I have been hoping for something like this for AGES, finally found it exists in small forms already: there was a negative result workshop at CVPR2017 http://negative.vision
(and a journal for intersting negative results: http://jinr.site.uottawa.ca/, but it seems dead).
arxiv.org helps a lot in reducing publication bias in ML research, but I think we should incentivize people keeping track and publishing all their silly or unfinished ideas as well.
There is a dissenting opinion on whether they are worth it though: https://luckytoilet.wordpress.com/2018/03/07/publishing-negative-results-in-machine-learning-is-like-proving-dragons-dont-exist/
As a general instance, we are not centered on a specific theme, or a specific language. Everyone is welcome as long as they follow the few rules we have.
We always run the latest stable version of Mastodon, providing you a stock experience, with the following modifications:
Although we are a medium+ sized instance, we'd like to keep the feeling that you are at home and safe here.