Hacker News Reader

It is not monkey work. But A/B testing can often be more a trapping of scientific than the real thing. Google UXR is hit or miss. They have done a number of poorly run studies on what shade of blue to color buttons without thinking about multiple comparison corrections, seasonality or evolution in display technologies. I am increasingly unconvinced that these efficiencies are meaningful for UX. Compilers are different and those optimizations are real and matter for both the bottom line and UX.

I was using the phrase “turning the red button blue” as a description of the fundamental toil that goes into maintaining complex systems. There might be thousands of people who find that work interesting, but there are millions of buttons.

Learning whether the button should change colors is different from actually changing the colors.

I agree that the learning part is interesting but at google I suspect those are two different jobs.

> I could be wrong as I never worked in the hyper-optimized space, but I don't think that's necessarily a boring problem to work on. It's highly scientific and I'm sure a lot of talented engineers are interested in A/B testing.

That's why Google ends up with 25 shades of blue because each button is "scientifically interestingly A/B tested to death": "When you desgined 5 screens in Figma, but couldn't be bothered to chose the better buttons, fought with managers and requested an A/B/C/D/E test" https://twitter.com/nikitonsky/status/1555686876274630656 and https://twitter.com/nikitonsky/status/1557755746686754818