I used to be an ORM-hater, but my experience with Django has changed my mind, somewhat. I still think there are projects where ORM is unnecessary or even harmful, but for some projects, being able to lean on an ORM to create simple queries/updates or to handle DB migrations is a big time saver. And you can always fall back to hand-written SQL when you need to as long as the ORM allows it, which it absolutely should.
And it’s written in Java. Even though I’m not a huge fan of Java, it’s almost refreshing to see a new project claiming high performance that isn’t written in Rust or Go.
If there was an easy answer, someone would have implemented it already. Obviously, it’s a challenging problem, and I don’t claim to have the solution.
I think expanding the voting dimensions (a la Slashdot) would make it easier to create an algorithm, but it pushes complexity to the user, so that’s a tradeoff.
But, even with up/down votes, I think there are potential ways of identifying users whose votes deserve more weight. For instance, someone who up-votes both sides of an argument chain (because both sides are making good-faith responses and adding to the conversation) should be boosted.
We need the karma-equivalent of PageRank. Every vote should not be treated the same, just as Google doesn’t weight every link equally. The “one user one vote” system is the equivalent of pre-Google search engines that would rank pages by how many times they contained the search term. But it can’t be as simple as “votes from higher-karma users are worth more” because the easiest way to build insane karma is to build a bot or spam low-effort replies to every rising post. Still, the system needs to be able to extract the wisdom of the crowd from the stupidity of the crowd, and the only way to do that is to apply a weighting gradient to users and their votes.
I suspect there is at least one engineer who voiced concerns months or years ago, was not listened to, and is now having an “I told you so” moment.