What do you mean statistical settling? Fargo was used for many years before Shane's 824 in 2015. I think once you get 1000+ robustness and over 2MM data points, the rating would be fairly accurate. I could be missing something though.
Separately, given the Fargo creep, having 800 rated players in the 700s might give us over 1000 rated players now! That would be a hoot! Who is gonna be the first 1,000!!!
I don't think ratings have compressed. I think they have actually expanded, no?
-td
So Shane may have had 1,000+ games at the time, but in the early days there weren’t nearly as many pros or players in the system with high robustness, and many were still building from starter ratings. The issue isn’t Shane’s robustness, it’s the robustness and connectivity of the player network around him.
For example, you had players like Albin sitting in the high 700s who were primarily competing against other players who were also in the high 700s at the time, but many of those same players would later end up as 800+ Fargo players. If everyone in that group is initially clustered together, there is only so much room for separation until more results, more connections, and more reporting resolve where everyone actually sits relative to the broader player population.
My point is that Shane can only be rated relative to his competition. If his competition isn’t yet accurately positioned relative to the rest of the Fargo universe because the network is still maturing, that affects Shane’s rating as well.
And yes, ratings have expanded. What I meant by compression was mostly brainstorming how variables unique to the 70s and 80s might affect the distribution of ratings.
