I've said as much as to one extra incentive (besides retrain cost) as to why openai has frozen the training period for post-2022. I think it's trying to generat

I've said as much as to one extra incentive (besides retrain cost) as to why openai has frozen the training period for post-2022. I think it's trying to generate as much data as possible before itself has contaminated its training set. They'll effectively have a monopoly over it; it's interestingly a rare example of "we can only do this once, then it's forever degraded". You really want a clean & discrete start.