the protocol bit of the URL
There is no "protocol bit of the URL". There's a scheme portion of the URL1, but it doesn't necessarily identify the concrete protocol used on the wire. It names the abstract protocol - the language, semantics, and requirements - that the user agent and server use for their control and data flows. That can become something else under the covers.
the 'http://' bit is a pattern for humans, not machines
That's not really correct either. It's for both. The user agent uses it to decide what kinds of request messages it will send to the server, and how to interpret the server's responses.
This happens with standard HTTP/1.1 as well. A user agent talking to a server for the first time doesn't know if that server supports HTTP/1.1, so it has to send a request that's compatible with HTTP/1.0 as well. Once each side knows the peer supports HTTP/1.1, they can use message formats that aren't supported in 1.0, such as the chunked transfer encoding for message bodies.
1Some would say "of the URI", but consensus on the W3C and IETF URI mailing lists seems to be moving toward dropping that acronym.