Two common implementations of engagement farming are a) very contrarian/misinformed opinions to bait people to respond negatively and b) simple Q&A prompts like "What opinion about AI would get people mad at you?" which encourage people to respond. The organic discussion (positive or negative) is generally favored algorithmically so it gets boosted, and in the case of Twitter, there is a monetary incentive to do so, as Twitter Blue accounts can earn revenue from tweet views.
It used to be called "trolling," before trolling was redefined to "publicly disagreeing with important people." It's maximizing the ratio between the characters you type and the characters typed in response to you.
Is an ideal response one so thoroughly perfect that nobody has anything to say after it? Or is leaving some room for continuation good until it's excessively so?
The tweet author performed the former very recently and caused a news cycle by complaining about MKBHD's Humane Pin review with a bizarre take: https://daringfireball.net/linked/2024/04/16/vassallo