If you want a response to this and you don't see one here in the next few hours (I'll try, but can't right now), you're welcome to email hn@ycombinator.com. The short version is that you're partly right on #2 but probably less than you think, and definitely not on #3.
Edit: ok, here you go. Long, but hard to make shorter.
Here are all the things that can make a post fall in rank: user flags, software penalties (for voting rings, flamewars, etc.), and moderator downweights. All try to optimize HN for the core value of the site—intellectual curiosity—and all are mechanisms to countervail the upvoting system in some way. Why? Because pure upvoting optimizes for other things than intellectual curiosity, such as outrage, gossip, and gaming the system.
Think of HN like a flower garden. The flowers are stories that gratify intellectual curiosity, and to have these we must protect the garden against weeds, rabbits, motor vehicles, and other things that otherwise would soon take over.
User flags and software take care of much, but not all, so moderation is needed. (We'd love to make it not be needed, but that's a hard problem.) Our criteria for downweighting come from a general sense of what attracts upvotes for reasons other than intellectual interest. It's not about what we personally believe, or have a vested interest in, or even what we happen to be interested in ourselves, with the possible exception of APL.
When we downweight, we try to use a light touch. When you see a story plummet off the front page, it probably wasn't moderators who did that (or there was a clear-cut reason, such as a duplicate). In judgment call cases, we're more likely to make a story go from #1 to #4. It's all about balance.
You may reply that "a general sense" is hopelessly subjective, and I'd have to agree, yet it's not as subjective as you might think. For example, we wouldn't downweight a substantive article about startups just because it disagreed with YC's philosophy. We might do so if it included a lot of tricks to stir up controversy. Do you see the difference? The latter is procedural; not much different than editing linkbait out of titles. If you do this job across tens of thousands of stories, and you're conscious of the obvious biases, you can mostly guard against them, especially since the same patterns come up over and over.
Not sure if I've succeeded here in getting across the gist of what we do, but I hope it helps a bit.
As for #3, that's an easy one. We don't give YC startups extra favor when it comes to boosting stories. If a Stripe API went to #1, that was because the community heavily upvoted it. We also are less likely, not more, to downweight stories when they are critical about a YC startup, such as the thread we're currently commenting in.
Hmm, no idea. But we don't see a lot of flag brigading on HN, and there aren't that many posts in that list, so I guess the effect is minimal?
A surprising thing about HN is that the flagging system works remarkably well compared to the voting system. You'd think (well, I'd have thought) that they wouldn't be so different, but they are.
Edit: ok, here you go. Long, but hard to make shorter.
Here are all the things that can make a post fall in rank: user flags, software penalties (for voting rings, flamewars, etc.), and moderator downweights. All try to optimize HN for the core value of the site—intellectual curiosity—and all are mechanisms to countervail the upvoting system in some way. Why? Because pure upvoting optimizes for other things than intellectual curiosity, such as outrage, gossip, and gaming the system.
Think of HN like a flower garden. The flowers are stories that gratify intellectual curiosity, and to have these we must protect the garden against weeds, rabbits, motor vehicles, and other things that otherwise would soon take over.
User flags and software take care of much, but not all, so moderation is needed. (We'd love to make it not be needed, but that's a hard problem.) Our criteria for downweighting come from a general sense of what attracts upvotes for reasons other than intellectual interest. It's not about what we personally believe, or have a vested interest in, or even what we happen to be interested in ourselves, with the possible exception of APL.
When we downweight, we try to use a light touch. When you see a story plummet off the front page, it probably wasn't moderators who did that (or there was a clear-cut reason, such as a duplicate). In judgment call cases, we're more likely to make a story go from #1 to #4. It's all about balance.
You may reply that "a general sense" is hopelessly subjective, and I'd have to agree, yet it's not as subjective as you might think. For example, we wouldn't downweight a substantive article about startups just because it disagreed with YC's philosophy. We might do so if it included a lot of tricks to stir up controversy. Do you see the difference? The latter is procedural; not much different than editing linkbait out of titles. If you do this job across tens of thousands of stories, and you're conscious of the obvious biases, you can mostly guard against them, especially since the same patterns come up over and over.
Not sure if I've succeeded here in getting across the gist of what we do, but I hope it helps a bit.
As for #3, that's an easy one. We don't give YC startups extra favor when it comes to boosting stories. If a Stripe API went to #1, that was because the community heavily upvoted it. We also are less likely, not more, to downweight stories when they are critical about a YC startup, such as the thread we're currently commenting in.