I was thinking about this and imagined the federated servers handling the index db, search algorithms, and search requests, but instead leverage each users browser/compute to do the actual web crawling/scraping/indexing; the server simply performing CRUD operations on the processed data from clients to index db. This approach would target the core reason why search engines fail (cost of scraping and processing billions of sites), reduce the costs to host a search server, and spread the expense across the user base.

It also may have the added benefit of hindering surveillance capitalism due to a sea of junk queries from every client, especially if it were making crawler requests from the same browser (obviously needs to be isolated from the users own data, extensions, queries, etc). The federated servers would also probably need to operate as lighthouses that orchestrate the domains and IP ranges to crawl, and efficiently distribute the workload to client machines.