Inferring Tweet Quality from Retweets

evanmiller.org

33 points·sebst··8 comments

tweet's quality (percent of readers who will retweet it)

That seems a very odd way to define "quality", and the phrase that most comes to mind is "quantity is not quality". Perhaps virality would be a better word for this metric.

It doesn't strike me as odd. Retweeting is something you do when you like a tweet for one reason or another. What seems odd is invoking "quantity is not quality" for retweets since that is a second order phenomenon. It would seem more appropriate to apply it to people who tweet a lot, not people who are retweeted a lot.

It sounds like a feature Twitter might be interested in - Especially given all the recent talk about users being 'overwhelmed' with too many irrelevant tweets.

Having a tweet score based on those formulae could allow twitter to filter out less interesting tweets from users' feeds. It would be a nice optional setting to have for those users who have too many followers.

I did some similar analysis on gauging the quality of a user from their retweet behaviour [1]

[1] http://bit.ly/Pzea7o

it'd be interesting to get more metadata for Nth level retweets as another measure of quality

That essentially reduces to PageRank, replacing links with tweets.

Yes it does. But the Nth level metadata is not publicly available today. I'm not even sure its available in the Twitter API.

Second-order retweets as a measure of retweet quality has been a thing since as far back as 2011.

Reply on news.ycombinator.com