Perhaps a way forward is instead much greater personalization?
Yes, reddit, HN, and the NYT, are echo chambers. But is that the problem, or the eternal-Septemeber, clueless reporting, low-signal high-noise level of the echos?
Is the problem that the LA Times publishes a propaganda piece on Copyright Office staffing? Or that it's so very hard to find high-quality commentary? With which, if you cared to, you could to discredit the dreck? Is the enormous role that groupthink plays in Washington the problem, or that even when people wish to break out of it on some topic, the information infrastructure doesn't exist to support them?
Imaging an HN which, instead of pretending its quality hadn't declined over the years, was instead trying to technically address the decline. What might be done?
Well, why couldn't a high-quality discussion community exist embedded in a reddit? A necessary condition is "we don't see the dreck". Personalization. Everyone has control of what the site looks like to them. So you can be fine with reddit-style puns on topic A, but only see insightful expertise on topic B. And I can make the opposite choice. On Tuesdays.
A core idea of machine-learning-based human-computer-hybrid systems, is that the system learns about the people at the same time that the people help the system execute the process. The candy machine bribing CS undergraduates to grade CS 101 exams, learns both about the answers being graded, and about how much to trust different people's grading of different subjects.
And the system can recognize when it needs advice. "This clearly goes in bucket M, and this in bucket N, but hmm... I'm not sure where to put this one. Hey, statistically well-executed sample of people, which bucket?".
Perhaps it's overly optimistic to think that people might choose to not live in mind ghettos if given the choice. But we won't have a choice until we get better at cultivating quality. Instead of chance "Oh, look over there, by the petunias! A thoughtful thread! Oh, no, it's dead now".
Yes, reddit, HN, and the NYT, are echo chambers. But is that the problem, or the eternal-Septemeber, clueless reporting, low-signal high-noise level of the echos?
Is the problem that the LA Times publishes a propaganda piece on Copyright Office staffing? Or that it's so very hard to find high-quality commentary? With which, if you cared to, you could to discredit the dreck? Is the enormous role that groupthink plays in Washington the problem, or that even when people wish to break out of it on some topic, the information infrastructure doesn't exist to support them?
Imaging an HN which, instead of pretending its quality hadn't declined over the years, was instead trying to technically address the decline. What might be done?
Well, why couldn't a high-quality discussion community exist embedded in a reddit? A necessary condition is "we don't see the dreck". Personalization. Everyone has control of what the site looks like to them. So you can be fine with reddit-style puns on topic A, but only see insightful expertise on topic B. And I can make the opposite choice. On Tuesdays.
A core idea of machine-learning-based human-computer-hybrid systems, is that the system learns about the people at the same time that the people help the system execute the process. The candy machine bribing CS undergraduates to grade CS 101 exams, learns both about the answers being graded, and about how much to trust different people's grading of different subjects.
And the system can recognize when it needs advice. "This clearly goes in bucket M, and this in bucket N, but hmm... I'm not sure where to put this one. Hey, statistically well-executed sample of people, which bucket?".
Perhaps it's overly optimistic to think that people might choose to not live in mind ghettos if given the choice. But we won't have a choice until we get better at cultivating quality. Instead of chance "Oh, look over there, by the petunias! A thoughtful thread! Oh, no, it's dead now".