Eleven things a day

In January 2020 I had a question I could not find a satisfying answer to: how much can a person actually learn in a day?

Not in theory. Mine, specifically, measured.

So I ran it as an experiment. I picked three subjects — Python, Russian and JavaScript — on the assumption that three simultaneous learning tracks was roughly the ceiling for a working adult.

It was not. Over the following weeks I kept adding. I finished at eleven.

What came out of it

By the end I had the basics of Python, the Russian alphabet and enough reading to sound out words, JavaScript, HTML and basic CSS, and an introduction to algorithms and binary search.

Two of the eleven failed outright: reverse engineering and statistics. I could not get traction on either at the time.

I want that failure rate on the record, because a version of this post that only lists the successes would be useless. Two in eleven did not work. The reason matters less than the fact that the experiment still succeeded overall — nine subjects moved from nothing to something, which was nine more than the “focus on one thing at a time” advice would have produced.

Why parallel worked better than serial

The conventional advice is to learn one thing properly before starting another. In my experience that advice quietly assumes your motivation is constant, and it is not.

Running several tracks at once meant that when I hit a wall on one — and there is always a wall — I had somewhere to go that was not “stop learning today.” Progress on Russian vocabulary on a day when Python was not working still ended the day with something achieved. Momentum survived the bad days, which is the only thing that matters over a span of months.

It also produced connections I did not plan for. Learning how websites work made SQL injection make sense, which pulled me toward security, which is now a significant part of what I do. None of that was on the roadmap. It came from having several doors open at once.

The stopwatch

The other half of the method is embarrassingly simple: I time things.

In June 2020 I needed to update my CV, and I did not have Office installed. It would have taken an hour to download. Instead I decided to learn LaTeX — which I had never used — and write the CV in it. It took about an hour, start to finished document.

There were easier ways to write a CV. That was not the point. The point was to put a clock on an unfamiliar task and find out what an hour of concentrated effort against something new actually buys.

I have kept doing this for six years. Java basics in four days. Android Studio in five. A full augmented reality project in four. A framework I had never touched, shipping a client project in one. Odoo, from no API knowledge to a working integration, in two weeks.

The times are not there to impress anyone. They are there because a measured time is the only defence against the feeling that you are not making progress. The feeling lies. The stopwatch does not.

What I would change

Two things, honestly.

Eleven was too many. It worked, but the depth was thin — I got the basics of nine subjects and mastery of none. That was the right trade in 2020, when the objective was to find out what was possible and to cover enough ground to know what I actually wanted to pursue. It would be the wrong trade now. Breadth is a phase, not a permanent strategy.

I under-planned the recovery. Adding tracks is easy. What I did not build in was what happens when a track stalls for a month. I worked that out later, and it became the most useful thing I know about goals: a sub-goal that overruns has not failed. Pause it, keep the others moving, come back. Drifting off the path is not the same as the plan failing.

If you want to try it

Start with three, not eleven. Add a fourth only when three feels comfortable rather than when it feels heroic. Put a clock on the first session of anything new so you have a real number instead of an impression. And decide in advance what you will do when one of them stalls — because one of them will.

Then accept that some of them will not work. Two of mine did not, and it is still the single most useful month I have ever spent.