I'm all in favor of doing more to help obscure sites and having less major-media and $BigCo stuff, but there are limits. A site being obscure or having original content by no means implies that it is interesting in HN's sense. If you try to encode those criteria into software (and we've tried many times) the median-quality post comes nowhere close to clearing that bar, so you still need human curation, and that is basically the status quo. If you look at https://news.ycombinator.com/pool you should see a lot of such sites.

Also, a lot of those media and BigCo stories really are of interest to the community. We try to dampen the stuff that's repetitive, and most of those sites are downweighted by default, but HN would not get better if they were excluded. It's all just more complicated than it seems like it might be.

What ultimately matters is how interesting a story is, not what site it comes from. I'm suspicious of encoding proxies for that, because it would be easy to end up optimizing for the wrong things. https://hn.algolia.com/?dateRange=all&page=0&prefix=true&sor...