Well…cheating is something that demands control. And when speaking about virtual gaming the “path to cheating” becomes easier than in real life. So…, with all due respect to all, but “rely on honesty” is a a very honest position ))), but contradicts slightly with the reality and the so-called “social-mass behavior”.
I do think that there should be some mechanism, automated one. Also, maybe something close to search-engines algorithms with the automated analysis of users past behavior vs present.
In addition, it would be really great to have ONE system and not some third party website where the users should go to check their results, although…even with the third party tool it could be good in case if the integration is made inside Zwift directly.
Bottom line: it would be great to have a smart-type of system categorization. And I do not think that average w/kg categorization could be consider as a “smart system”.
I would take more data than just average w/kg (lets say in past 30 days). Maybe also last 10 races positions vs types of rides in those races. Could be wrong, but I think that Zwiftpower already has in place some type of parameter with the “race difficulty” - since different races have different difficulties even if those races are in same category. But not sure what is included in the “difficulty parameter”, is it just riders, or maybe the distance + elevation also?..