These task platforms are really starting to feel more and more like clocking in at work: check-ins, reposts, interactions, screenshots—and finally they give you a “score.” Plainly speaking, it’s just screening witches. The problem is that many project teams don’t bother to explain the scoring logic clearly either; in the white paper they brush it off with a single line like “anti-witch model,” and even I find it exhausting.

What’s even funnier is that recently they’ve been hyping AI Agents and automated trading, saying it frees up your hands—yet in reality, on-chain interactions become even more frequent, and it starts to look more like a production line. On the security side, though, nobody goes into details: who gets the permissions, whether the contract can be paused, who’s responsible if something goes wrong… In any case, the moment I see the words “automated,” I immediately start to be suspicious.

My biggest fear isn’t losing money—it’s working away for half a month and then getting wiped out by a single sentence like “suspected witch,” with no way to appeal. Right now I’d rather do fewer tasks and first翻 clear the rules and the contracts; even if it means it’ll be slower, so be it.