The bootstrap webclient https://github.com/yacy/yacy_webclient_bootstrap
now points by default to http://search.yacy.net
using the solr endpoint. It can be hosted as github pages and thats the case with that repository. It can be tested at http://yacy.github.io/yacy_webclient_bootstrap/
This shows how fast a YaCy search page could deliver search results, if no p2p activity is done. I believe we should consider this method as default in the future and abandon the old YaCy search page to move on to a modern architecture of web pages using a JSON service and a standard search engine backend (here: Solr. elasticsearch may come someday...).
The question is: how can we still integrate a p2p search? Doing p2p and collecting the results would make it necessary to push their results into the index before retrieving the combined result. I did not test that yet but I doubt that this is fast.
An alternative would be, to search in the p2p network _after_ the result has been delivered - just to enrich the search index further for the next search. But that would work only for large search portals with many users.