I agree that “For You” used to have a much better signal to noise ratio.

I agree that “For You” used to have a much better signal to noise ratio. I’m tempted to blame the increased size of Medium’s user base — after all, with a small user base it’s a lot easier to tweak the algorithm for good results, a small sample set is a good enough proxy for recommendation quality, and few people are going to be trying to game the system, while all of those things grow worse at at least a geometric rate as the number of users grows.

That said, I’m seeing a particular kind of bad post pop up much more frequently on Medium in the past ~2 years than previously: the short post that should be much shorter. I see a lot of medium posts that are two paragraphs long, making a point that would be clearer as a tweet, along with a couple huge and irrelevant images and somebody’s affiliate link.

I used to see these mostly as HN links (since I browse HN’s “newest”), and chalked it up to the culture of shallow self-promotion that’s all too common on HN (combined with the fact that, like Blogger a decade ago, Medium provides people who have no technical skills with a free blog that looks relatively professional). Because I rarely interact with such posts other than viewing them, I doubt that my feed is full of them due to this early exposure; instead, I suspect that this behavior has become normalized — it’s now expected to post “one minute reads” that are actually ten second reads, and my darling “twenty minute reads” and “eighty minute reads” are seen as bad for business, creating a low view to read ratio — nevermind the fact that there aren’t any ads on Medium (other than native ones) so impression metrics don’t really matter.

Another possible source is following people who have lower standards for recommendations than I do. Such people may typically write nice, long, well-considered posts, but yet recommend all the crap they agree with, whether or not it’s worth reading. Medium doesn’t distinguish between following somebody for their posts and following them for their recommendations.