Whyday August 19

Whyday in Two Hours

Most of this site assumes you have an evening. Plenty of people do not. There is a job, or a child, or a commute, or the nineteenth lands on a bad week, and what you actually have is the gap between nine and eleven.

Two hours is enough. It is not a lesser Whyday. It is a different shape, and the main thing to understand is that the shape changes what you should attempt — not how much of the same thing you get through.

For teams trying a short creative session inside a normal workday, work hours in a year is one practical reference for keeping the experiment distinct from routine delivery.

What two hours is not

It is not a quarter of an eight-hour project. Small blocks of time do not scale linearly, because the fixed costs stay fixed: deciding what to build, remembering how the tool works, getting the first thing on the screen. In an evening those costs are noise. In two hours they can be the whole thing.

So the two-hour version is not "start something and get a quarter of the way in." A quarter of a project is nothing, and it will make you feel worse than not starting, which is the opposite of the point.

The two-hour version is a complete small thing.

For broader programming and making context related to this topic, The Odin Project is an independent reference worth comparing with the material here.

The shape

Zero to fifteen minutes: nothing new. No new language, no new framework, no new editor setup, no new machine. Whatever is already installed and already familiar. This is the single highest-leverage rule on the page — installing something is how two hours becomes twenty minutes.

Fifteen to twenty: the first output. Get something on the screen or into the speakers within five minutes of starting to type. Anything. A window, a printed line, a tone, a coloured rectangle. If you cannot get any output in five minutes, the project is wrong for the slot; change it now rather than at minute fifty.

Twenty to ninety: the actual work. One thing. Not a thing with a settings screen.

Ninety to a hundred: stop adding. This is a hard boundary and it is the one people miss. At ninety minutes you stop building features, whatever state you are in.

A hundred to a hundred and twenty: make it showable. Name it. Write three sentences about what it is. Take a screenshot or record ten seconds of it. Put it somewhere. This half-hour feels like waste and it is the difference between having made something and having had an evening.

What fits in two hours

Things that fit:

Things that do not:

The dividing line is roughly: does this need infrastructure, or just logic? Logic fits in two hours. Infrastructure does not.

Pick before you sit down

This matters more here than anywhere. In an evening you can afford fifteen minutes of choosing. In two hours that is an eighth of everything you have.

Decide on the commute, in the shower, in a meeting. The list of twelve is there for this; so is the constraint list, and for two hours the useful constraints are the ones that remove decisions rather than the ones that make things strange. One file. Standard library only. No configuration. Save the weird ones for a year when you have the whole evening.

If you have a long-postponed idea, two hours is usually not its night — work out which pile it is in first, because the "it would take too long" pile often contains something whose smallest convincing version does fit.

Set the alarm properly

Not a mental note. An actual alarm at ninety minutes and another at a hundred and ten.

The reason is that two hours passes in one block of attention. There is no natural break in it, no point where you look up. Without an alarm you will look up at one in the morning having added a configuration file to something that will never be configured, which is a specific and very common way to lose the night.

Two hours a week beats eight hours a year

The last thing, and it is really the argument for the whole day.

If two hours turns out to be a good shape for you, do not wait until next August. The date exists because play does not survive a calendar unless something puts it there, and a recurring Tuesday slot does that job better than an annual one.

The 2010 text made this point itself: the things worth celebrating are not for one day a year. The day is a starting mechanism, not a ration.

The short version