I led the BrowserID/Persona effort at Mozilla from its first days first building, then as product lead until shortly after Persona launched. Plenty of people contributed as much as I did, and more on the technical side; but I was in the room for just about every protocol and product decision, so I know what we got right and where it failed. What we got right, I'd still defend today: email as identity, and shipping the scaffolding ourselves it worked with any email, in any browser, from day one, and the only people we had to convince were websites. Why it died is a longer story: timing and priorities inside Mozilla at least as much as anything in the market, a too-late bridge to the big mail providers, and, yes, passwords that were just barely tolerable so an open protocol stayed optional. The idea remained with me though, of what could have been.
A decade of other work later, agents changed the terms. You cannot hand an agent your password and call it fine; credentials that merely annoyed humans become indefensible at machine speed. The identity question is open again, and the answer will harden into infrastructure that lasts decades. browserid-ng is my second run at it: I picked up where Persona left off, rebuilt it in Rust, extended it to enable you to give identities to and control agents acting on your behalf.
Check it out at browserid.me.