Programming competitions correlate negatively with being good on the job

Another way to see it is that there is a world of difference between writing a program that has to work once for a competition and writing a program that has to work and be maintained for years.

That's not a surprise really. A typical competition may be won by a programmer scoring 90/100 on four out of five tasks and 20/100 on the last one. Such performance at work would likely result in a quick termination.

That's Google, so probably quarterly review scores.

That's very interesting, especially as someone who did pretty well on those competitions. I don't doubt the finding, but I do think that it possible to put your mind in different modes. I frequently think about "running" vs "going slow" when solving problems at work. I use the former for whipping up a single use script and sketching solutions, and the latter for permanent solution, especially in hard domains.

I think knowing both modes can be useful, and knowing only one can be limiting, no matter which one it is.