Lifestream, Tweets

Hmm – I blogged about the article here..  To add to that discussion, how do we separate human agency here? Here’s my attempt:

  • Human agency: choosing to put a sticky comment on a post; choosing to comment with links; choosing to upvote or downvote (even if only doing so out of “reactance“); Reddit decision to change algorithm.
  • Algorithmic agency: within the code there is a (secret) decision about what to count when calculating a post’s ‘score’; based on this score, the algorithm promotes/does not promote posts, influencing what material is read.

However,

  • Humans did not act as expected in response to encouragement to downvote genuinely unreliable material.
  • The (updated) algorithm did not act as expected in not increasing posts’ scores based on changed commenting behaviour in response to stickies. Or rather, it did not act as an external observer (Matias) expected; perhaps the update performs exactly as Reddit intended.

I’m not happy with this breakdown, however – there needs to be something about collective agency, which the algorithm seems to negate.