Couchbase 2.0 is Membase Server (http://www.couchbase.com/products-and-services/membase-serve...) merged with CouchDB, and as I understand it, Couchbase 2.0 more closely resembles Membase Server at the moment (http://www.couchbase.org/get/couchbase/2.0.0).

I agree. This is getting overly complicated.

We're certainly hoping to reduce the complication by removing the boundary between (clustered) BigCouch and Apache CouchDB by making the clustering an optional and transparent part of the Apache CouchDB project.

Its not that complex. CouchDB is the core technology that Couchbase and BigCouch, et. al are using to power their more "vertical" offerings, in various degrees. CouchDB is not going anywhere, and its a great standalone product.

Yeah, this makes me want to learn MongoDB instead.

I view CouchDB and BigCouch to be very similar since the client APIs and libraries are the same, so both have the same developer experience. Couchbase uses different client libraries and doesn't seem to have the language coverage that CouchDB and BigCouch do.

All three are open source and you can build them yourself.