Yeah, if you want to do something impactful but have no idea if you’re going to be a net positive or a net awful, perhaps it would be wise to consider not doing that?

The thesis is more that we don't actually know how to "do good" because the world is complex (nth order effects, unintended consequences, etc). So the intention to do good isn't enough. (Of course, that doesn't mean we should purposefully have bad intentions.)

Instead, let's look at impact as an indirect outcome of a ton of experimentation, evolutionary pressure and a complex system. Then the way to have impact is to experiment. So focus on doing that instead of getting stuck in figuring out how to "do good".