Thread

So, Vertex uses pagerank which is a probability, which implies it's a number between 0 and 1. Pagerank is distributed as a power law, which means that between an average pagerank and a high pagerank there are 4 or more order of magnitude difference. Jack dorsey has 1000-10000x the pagerank of a normal person. Power laws are hard to comprehend for the human mind, which is why Profilestr tried different formulas to normalize this score to be between 0 to 100, in ways that make sense. These formulas were not great, so I suggested they use something like this. It's poorly written but I don't have much time to work out the details.
Gist
How to display Pagerank by Vertex
How to display Pagerank by Vertex. GitHub Gist: instantly share code, notes, and snippets.
 TLDR; Vertex didn't change anything, the ranks are computed as before, they are simply presented differently to the end users. This whole conversation is making me more adamant in my belief that ranks should not be displayed to end users.

Replies (4)

It's hard to understand indeed. People were asking this because Trust Score some weeks ago was high for almost all users and suddenly it dropped hard, I even read that some people had 0, which raised alarms and people wanted an explanation. I didn't understand the maths, but I appreciate the transparency in this matter.
🛡️
These kinds of scores create wrong incentives, it reminds me of how it was done with a service called Klout in the early web 2.0 days. People will do strange things to get better scores, but it’s pointless as every score should be contextual. It should be clear that these scores have nothing to do with reach either (as many assume it is).