What is pouets logic about the #rank in the alltime-top?
category: general [glöplog]
It's not number of views (Pouët doesn't divulge that directly), it's viewer-rank. Focused is the 10699th most viewed prod on the site. Aquaverse is the 5655th most viewed. Debris is the most viewed (the 1st). So lower is better.
One thing that surprised me a bit is how small the gap between the popularity of top-10 and top-11 is. I had assumed there would be a huge bonus of always being visible from the front page (and thus a cliff between top 10 and the rest), but external sources seems to count for a bunch. E.g., MySQL Raytracer is 20th in viewer-rank despite not being very visible at all, probably boosted from non-scene sources like Twitter or Hacker News, and gets into 800th-ish overall mainly because of that.
One thing that surprised me a bit is how small the gap between the popularity of top-10 and top-11 is. I had assumed there would be a huge bonus of always being visible from the front page (and thus a cliff between top 10 and the rest), but external sources seems to count for a bunch. E.g., MySQL Raytracer is 20th in viewer-rank despite not being very visible at all, probably boosted from non-scene sources like Twitter or Hacker News, and gets into 800th-ish overall mainly because of that.
But what is the viewer-rank exactly?
The prod with the most page views (highest popularity) gets viewer-rank 1. The prod with the secondmost page views (second highest popularity) gets viewer-rank 2. And so on.
Ah ... is this something that is visualised with the popularity-bar? So that debris has a popularity of 100% and the other prods have a lower value in relation to their viewer-rank?
Yes, popularity follows directly from number of views (I double-checked in the source). If prod A has higher popularity than prod B, it also has more views. On the web, I think you don't get any decimals, but in the API, you get full precision.