Revising in Public
This revision has been superseded. Read the current one.
An argument is rarely right the first time it is written down, and the usual remedy — quietly editing the page until it says something else — hides the part a reader might learn the most from.
So this piece keeps its drafts. This is the first one, and it is wrong in at least one interesting way: it assumes the reader has already agreed that the revision is worth showing at all.
The claim
Publishing a revision is cheap. Publishing the record of a revision is what costs something, because it commits the writer to having been wrong in public.