothing is allowed to quietly vanish, because quietly vanishing is precisely how a plan becomes fiction without anyone noticing.

The fourth defence appears to contradict the third, and working out how they fit together is the actual design.

I had been building systems for a person with unlimited willpower, because that is the person I imagined while writing them. Real days vary. Energy varies, motivation varies, conditions vary. My older systems handled this badly in a specific and mechanical way: I would be consistent for a while, hit a day where I was too depleted to continue, stop, feel bad about the missed day, and then abandon the project entirely after accruing too much debt.

That last phrase is the mechanism. Missed days accumulate like a liability. Each one raises the cost of returning, because returning now means facing the pile, and at some point walking away is cheaper than paying it down. The collapse is not a failure of will at the moment of quitting. It is a debt calculation that came out the wrong way, and it was made inevitable several days earlier by a system with no way to make a small payment.

Degraded modes

So the fourth defence is that the system has to offer less-demanding forms of the original commitment on days when the full version is not going to happen.

The constraint that makes this work rather than turning into permission to do nothing is that the reduced version must keep a minimal element of the original. It is a smaller instance of the same act, not a substitute for it. An hour at the gym degrades to forty-five minutes, or thirty, or fifteen, and at the bottom of the ladder it degrades to walking through the entrance door and turning around. The last rung is almost entirely symbolic and it is the most important one, because it is the one that keeps the debt at zero on the worst day.

Both of these are true at once, and neither is negotiable. Every commitment gets accounted for, and every commitment has a floor low enough to reach when depleted. Accountability without degraded modes produces the collapse described above. Degraded modes without accountability produce a project that has quietly become optional. The pair is the design.

The time grammar

Underneath the defences is a set of units, and the reason there is a set rather than just days is that different questions need different resolutions.

Days are the fundamental unit but not the smallest. Projects are sized in an exact number of days, and dividing elapsed days by total days gives a progress figure at a glance. Inside a day I track in minutes rather than hours, because a minute figure feels more significant than the same duration expressed in fractions of an hour.

Three days is a sprint, and it has a phase per day rather than being a round number. Day one is for clarity: research, design, refining the problem statement, planning, prototyping, with the goal of knowing what you are doing and having built some momentum by the end of it. Day two is pure execution against day one's plan, aiming for something working and deliberately avoiding refactoring and optimisation. Day three continues day two through the first half and then turns to delivery: packaging, tidying the interface, fixing bugs, documenting, and at least thirty minutes writing up what happened, what was hard, what was learned, and what is left.

That last half hour is the part I would defend hardest. Without it a sprint produces work and no knowledge, and the next sprint starts from the same place.

A week is two sprints and a rest day, arranged either sprint, sprint, rest or sprint, rest, sprint depending on conditions, and I made a point of deciding which one at the start of the week rather than discovering it on Wednesday.

Two weeks is the interval at which results become visible, and I want to be honest that this number is a rough extrapolation from one domain. I had noticed that training in the gym showed up as visible improvement about two weeks after the work, and I generalised that to other habits as a rule of thumb for how long to wait before reading the instrument. Two weeks is also roughly when I stop feeling like I am doing a thing and start feeling like I am a person who does it.

The consequence matters more than the number. A system that reads its own effects sooner than the effects can appear will chase noise, and then adjust, and then chase the noise its adjustment made. The sampling interval has to be longer than the lag or the feedback loop is measuring itself.

Months are useful for looking back and poor for planning. Trimesters and quarters divide a project into phases: three phases when the project has a clear goal and a deadline, because three gives you a beginning, a middle and an end, and four phases when the project is about improving something ongoing rather than arriving somewhere. A phase should be at least two weeks, so that something can happen inside it, and at most eight, so that it does not become the whole project again.

Written out as a reference, the whole grammar is six units and each one answers a different question:

Unit	Span	The question it answers
Day	1 day	how far through the project am I
Sprint	3 days	day 1 clarity, day 2 execution, day 3 delivery and write-up
Week	7 days	two sprints and a rest day, in an order committed to in advance
Progress interval	14 days	has anything I did actually shown up yet
Month	about 30 days	retrospective only, not planning
Phase	2 to 8 weeks	3 phases against a deadline, 4 phases for ongoing work

The reason this is a grammar and not a list is that the units nest and the nesting is what carries the meaning. A sprint is not three arbitrary days, it is a beginning, a middle and an end. A week is not seven days, it is two of those plus recovery.

Daily renewal

The last piece addresses a failure I had never named before: I would forget a project was running. Not decide to stop it. Forget. And because I had never decided to stop, there was never a moment at which I had to admit anything.

So the project gets renewed every day, out loud, and the renewal contains an explicit clause saying that if I no longer want to continue I will formally resign and terminate it, with reasons. The declaration ends by saying the decision to continue is valid only for today and has to be made again tomorrow.

The point is not the ritual. It is that a project you can formally quit behaves differently from one you can drift out of, because drifting is no longer available. Every day you are either in, which requires an act, or out, which also requires an act. There is no middle state where the project is technically alive and actually dead, and that middle state is where every previous attempt of mine had gone to die.

Adjacent work: Atomic Habits

There is one inconsistency in all of this that I noticed at the time and decided to keep.

James Clear's Atomic Habits argues for gradual intensity and minimal mental exertion, on the grounds that consistency beats intensity and that anything requiring willpower is fragile. I am convinced by that argument and it is the main strategy underneath everything above. The line from the book that moved me most is that we do not rise to the level of our expectations, we fall to the level of our systems, which is the sentence this entire document is a response to.

And yet whenever I start something I begin with a burst of effort well out of proportion to what the plan calls for. A very long first session. A document written in one sitting. It functions as a signal to myself that something has started, and it works, and it is exactly what the book tells me not to do.

I kept both. The burst is a useful tactic for crossing the line from not-started to started, and gradual consistency is the strategy for everything after. The failure mode is confusing the tactic for a requirement, because then a day without the capacity for a grand gesture becomes a day you cannot begin at all, and that produces procrastination out of what was supposed to be momentum.

This document was written in one four-hour sitting after about a month of not journalling, which makes it an instance of the thing it is describing.