One of the problems with "karma" sites like these is that everyone sees the same metric. But the comments you love aren't likely to be the same ones I love.
I'd like to see a system closer to a movie/music suggestion system. Personalize the crap out of everything, show me the stuff I'll probably love, and don't even bother showing me trash I'll obviously hate. And for God's sake, make it trivially easy to permanently killfile other users.
Yes, this is computationally more involved. But the result would be something that would make everyone happier (except perhaps inveterate trolls).
make it trivially easy to permanently killfile other users
My wish list would also include being able to filter based on URLs and words in the article's title (using regular expressions). If I had that, I don't think I'd even need a fancy personalized recommendation system.
I actually like knowing that what I see is the same as what other people see. Among other problems, it's hard for a site to cultivate a consistent culture if everyone has their own filter bubbles in it. I don't mean to rail against customizability completely, but I see what's basically "just let everyone make their own filter bubbles" brought up way too often as a one-size-fits-all solution for culture problems on websites with the downsides almost never brought up.
There's been a few times where I was a heavy user of a small filter-bubble of a site, recommended the site to others, and then later realized most people associated the site with certain kinds of content or idealogical positions (that I had avoided in my unique view of the site). That can be uncomfortable.
As a much more minor and kind of silly case: there was a forum I was a very long-time user of, and it had an old-school visual design that I somewhat associated with the community. I expected that other users would easily recognize screenshots of it or even things with a similar color scheme. I wondered if the old-school design of it influenced the community and had considered writing something about that if I wrote about site design. Then one day I visited the forum while logged out. It had a much more modern and conventional visual design. It turned out that a new theme had been made for the site many years ago and made default only for new users which now made up the bulk of the community. I felt uncomfortably solipsistic to learn that after so much time.
Valid concerns, to be sure. In my case, I'm trying hard to get away from "outrage"-type posts, and it would be nice to have a better way to filter these out. Currently, I pretty much just drop sites entirely once this sort of thing starts.
show me the stuff I'll probably love, and don't even bother showing me trash I'll obviously hate
I think this is what I use Hacker News for. I want to escape all of the weird suggestion tech that I've only ever seen be done poorly by the most successful companies applying it (Youtube, Facebook, Netflix). Hacker News (and hckrnews.com) let me see an unfiltered list of everything ~10 people gave enough of a shit about to show up on the front page.
I'd rather solve the signal to noise problem by reducing the noise, than by reducing the signal and hoping I'm only getting signal.
(2009)
Original discussion, which may be of interest:
https://news.ycombinator.com/item?id=495053
Wow, originally published 10 years ago.
Ouch, please add 2009 to the title...
One of the problems with "karma" sites like these is that everyone sees the same metric. But the comments you love aren't likely to be the same ones I love.
I'd like to see a system closer to a movie/music suggestion system. Personalize the crap out of everything, show me the stuff I'll probably love, and don't even bother showing me trash I'll obviously hate. And for God's sake, make it trivially easy to permanently killfile other users.
Yes, this is computationally more involved. But the result would be something that would make everyone happier (except perhaps inveterate trolls).
My wish list would also include being able to filter based on URLs and words in the article's title (using regular expressions). If I had that, I don't think I'd even need a fancy personalized recommendation system.
I actually like knowing that what I see is the same as what other people see. Among other problems, it's hard for a site to cultivate a consistent culture if everyone has their own filter bubbles in it. I don't mean to rail against customizability completely, but I see what's basically "just let everyone make their own filter bubbles" brought up way too often as a one-size-fits-all solution for culture problems on websites with the downsides almost never brought up.
There's been a few times where I was a heavy user of a small filter-bubble of a site, recommended the site to others, and then later realized most people associated the site with certain kinds of content or idealogical positions (that I had avoided in my unique view of the site). That can be uncomfortable.
As a much more minor and kind of silly case: there was a forum I was a very long-time user of, and it had an old-school visual design that I somewhat associated with the community. I expected that other users would easily recognize screenshots of it or even things with a similar color scheme. I wondered if the old-school design of it influenced the community and had considered writing something about that if I wrote about site design. Then one day I visited the forum while logged out. It had a much more modern and conventional visual design. It turned out that a new theme had been made for the site many years ago and made default only for new users which now made up the bulk of the community. I felt uncomfortably solipsistic to learn that after so much time.
Valid concerns, to be sure. In my case, I'm trying hard to get away from "outrage"-type posts, and it would be nice to have a better way to filter these out. Currently, I pretty much just drop sites entirely once this sort of thing starts.
I think this is what I use Hacker News for. I want to escape all of the weird suggestion tech that I've only ever seen be done poorly by the most successful companies applying it (Youtube, Facebook, Netflix). Hacker News (and hckrnews.com) let me see an unfiltered list of everything ~10 people gave enough of a shit about to show up on the front page.
I'd rather solve the signal to noise problem by reducing the noise, than by reducing the signal and hoping I'm only getting signal.