I had that graphed and then pulled it out because I think it needs to be it's own post and I need to pull some data from a different source.

An important characteristic of all the articles I was looking at is that they all had been accepted for curation. It's basically true across Medium that an article doesn't get amplified across the network unless a human editor accepts it. Medium has an internal team that does this and then also has some partner publications that can do it, like ours.

So, in order to say shorter is better you also have to take into account curation rates. One and two minute articles performed well on a per-effort basis, but I'm nearly positive that they are rarely curated and so factoring that in, they don't perform well at all.