I don’t know that I would go so far as to call them subject headings…the experiment is still too new to know how this will all shake out. However, these tagmashes have a great deal of potential, when enough have been generated, to provide a far more useful search tool than (in my opinion) LCSH does. Tagmashing is a chance to combine a lot of what makes tagging great (user-friendly, current language, allowing for multiple perspectives, users decide what is meaningful/interesting) with what makes subject searching great (potential for less ambiguity, combining frequently searched/related terms).
How it works:
When users search for two or more tags, LibraryThing creates a persistent URL for that “tagmash” (for example, this page of the tagmash for ‘France’ and ‘WWII’). This is a bit different from Flickr’s clusters, which are (at least from what I understand) fully automated, as these are the aggregated results of user searches.
Why it’s not not a silver bullet (yet):
One of the commenters on Weinberger’s blog post made an excellent point: at the moment, at least, this process has the potential to even increase ambiguity, depending on the subject: “philosophy, history” can be about the philosophy of history (historiography, etc.), “older” philosphy (Plato’s ‘The Republic’), or a history of the discipline of philosophy.
How I think this would (ideally) impact the archival world:
This is an especially interesting development to me, as my first thought after reading about tagmashing was of Elizabeth Yakel (and co-authors’) D-Lib article “Creating The Next Generation of Archival Finding Aids” was how wonderful it would be if the next-generation finding aids that Yakel discusses could incorporate this feature.* Having created EAD finding aids, I found myself wondering about what other possibilities these types of finding aids had for users, especially in terms of subject headings (the finding aids I created used LCSH) that, as far as I could tell, were not often a good fit for very regionally or subject-specific collections, especially those with creators who were not well-known.
One problem I can see right off the bat is that these types of collections tend to be (relative to the types of materials on LibraryThing) low-usage, and so the statistically less useful user-generated tags (“my aunt Alice and her brother”) would not necessarily be knocked out of place by the potentially more-useful tags/tagmashes.
The way I choose to believe that problem would be solved, should archival tagmashing actually arise, is that one would just market the heck out of your collection and make it the “it” place to be online.
*Yakel points out multiple other possibilities for online finding aids in the article…which is fantastic and thought-provoking, and the Polar Bear Digital Collections project, on which the article is based, are fascinating. Read the article and visit the collections now, if not sooner.