It’s a nice idea, but I don’t see how it works in practice. A policy banning someone until they sign an oath saying they’ve read and understood it? Should they pass a quiz?
I’m skeptical that moderation does very much to teach people how to have a discussion. It teaches people that they will be reprimanded for certain behaviors on certain forums, and if it’s applied consistently that can be enough. But the audience isn’t captive enough, and the stakes aren’t high enough, for moderator intervention to have the effect of changing someone’s style of discussion. At the very least, that would require a good faith effort on the part of the poster, and the people who come here to make a good faith effort to post the best philosophy they can don’t really need moderator help to improve their posting; most of what makes for good discussion is just good faith.
Most of moderation’s effect comes from excluding the small percentage of people who cause most of the disruption, but that too is moving the ocean with a spoon given how easy it is to circumvent any exclusion – and that problem also hits any attempt to elicit some promise in exchange for lifting a ban: anything more onerous than creating a new account can be ignored at a net gain.
And either way, the bottleneck is time. We have a handful of part-time volunteer moderators, and there’s more posted here than can be read in its entirety. We can bring on more moderators, but that brings its own challenges (over-moderation also gets complaints), and in any case we’ve had trouble finding people interested in moderating. This all makes me skeptical of anything that costs moderator time but is unlikely to make a meaningful difference.
And so: we return to ignoring. The time cost is distributed, and paid most by those who would request the most intervention. It avoids issues of over-moderation, but directly addresses the existence of bozos. It works best paired with moderation, because there are efficiencies in centralizing the exclusion of those who everyone agrees are a source of disruption, but it also works in the absence or delay of moderation.
If the alternative suggestion is “change all the bozos into not-bozos”, ignoring the bozos strikes me as a pretty good option.