I understand your argument, but I think this is perfection being the enemy of good. As I mention in the article, it's not necessary to monitor every account or every lie. Most content on Twitter for example, is retweets from the most popular accounts. Remove THOSE lies, and you remove a significant amount of the lies on the platform.

Of course, that leaves the valid question of what qualifies as a lie. But this question is also answerable, if not perfectly. Trump's election fraud lies were being flagged by Twitter, just in a wholly ineffective way which did nothing to stop them being spread. Twitter eventually demonstrated that they coud impede this spread by blocking the retweet function on his flagged tweets.

The idea that you could remove all misinformation from a social network is, as you say, unachievable (I think you'd end up with a lot fewer than a million accounts able to post😅). But not all misiniformation is created equal. The goal should be to target the most pernicious, lies when they are spread by the most influential accounts. I think that alone would have an enormous impact.