Thread by @benadida

1/ On the occasion of the launch of @apple's "Sign in with Apple," allow me to indulge in a walk down a memory lane called @MozillaPersona -- the project I loved and led at Mozilla 6-8 years ago, the project that broke my heart. This is my take, I'm sure it's incomplete.

2/ I joined Mozilla (Labs) in 2011, and I immediately clicked with @thunder, @michaelrhanson, and Lloyd who were working on the verified email protocol. The idea was simple and powerful: signing into web sites often boils down to typing and then verifying an email address.

3/ so why not make that a one-or-two-click operation rather than a rigamarole of typing your email, going to your inbox, clicking the verify link, etc. The browser could mediate some proof of email ownership. We could make it super privacy protecting. It was called BrowserID.

4/ Given my background and, shall we say, somewhat impassioned interest, I became the Director of that effort. We built an amazing team, one of the best I've ever worked with. BrowserID was branded Persona.

5/ as a web site, you would make one JavaScript API call, get a blob of data, hit a Mozilla API endpoint with that blob, and boom, get a verified email address.

No consumer registration or API keys or ID tokens. Easier than anything else on the market even today, 7 years later.

6/ As a user, you'd click a "sign in w/ Persona" button, select amongst your email addresses, and go. If you had never used an email address with Persona before, we'd do the email verification dance once and then you could use that email address everywhere (that supports Persona)

7/ It was a magical experience for both web site & user. As long as both web site & user were set up with Persona. Because remember, identity is a transaction between a user and a service provider, where interests are not necessarily aligned.

8/ Anyways, we even had a prototype pseudonym feature where Mozilla would give you a pseudonymous email address to protect your privacy at web sites where you didn't want to share your real email! It was grand.

9/ Only, past the first couple thousand web sites, adoption was slow. And support internally at Mozilla was far from uniform. This led to deep and painful discussions about how we should grow.

10/ Some wanted Persona to be built deeply into Firefox, so that "Firefox would be the best Persona experience." I saw little value in this. It didn't really change the user experience to be in the browser vs in a JavaScript polyfill library.

11/ We wasted an inordinate amount of time on this in-browser vs. Polyfill issue, and ultimately I failed to convince leadership of my view. This was the proximate cause of Persona's demise. We didn't have leadership support because I failed to make the vision compelling enough.

12/ to this day, many Mozillians will tell you that our mistake was not building natively for Firefox. I think that's even in the official Mozilla postmortem. With all due respect to my Mozillian friends, I think that's ludicrous.

13/ But whatever, because that wasn't the only problem.

14/ We needed identity provider adoption. See, Persona was built to be federated: each domain would vouch for its users, e.g. adida.net vouches for [e-mail]. Mozilla was bootstrapping provisioning only as an intermediate step, because two-sided networks.

15/ we all knew this was smoke & mirrors until one big email provider came on board and directly vouched for its users using Persona. We should have been more aggressive about that. We should have used the Mozilla muscle to make it happen. We didn't. We opted for neat hacks.

16/ one of my proudest ideas was bridging to Google and Yahoo via OAuth, so that Google and Yahoo users would have close to the ideal experience, never doing an email verification link dance. It was lovely. We even have some great initial strategy conversations with Google.

17/ but that wasn't quite enough to take off. It was a cute hack that didn't indicate to the world at large that Persona was the right bandwagon.

18/ another problem: we assumed everyone wanted a privacy-first identity solution. We were wrong. Web sites wanted -- especially at the time -- your name, your friends, your calendar, etc. We had only an email address. Identity is a transaction, and we were offering too little.

19/ and then there was mobile. We worked out butts off to make it work well on mobile web (and IE 6!!), but native app support was pretty tricky at the time. (With today's native APIs ... Oh we could do so much better.)

20/ so as a web site / app, you could implement Persona, but it would only solve 1/2 your problem. And now the remainder of the problem (native) was harder, because how do you let users log in if they're accustomed to a Persona button on the web, and it doesn't show up on native?

21/ we should have done way better on native. Way better. Oh my goodness did we screw the pooch on that one.

22/ late in the 2nd year of Persona, we vaguely started to understand that we couldn't settle for "now you have 10% fewer passwords," which is unfortunately how it looked. We needed to be either the complete password killer for web sites / apps, or for users.

23/ We started thinking about native app support. We started building a password manager. But by then it was getting too late. Our executive support was running out. And we hadn't gotten a win in a while. A couple executive departures and reorgs sealed the project's fate.

24/ my conclusion about identity systems: you can only make one happen if you have a lot of weight to throw around like FB or Google with a lot of users' data. Or if you've got a unique offering like @Clever that brings sanity to a fragmented K-12 classroom environment.

25/ or, if you're Apple, and you can just tell developers what they can or cannot do on the most important computing platform ever *and* you've got a metric ton of users, then yeah you can probably make a privacy-first identity system happen.

[Cheers to my Persona teammates.]