4 ms·

That is an issue, but the real fundamental issue is algorithmic timelines and recommendations that bias content to maximize engagement. What maximizes engagemen

by api 6y ago

That is an issue, but the real fundamental issue is algorithmic timelines and recommendations that bias content to maximize engagement. What maximizes engagement? Hate, fear, and bombast.

im3w1l 6y ago
https://en.wikipedia.org/wiki/List_of_most-liked_tweets https://en.wikipedia.org/wiki/List_of_most-liked_tweets https://en.wikipedia.org/wiki/List_of_most-retweeted_tweets https://en.wikipedia.org/wiki/List_of_most-retweeted_tweets Doesn't seem like you are right about that. Edit: Going by replies might yield a different more negative result, the word "ratiod" imply that many see a reply almost as a downvote. But I was not able to find a list of most replied-to tweets.
- true_religion 6y ago
  According to that list, K-pop maximizes engagement so Twitter should pivot to being MTV online.
- hshshs2 6y ago
  replies are only similar to downvotes in the absence of likes, that’s why it’s “ratiod” and not “replied”
- ben_w 6y ago
  Interesting data — and heartwarming to see! — but I’m not sure it’s the right measure. A thousand tweets that are each seen by a different (non-intersecting) group of a thousand people, each tweet being an accusation of evil directed at a specific person or group, does as much damage as one tweet seen by a million. I hope that this measure would show the same result as the two lists you linked to; my current guess is that if someone did this analysis, the result would be almost as positive, but not quite, on the grounds that hate-inducing tweets are against the TOS and will be hidden or deleted when they reach a visibility threshold.
  - AnimalMuppet 6y ago
    A thousand tweets, each seen by a different, overlapping group of a thousand people, reaches fewer total people, but to those people it looks like multiple independent sources confirming the same information. That is, the information looks more trustworthy. (That was one of the points of the Soviet "active measures" propaganda during the Cold War. They not only had western sources giving the Soviet line, they had multiple sources giving the same line.)
    - ben_w 6y ago
      Thanks for the insight, I thought that might be the case but didn’t know. (I also felt it would be an Apples-to-Oranges comparison when the alternative was a single account tweeting to a million followers).
- dahfizz 6y ago
  I don't think looking at the 30 most liked / retweeted tweets is a good measure for deciding what kind of content is most engaging. BTS makes up more than half of both of those lists - that's just a list of popular twitter accounts. The most tweeted about people on twitter are Trump, Biden, and George Floyd[1]. I think that gives a pretty good idea of the content occupying people's minds. [1] https://blog.twitter.com/en_us/topics/insights/2020/spending-2020-together-on-twitter.html https://blog.twitter.com/en_us/topics/insights/2020/spending...
w_t_payne 6y ago
This is the fundamental design flaw of the system. Fixing it isn't going to be easy, though. Look at the amount of effort that the scientific community tries to put into improving the quality of communication:- peer review, standards for content ... and there's still nagging issues with reproducing results and even outright fraud in some cases. Whatever solution we come up with will likely increase friction in the user experience, but there's lots of room for innovation here.
- api 6y ago
  Fixing this isn't a matter of adding complexity but of removing it. When timelines were sequential or user-moderated and YouTube was just a search engine and host for videos it was fine. Algorithmic prioritization algorithms were introduced around the 2008-2012 time frame and all Internet discourse went to hell shortly afterwords. What we did was to insert a biasing function between all people that biased content for engagement. What maximizes engagement? Hate, fear, and conflict.
- AnimalMuppet 6y ago
  The fundamental flaw is human nature. That's not going to be easy to fix.

by api 6y ago

That is an issue, but the real fundamental issue is algorithmic timelines and recommendations that bias content to maximize engagement. What maximizes engagement? Hate, fear, and bombast.

im3w1l 6y ago

https://en.wikipedia.org/wiki/List_of_most-liked_tweets https://en.wikipedia.org/wiki/List_of_most-liked_tweets https://en.wikipedia.org/wiki/List_of_most-retweeted_tweets https://en.wikipedia.org/wiki/List_of_most-retweeted_tweets Doesn't seem like you are right about that. Edit: Going by replies might yield a different more negative result, the word "ratiod" imply that many see a reply almost as a downvote. But I was not able to find a list of most replied-to tweets.

true_religion 6y ago
According to that list, K-pop maximizes engagement so Twitter should pivot to being MTV online.
hshshs2 6y ago
replies are only similar to downvotes in the absence of likes, that’s why it’s “ratiod” and not “replied”
ben_w 6y ago
Interesting data — and heartwarming to see! — but I’m not sure it’s the right measure. A thousand tweets that are each seen by a different (non-intersecting) group of a thousand people, each tweet being an accusation of evil directed at a specific person or group, does as much damage as one tweet seen by a million. I hope that this measure would show the same result as the two lists you linked to; my current guess is that if someone did this analysis, the result would be almost as positive, but not quite, on the grounds that hate-inducing tweets are against the TOS and will be hidden or deleted when they reach a visibility threshold.
- AnimalMuppet 6y ago
  A thousand tweets, each seen by a different, overlapping group of a thousand people, reaches fewer total people, but to those people it looks like multiple independent sources confirming the same information. That is, the information looks more trustworthy. (That was one of the points of the Soviet "active measures" propaganda during the Cold War. They not only had western sources giving the Soviet line, they had multiple sources giving the same line.)
  - ben_w 6y ago
    Thanks for the insight, I thought that might be the case but didn’t know. (I also felt it would be an Apples-to-Oranges comparison when the alternative was a single account tweeting to a million followers).
dahfizz 6y ago
I don't think looking at the 30 most liked / retweeted tweets is a good measure for deciding what kind of content is most engaging. BTS makes up more than half of both of those lists - that's just a list of popular twitter accounts. The most tweeted about people on twitter are Trump, Biden, and George Floyd[1]. I think that gives a pretty good idea of the content occupying people's minds. [1] https://blog.twitter.com/en_us/topics/insights/2020/spending-2020-together-on-twitter.html https://blog.twitter.com/en_us/topics/insights/2020/spending...

w_t_payne 6y ago

This is the fundamental design flaw of the system. Fixing it isn't going to be easy, though. Look at the amount of effort that the scientific community tries to put into improving the quality of communication:- peer review, standards for content ... and there's still nagging issues with reproducing results and even outright fraud in some cases. Whatever solution we come up with will likely increase friction in the user experience, but there's lots of room for innovation here.

api 6y ago
Fixing this isn't a matter of adding complexity but of removing it. When timelines were sequential or user-moderated and YouTube was just a search engine and host for videos it was fine. Algorithmic prioritization algorithms were introduced around the 2008-2012 time frame and all Internet discourse went to hell shortly afterwords. What we did was to insert a biasing function between all people that biased content for engagement. What maximizes engagement? Hate, fear, and conflict.
AnimalMuppet 6y ago
The fundamental flaw is human nature. That's not going to be easy to fix.