It IS fixable. Let's see if this the latest episode provides enough motivation for devs and Bitcoin Foundation to start even talking about BPS 1.0, so far there's been clear lack of any will for a spec, instead all the usual ("we are all just victims of the original C++ hairball dropped on us by Satoshi", or "there is too many bugs we cannot ever fix b/c some ancient client might depend on them" are brought out and reheated over and over again)
I went through the self imposed pain of implementing the protocol from scratch and learned the hard way that the behavior of the Satoshi client is not captured by any written sources other than its code.
Documenting the protocol is a major effort and it would have to be code rather than text in quite a few details to be precise, but I think it is doable and should be done.
Yes, the foundation should fund the effort.
+1. See latest emails in bitcoin-dev mailing lists (another little refuge where "big guys" and "magicians" hangout)