You can take a look at PubSub which allows you to broadcast data (whether it be a file, a new IPNS record, a message,…) to peers who are interested in it. But yes, if noone downloads your file, it will only be on your computer, but I don’t see it as a shortcoming. We don’t want to make peers download and seed files without their consent.
It depends what you mean by search. If you search a website by knowing already what you have to type in the URL bar, then IPFS already does that with mutable links like IPNS or DNSLink. But if you search a website using only keywords, then it becomes complicated. Either you wait that a centralized search engine like Google starts to index IPFS websites, or you develop your own decentralized search engine, like YaCy, or the search engines used for IPFS that crawl the DHT to index new content. But I don’t think a distributed search engine will ever be as efficient as a centralized one, because the latter use gigantic databases called indexes, and with IPFS you would have to download this database in its entirety before using it.