David, good article and glad to see you reference our discussion on Wikipedia (including my discussion of the copyright database). I'm the guy that's spearheaded a lot of the research and verifiability issues (along with Xenon54 who has been very interested in getting to the bottom of this). For my background, I work in the industry as a writer and programmer. I'm also a professional industry historian and archiver, as well as a member of the IGDA's game preservation SIG. So I take my research very seriously.
There's actually two parts to the documented research at Wikipedia, which includes a section that happened on the video game project's discussion page as well.
As far as the Oct. 18th date, it's simply impossible given what we know. As for the current Nintendo's repetition of it, as I reported over at Wikipedia after a lengthy networking session through reps of different divisions of NOA, they're simply getting their info from an internal database. Nobody knows when it was created, or who gave the information for it.
Regarding Kent's followup statement on his blog that you stumbled on, that was prompted by me as well (my name over there is wgungfu) and was also a direct response to the discussion that was going on at Wikipedia.
Regarding your guess of October 31st, there's a problem with going with that as the avaliability date. What the copyright database states is when it was first made available for public consumption. This can be off-press date, your shipping date, or your actual publication date (i.e. available for purchase). We have no way of knowing which was intended and normally in video games you want to get the copyright done as early in the process as possible. So I'm inclined to believe Oct. 31st is the off-press date (same with the slightly earlier date for the box). Additionally it's still just the date given for the manual not the entire product. As shown by the earlier box publication date and the discussion about the chip manufacturing (42nd week of the year 1985 would be the final week of October), it simply shows a timeline of when the reources for the finished product were being produced. Not their assembly (as a single finished product), not their shipping and distribution, and not the selling of the finished product. In the end, we're still left with the only form of evidence for public availability (i.e. sales) that doesn't require any sort of interpretation being the Macy's launch.
Regardless, there is no contradiction between Tilden's statement of "shortly after", Kent's statement in his book, and the facts. When "the launch" is being stated, that's referring to the actual official launch date "event", October 18th, when they first began hitting up retailers in the New York area - which it was not available for. That in no way states it wasn't available during the entire test marketing process there, which is clearly shown by the Macy's ad.