Thanks for playing! I agree there is some risk of confirmation bias, and the option to hide explanations by default is very interesting.
The reason it is designed the way it is now is because I’d prefer to avoid too many duplicate explanations. Currently, you can only submit explanations that are not exact duplicates, though you can submit explanations that are very similar -e.g, “banana” vs “bananas”.
The first downside would be that duplicate explanations may clutter up the voting options. The second downside is when someone is looking at the two explanations later, the vote may be split between the two similar explanations—meaning a third explanation that is worse might actually win (e.g, “cherry” vs “banana(s)”).
HOWEVER—those are not insurmountable downsides. the server just has to have a better duplicate/similarity check (maybe even asking GPT4), like check for plurals—and if you explain similarly to an existing explanation, it just automatically upvotes that. I think it’s definitely worth experimenting. The similarity check would have to not be too loose, otherwise we may lose out on great explanations that appear to only be marginally different but actually score very differently.
Please keep the feedback coming and join the discord if you’d like to keep updated.
Hi Martin,
Thanks for playing! I agree there is some risk of confirmation bias, and the option to hide explanations by default is very interesting.
The reason it is designed the way it is now is because I’d prefer to avoid too many duplicate explanations. Currently, you can only submit explanations that are not exact duplicates, though you can submit explanations that are very similar -e.g, “banana” vs “bananas”.
The first downside would be that duplicate explanations may clutter up the voting options. The second downside is when someone is looking at the two explanations later, the vote may be split between the two similar explanations—meaning a third explanation that is worse might actually win (e.g, “cherry” vs “banana(s)”).
HOWEVER—those are not insurmountable downsides. the server just has to have a better duplicate/similarity check (maybe even asking GPT4), like check for plurals—and if you explain similarly to an existing explanation, it just automatically upvotes that. I think it’s definitely worth experimenting. The similarity check would have to not be too loose, otherwise we may lose out on great explanations that appear to only be marginally different but actually score very differently.
Please keep the feedback coming and join the discord if you’d like to keep updated.