4 ms·

I miss SO. I spent a good amount of time on there as a 15 - 17 year old thinking the stats would mean something to a future employer. I remember revering Jon S

I miss SO. I spent a good amount of time on there as a 15 - 17 year old thinking the stats would mean something to a future employer. I remember revering Jon Skeet. I would post questions that I couldn't crack and I'd usually get a quick reply with someone explaining kindly and in-depth about the issue. Barely into college was when they started auto-closing and marking as duplicates along with mods being little snotnoses. Big miss. I genuinely think volunteer / unelected mods are the downfall of any forum. On my backlog of ideas, I have "democratic Reddit"... doubtful I'll ever do it but if anyone wants to: there you go (the concept being mods are elected by and can be removed from their roles by a vote of active posters) I am not sure HN's mod situation but my understanding is that it is essentially just two people and they are paid employees of YC.

blackeyeblitzar 2y ago
I would like to learn more about your ideas for solving the mod problem. Although I do think it’s unrealistic to overcome the network effects of existing huge platforms. When a site like Reddit has tens of millions of users, I personally think it should be regulated like a utility. They should not be able to just ban users arbitrarily, when they’re just operating a communication service effectively, and one protected from competition by network effects.
mncharity 2y ago
> On my backlog of ideas, I have "democratic Reddit" Hybrid computer-human social computing might permit that and much else. Consider a CS break-room candy machine which bribes undergraduates to grade CS 101 exam answers (eg [1]). It learns the characteristics, the strengths and weaknesses, of the graders, the gradees, the answers, and the questions. And can use highly targeted redundant cross-checks to address uncertainty. Eg "graders are giving this answer very different partial credit, so I'll show it to someone who has a history of doing partial credit well". An intelligently-applied small-integer redundancy of novice graders can thus emulate scarcer expert graders. And processes can be crafted from a mix of tooling and AI and humans - "this answer was left blank, and this one is clearly right, but this one has higher uncertainty, so involve wetware here". So one could do "democratic" tracking of poster culture. Or elected mods. Or mods with rich AI tooling. Or sampled direct democracy. Or eternal-September filtering. Or anchor to a subculture defined by individuals or policy. Or... whatever. Or perhaps even allow colocated participants to live in their own little flexibly-specified worlds, with preferences on topics and people and thread characteristics - my netflix isn't yours, so why are we eyeballing similar threads on this page? Entangling location with personal interests seems so... "You are in a library and interested in topic X. So leave the library and drive across town to the library". At least this all seemed a rich space of possibilities... more than decade ago. Someone else will have to comment on why that seeming opportunity seems so little mined. [1] https://people.eecs.berkeley.edu/~bjoern/papers/heimerl-umati-chi2012.pdf https://people.eecs.berkeley.edu/~bjoern/papers/heimerl-umat...

blackeyeblitzar 2y ago

I would like to learn more about your ideas for solving the mod problem. Although I do think it’s unrealistic to overcome the network effects of existing huge platforms. When a site like Reddit has tens of millions of users, I personally think it should be regulated like a utility. They should not be able to just ban users arbitrarily, when they’re just operating a communication service effectively, and one protected from competition by network effects.

mncharity 2y ago

> On my backlog of ideas, I have "democratic Reddit" Hybrid computer-human social computing might permit that and much else. Consider a CS break-room candy machine which bribes undergraduates to grade CS 101 exam answers (eg [1]). It learns the characteristics, the strengths and weaknesses, of the graders, the gradees, the answers, and the questions. And can use highly targeted redundant cross-checks to address uncertainty. Eg "graders are giving this answer very different partial credit, so I'll show it to someone who has a history of doing partial credit well". An intelligently-applied small-integer redundancy of novice graders can thus emulate scarcer expert graders. And processes can be crafted from a mix of tooling and AI and humans - "this answer was left blank, and this one is clearly right, but this one has higher uncertainty, so involve wetware here". So one could do "democratic" tracking of poster culture. Or elected mods. Or mods with rich AI tooling. Or sampled direct democracy. Or eternal-September filtering. Or anchor to a subculture defined by individuals or policy. Or... whatever. Or perhaps even allow colocated participants to live in their own little flexibly-specified worlds, with preferences on topics and people and thread characteristics - my netflix isn't yours, so why are we eyeballing similar threads on this page? Entangling location with personal interests seems so... "You are in a library and interested in topic X. So leave the library and drive across town to the library". At least this all seemed a rich space of possibilities... more than decade ago. Someone else will have to comment on why that seeming opportunity seems so little mined. [1] https://people.eecs.berkeley.edu/~bjoern/papers/heimerl-umati-chi2012.pdf https://people.eecs.berkeley.edu/~bjoern/papers/heimerl-umat...