Not true. A person can play tens of thousands of games and not have a single sandbagging violation simply by not sandbagging. The mistake you make here is in thinking that a violation is equivalent to a single game that looks like sandbagging. People are cited for rank manipulation sandbagging based on a large pattern of behavior. (Alt sandbagging is a little different but also judged based on a pattern of behavior.)
Ridiculous. A game either is or it isn’t a sandbagging (read rank manipulation) game. How it comes to the attention of the mods is irrelevant. For example, a mod might independently notice that a game ended suspiciously, decide to investigate, and discover the sandbagging pattern. The triggering game was no less a sandbag game just because it wasn’t reported. As I already discussed in my previous post, it can be very difficult to tell at times, so great care must be exercised, but that is why a large pattern of behavior is used and not just a few games.
Correct. The rank-manipulating sandbagger isn’t being banned for the one you reported. He is being banned for a bunch of recent rank manipulation games that came after a previous warning also based on a bunch of rank manipulation games. Your report merely called attention to the situation, albeit inadvertently. Most people I hope (and expect) would be delighted to know that their report led to the banning of a sandbagger. Certainly I would be.
All the sentences in this paragraph are incorrect, except the first. Perhaps language is the barrier (I am a stunted English-only speaker, and do admire anyone who speaks another language), so let’s see if I can rephrase the example to make it more explicit. A 9k player (real strength) is a rank manipulation sandbagger. He has driven his account down to 14k by deliberately resigning (or timing out of) won games. He is on the road to winning because he is playing normally (at a 9k level), but his opponent surprises him by resigning sooner than expected, spoiling his (the sandbagger’s) plan to resign just before the scoring phase. This is normally a set-back for the sandbagger. But your button comes to the rescue. KataGo says that there was a way for his opponent to win, even though neither player saw it because it would require the skill of a 1d player. Nevertheless, the button is now available, and the sandbagger uses it to annul and thereby eliminate the set-back to his plan. He has lost a little time, but he is pleased that he didn’t rank up.
Sandbaggers rarely play badly to rank down in my opinion. In the suspicious instances I have seen, it would take the discernment of a high dan to tell with any confidence. That is a poor basis for an accusation, and why bother when innumerable sandbaggers use less ambiguous techniques. Playing badly goes entirely against the psychology of sandbagging. Sandbaggers are insecure people who want an easy win to feel good about themselves and/or they are incipient sadists who find ecstasy in humiliating others. Consequently, they take a rank manipulation game to a winning stage or to a won stage (where they resign instead of pass), because in that way they demonstrate their “superiority.” If they don’t have time to do that, and just want to rank down, they resign from a non-game (2 or 3 moves).
Now that sounds like an excellent idea. There may be some danger that it would swamp the mods, but perhaps it could be refined.
All of the categories of annulment that I mentioned involve cheating. The characteristic that makes rank manipulation different is that an innocent party has a bogus win annulled. As I mentioned, bogus wins are universally condemned as illegitimate in sports and games.
Except for, perhaps, the recent annulment mistake already mentioned, I know of no case in which a game was annulled because KataGo said it was the wrong result. As far as I know, KataGo only draws attention to suspicious results that need investigation (if, for example, one looks though a game history), or it confirms the conclusion already reached based on other evidence (which includes the large pattern of behavior mentioned above).