I fully understand the issue that you are bringing up. By the design of the IRC servers we’ve never guaranteed that +o, -o, JOIN, nor PART arrive in real-time. Since the chatters API endpoint is based off of the same (potentially stale) data, it too can have issues. Hence, I am proposing bot authors use IRCv3 to validate if a user issuing a command is a mod for sake of authentication.
As @trumpi27 brought up, this doesn’t help in the case of trying to find out if the bot itself is modded. The only two solutions I’d know for this case is to either always assume you aren’t modded, or to have two connections open and read your own messages to see if the server considered you a mod or not. I understand these solutions aren’t great, but they’re the only foolproof methods of dealing with “Am I a mod?”.
Finally, in regards to the “why did this suddenly get worse?”, our chat team is aware of the issue and is working on it. I’m afraid I don’t have more information for you at this time.