[abathur]

Maybe I'm misreading this comment or the context (because I think the biggest single piece we're currently lacking in platform moderation is maximal transparency into what/who gets moderated and how)...

but how would this "solve" spam in any case where the ratio of spam-to-signal crosses whatever threshold it takes to start breaking a platform's network effects, reducing engagement, driving out existing users, warding off new ones, shifting how the community views what the platform is for, etc.?

[repolfx]

Spam didn't break email's network effects. Yeah, it hurts it a bit, but there were lots of attempts to build competing email networks that didn't go anywhere.

This is partly because email isn't "for" anything except communication. The idea that platforms must stand for something is new and wrong. A good platform stands for nothing and is open to everyone.

[abathur]

This might just come down to why I'm not sure I'm interpreting you right.

It's one thing to have the platform equivalent of a spam folder for new top-level posts that smell like junk. But these platforms have more significant design challenges: to cleanly handle replies, retweets + commentary, mentions, comments, threads, and anything else that is sort of inherently contextual (including the possibility that there are legitimate non-spam posts that interact with a spam post to quote, comment, reply, warn others, and so on).

I'm a little skeptical about how well anyone can meet that design in a way that makes it easy to see what was flagged as spam and isn't also sensitive to the ratio of spam to legitimate posts...

Its possible you're imagining that spam posts don't show up at all in a thread unless you hit a single toggle that re-renders the thread with any pruned replies and branches in place. Interfaces like this don't spiral out if the ratio changes, but I also don't think they make it easy to spot the flagged posts.

An interface that marks the posts in-thread can make it easier, but they're sensitive to that ratio. Very sensitive if they shows the full post, and a little sensitive if they replace the post with a clickable indicator that there's suspected spam there.

[repolfx]

Yes, for sure any implementation is very much about the design subtleties. I did work once on an email spam filter so I have a pretty good idea of the complexities involved in that space, which includes things like some people's spam being other people's ham, people replying to spams, problematic user interfaces and so on.

My point is a more general one: people argue for censorship as the only way to maintain a workable forum, but, I don't believe that, based on my prior experience. I'm not saying it's easy to build a really great spam filter for social forums, but Slashdot proved out a lot of good techniques, and anyway censoring stuff en-masse just creates a different set of problems: social rather than technical.