How To Search Facebook Photos?

Is Facebook's display of "This search isn't available yet-until then, use simple phrases to search for people or photos." a form of Minimum Viable Product (MVP)?

• As an information-retrieval "noob" (illiterate), it would seem that the difficulty of parsing natural queries would be harder than the difficulty of creating the database backend to support searches of "my friend's posts about topic X". But it appears that Facebook is able to parse natural queries, yet has chosen not to rollout the capability to return these search results.

• Answer:

  Post search has been announced as part of Graph Search (http://newsroom.fb.com/News/728/Graph-Search-Now-Includes-Posts-and-Status-Updates). It will roll out slowly to users that already have Graph Search. This wasn't part of the original launch because, contrary to the assertion made in the subtext, natural language parsing is in fact easier than indexing the whole of Facebook's status updates with privacy context and returning results to you in a punctual manner. Natural language parsing is mostly a syntactical rule set that remains fairly static. Indexing a continually growing set of data like all the status updates on Facebook is a moving target that requires continual expansion. The data set of all status updates is absolutely huge. Larger and more complex than what most search engines need to deal with. And it needs to scale up to the millions of people using Facebook at any given time. The amount of infrastructure required to support post search is many times larger than the amount required to provide the initial object search capabilities. I've been trying to find public numbers for illustration, all I can really come up with is this old post (http://searchengineland.com/by-the-numbers-twitter-vs-facebook-vs-google-buzz-36709) from 2010 that references our old statistics page. It says 60 million status updates per day. (I'm making up all the numbers below) FB has grown a lot since then, but let's just say 2x growth. So 120 million status updates per day. Let's say on average each update is 500 bytes. That's 600GB per day of new content to index. That may not seem like a lot in absolute terms, but that's a rough example of how much the index could grow by every single day. Keeping the index fast and able to scale up in terms of both data set and search volume requires a lot of forethought in architecture and significant capital investment. It's not like you can buy a bunch of 3TB drives from Fry's and call it a day to store the data. The data has to be loaded in memory in order to return timely results. The data has to be replicated several times to provide redundancy and scale horizontally. I just want to get across the point that this is one of Facebook's largest technological feats to date. I'm sure there will be a https://www.facebook.com/Engineering post on it in the future from the engineers that built it that will have plenty of the ooey, gooey technical details.