Prospective users, however, will weight the cost of learning/remembering a new thing, and changing the way they browse, versus the benefit of "now it's slightly easier to recall things that I've browsed before."
Imagine you saw some crazy weird looking mouse at the store. If you bought it, it'd feel weird at first, but it's pretty much guaranteed to be an awesomer mouse. However, your current mouse works fine. Would you buy this new mouse?
Probably not. It's easier to not switch, even if it would be beneficial to do so.
In terms of improving your conversation rate: If you are requiring people to create an account to try historious, you should immediately get rid of that. Let them try it with as low barrier to entry as possible, and hopefully get them hooked.
Regarding your switching point, I'd definitely switch (and, so would anyone who uses vi/emacs (maybe I should market to them)), but I realise I'm not my target market, and that your point applies to that very much indeed. This just means that I have to make every aspect of historious easy to get into, though (fully-functional demo, easy account creation, extensions for all browsers), as I'll never quit :P
On an unrelated note, I just took a look at moreofit, and I'm very surprised by how accurate it is, so props for that!
I don't see any reason to require a user to create an account to start using Historious. Just create a unique hash (user id) in the bookmarklet, and have that hash be sent along with any "historify" requests.
Perhaps once 30 days are nearing, or they are filling up their free account, you could remind them that they should upgrade to pro, or at at least create an account.
Hope that helps.