Why I built Drift

The starting point was not very original: I would spend half an hour in a feed and afterwards could not have told you much about what I had seen. Cutting down on it did not really help, because the amount was not the thing bothering me. Half an hour of reading something I had chosen would have been fine.

So the question I ended up with was whether a feed could work differently, rather than whether I should use one less.

Worth being specific, because the mechanics are well understood and they are not accidental. Three things do most of the work:

• Variable reward. You do not know whether the next item will be worth it. The uncertainty is what keeps you pulling.
• No end. There is no last item and no point at which the thing is finished, so stopping is always an interruption rather than a completion.
• Hidden selection. You choose from what you are shown, and you cannot see how that set was picked.

Roughly, it inverts each of those, and then does not trade them back when something easier comes along.

Visible options instead of uncertainty. Each card shows a few labelled directions you can go next. You are picking rather than finding out. This works less well than variable reward at holding attention, which is the intended trade.

A defined end instead of an open one. A session starts somewhere you chose, wanders, and finishes with a map of where it went. The map is only reachable by stopping.

A stated reason instead of a hidden one. Each card says why it is there: the thread you pulled, the field you picked, or “drifting” when nothing was chosen. The rule I use when deciding whether to build something is whether its reason fits in four words on a card. That rules out more than anything else does.

This was meant to be a small closing screen and ended up being the part I use most. Seeing a session drawn out, with the stops and which threads were pulled, makes it easier to remember afterwards than a list of pages would be.

It also sits in an awkward place commercially: the most interesting thing in the app is behind the decision to stop using it. Moving it earlier would almost certainly increase how long people stay, and would remove the reason it works.