The Best Thing We Never Built

The easiest mistake to make once you discover that you can build software quickly is, rather predictably, to start building software quickly, and I know that because I have been extremely good at making that mistake myself.

Creation has become cheap enough that building can become a reflex. Someone mentions a problem and, before they have finished the sentence, you are mentally designing the screen. I do this. I have to catch myself doing it.

The better sequence is less exciting, which is probably why it took me longer to learn. Remove, simplify, reuse, connect, build, buy. I do not mean that as a rigid six-step religion, complete with laminated cards and a consultant charging by the day, just a useful order in which to ask increasingly expensive questions.

Can we remove the thing causing the work? If not, can we simplify it? Is there something we already have that does enough? Can two existing things simply be connected? Only then should building become the obvious next move. And sometimes, after all of that, buying is still the smartest decision.

I had a request for a site that would let people check gift-card balances. Perfect little app territory. It would have been satisfying to build. Then I looked at the real volume, the login and security considerations, and the amount of effort I would introduce around a problem that a person could currently resolve faster. I did not build it.

There is no exciting demo at the end of that story, but I now think that is exactly why it counts as a successful one.

I also built a launcher for the growing collection of internal tools. Sensible enough as an idea, but perhaps a little ahead of its own need. Not enough had gone live in one brand to make the launcher essential yet. There is something wonderfully self-parodying about reaching the point where you need an app to find your apps. If this approach ever gets a corporate crest, perhaps that should be on it as a warning.

The highest form of this work is not a clever little application. Sometimes it is deletion. Sometimes it is deciding that the friction is cheaper than the fix. Sometimes it is admitting that the thing you were itching to build would mostly create another password.

Software is one possible intervention, sometimes a very useful one, but it was never supposed to become the objective.

The objective, for me, is to give people back some of the time and energy that unnecessary friction has quietly stolen from meaningful work.

Attention is a cost too. So is ownership. Every little thing you create needs somebody to remember what it does, where it lives, who can access it and what happens when it stops working. That cost may be tiny, but it is never quite zero. Ten tiny costs have a habit of becoming one medium-sized headache while everybody is congratulating themselves on avoiding the large one.

I now try to ask one slightly annoying question whenever an idea starts becoming a screen in my head. What happens if I do nothing? Sometimes the answer is that people keep losing hours every week and the case for action becomes stronger. Sometimes the answer is that somebody spends three minutes on it twice a month. In that case the cleverest technology decision may be to close the laptop and finish the coffee.