I'm still thinking about my "community" search engine, xobaque and wondering how to stick to some self-imposed design criteria (no crawling, no logins) and still make it more useful.
For example, would it make sense to provide a list of search buddies and forward queries to those other sites, merging the results? Scoring is probably going to be bonkers because each instance scores the articles independently and then we mix it all up. Does a search result exchange format exist? Or would it make more sense to merge at the HTML level? Looking at Wikipedia's federated search page I see even more challenges listed (pagination, timeouts).
More fundamentally, perhaps, is the notion that each community might already have a focus (all the blogs on Planet Emacslife focus on Emacs) and therefore what would be the benefit of also searching indieblog.page? Of course, people blog about Emacs without being listed on Planet Emacslife, so in this case, there's a potential benefit. But what if we search Planet Emacslife (about Emacs) and the RPG Planet (about role-playing games). Oh, I know a nerd or two who's into both, but really, is there a benefit to be had in this situation? That seems to point to an asymmetrical setup:
Planet Emacslife includes indieblog.page;
RPG Planet includes indieblog.page;
indieblog.page includes both Planet Emacslife and RPG Planet;
but crucially Planet Emacslife does not include RPG Planet;
and RPG Planet does not include Planet Emacslife.
🤔
Any other ideas? I'd prefer not to replicate the search indexes, to be honest.
31M for Planet Emacslife
770M for RPG Planet
901M for indieblog.page
#search #xobaque