There's an old saw in the AI community about Minsky talking to a student training a tic-tac-toe neural net. When he asked why the net was randomly-wired initially, the student responded that they didn't want it to have preconceptions of how to play. Minksy responded that it still had preconceptions; we just didn't know what they were.

Eliminating bias is a mu-goal. An unbiased electrical pin isn't a perfect signal; it's a floating input, passing along whatever radio flux slips through it. Our goal is always to seek out the right biases.

I must be missing something; doesn't everyone already think their biases are the right biases?

Generally. But some people are correct in that opinion and some are incorrect.

We can fool ourselves frequently. Nature cannot be fooled.

I'm not convinced.

Homo sapiens have had access to nature for 160,000 years. We've spent a lot of time thinking we've folded nature when we had no right to. Why should I think now is any different?

if one believes in objective reality, it stands to reason that it is possible for a person to believe things that match to that reality and it is possible for a person to believe things that do not match to that reality.

My claim is we should bias towards believing things that match to reality. Seeking "unbias" is seeking ignorance.

I've given up that people are capable of understanding basic concepts like operationalization let alone identify when and where they are doing it in their day to day life. The closer someone positions themselves as a rationalist, the further they are from understanding how they are participating in the collective action of coloring reality with their biases.

It might be theoretically possible for people to inch toward objective reality, I've never seen it work in practice. It's a failed ideology and for some reason folks believe that this, the millionth time, it's going to work.