I've heard about RethinkDB for the first time in 2013[1] and at the time it was described as "MongoDB with joins and auto-sharding".
I also heard that one of the initial "killer-features" of RethinkDB was supposed to be a stored history, just like a big persistent data structure (in my understanding, similar to Datomic). In the video[2] a lot of time was spent describing the benefits of an append-only structure.
I don't think it was ever supposed to be a history-saving data structure -- that log-structured stuff is (I haven't watched the video, I'm assuming) about talking nice to SSD's. It was at one point, after that video, a single-machine memcache-compatible persistent key/value store, before the pivot to being clustered and having its own query API.
What exactly what their previous goal? I thought they've always been for real-time search, at least that was the moniker when I looked at it 1 year previous.
I guess is because this guys have been jumping around to see what sticks, their focus nows seems to be the "the database for the realtime web".
Why should companies bother if rethinkdb doesn't show to have clear goals yet? tomorrow they can be the "database for the (whatever)".