Hacker News Reader

A stochastic approach would work quite well. Have each pixel computed by n people and take the mode.

I suppose by using a non-human shader and comparing the results :D

Well then why is showing so many mistakes? It makes the experience worse.

I mean if you want to do it perfectly with no chance of error, let a computer do it, or just find the reference image. This is communal mathematics, and the errors are a part of the experience.

Saying the mistakes make this worse is like saying "I'd really enjoy spaghetti, I think, but all the noodles are ruining it."

I suppose they won't use a non-human shader, since it is the whole point.

the final step in the worksheet provides an error correction heuristic and procedure:

> Thanks a lot for being part of the Human Shader, go find your pixel in the public canvas! Tip: if its color looks wrong to you, feel free to review your calculations and submit again with the same code!

By scrubbing through the duration I haven't found a single error pixel that has been corrected, though. Which is a pity, as they do stand out quite glaringly.

by eliminating the pixels where the proof-of-work is incorrect