We intend to replace the privately hosted federation search with a collection of partial indexes hosted on wiki farms with primary indexing of its own sites and collaborative indexing of sites known to its members.
Scrapes will focus on changed pages reflected in sitemaps within the federation but will also watch for reverted and deleted pages based on other signals internal to a farm.
Scrapes will catalog many properties such as types, tags, ids, links and assets. Word search will somehow cooperate with or delegate to our existing full-text search.