I'm Aza Raskin @azaaza. I make shiny things. I simplify.

I'm the Creative Lead for Firefox.

 

Identity in the Browser (Firefox)

Sponsored by

Identity will be one of the defining themes in the next five years of the Web. Nearly every site has a concept of a user account, registration, and identity. Searching for “sign in” on Google yields over 1.8 billion hits. And yet, the browser does nothing to make this experience better save for some basic auto form filling. The browser leaves websites to re-implement identity management, and forces users to learn a new scheme for every site.

Most current solutions involve lots of redirects or iframes, which leads to a confusing and phishable experience.

Besides the poor user experience, we are seeing market-moving effects of the identity/log in problem. Facebook Connect and Google’s Friend Connect both let you use your pre-existing identity and social graph to super-power other websites. The problem?

Your identity is too important to be owned by any one company.
Your friends are too important to be owned by any one company.

A Solution

The browser is your personal and trusted agent to the web. It’s the only actor on the Internet stage which both knows everything you do on the web, and never has to let that data leave the privacy of your desktop. Your browser knows you (or, at least, should).

At Mozilla Labs, we’ve been working on some potential integrations of identity directly into the browser. Note, this is an extremely rough draft. Some key points:

  • Identity is part of where you are, and what you are looking at (Amazon looks different depending on if you are signed in or not). That’s why we put it in the URL Bar.
  • For most sites, you’ll probably only have one identity, so login will be a single click or automatic.
  • Putting verbs into the navigation bar isn’t new. See Taskfox.
  • To increase visibility, webpages should be able to make a Javascript call that opens the login/signup bubble.
  • For webpages that want to own the login-process, the account creation simply acts as the ultimate form-fill.

Design Evolution

For those interested in the evolution of the idea, you can see an early mockup with comments as well as Alex Faaborg’s similiar mockups.

Next steps

Chris Messina and others has been advocating for a model which follows the Facebook Connect lead: a single verb, to connect. Once connected, you decide exactly what information to share in an asynchronous manner. Unfortunately this bleeds information — your name is known to all websites which which you connect. We’d like to explore what a connect metaphor in combination with the ability to remain anonymous but connected means.

Get involved here.

What are your thoughts? How would you expose identity in the browser?

RT @azaaza Identity in the Browser (Firefox) | Follow @azaaza on Twitter | All blog posts

View all 61 comments



Mike Beltzner

Exciting mockups, Aza and Alex. It’ll be great to see them in the context of a few scenarios. I’m especially happy with the fact that management of multiple “personas” seems to not be a primary factor in the design. I believe that to be fairly niche behaviour, and secondary to the pain point that most users have which is replicating signup and login requirements across various sites.

The technical roadmap behind the links also looks really smart. The insight that what’s making Facebook Connect so popular is that the browser can recognize a mechanism and respond to it is brilliant. I agree that the user should be in full control of that transaction, and not have to farm it out to a third party site like Facebook. This has real potential to – through disintermediation and repatriation to the browser – put the power and ownership back in users’ hands.



    Aza Raskin

    Thanks, Mike. Multiple “personas” is a weak use case. I do think there are two main modes you want to show to websites: anonymous and me. The “me” is interesting in that you may want to disclose more or less information to a website — but that is fundamentally different than having multiple personas.


About 10 days ago I started making a list of all the sites I log into. I’ve now logged 47 sites. The majority of these I log into with a username and password. This is not optimal, and I look forward to the future where the concept of identity is made stronger, where the user owns their identity, and where we can create new things based on this strengthened sense of who is who.



Kurt (supernova_00)

Very nice! I can’t wait to see where this goes because I hate logging into almost every single site that I visit.



Lennie

Is this based on the work of OpenID in Weave ? Will it be combined somehow ?



Ragavan

@Lennie – linking this with your Weave ID is one potential future that we’ve talked about. We’re still at the very early, exploratory stage of the browser playing an active role in managing your identity.


Are y’all talking to the Telepathy people? They are trying to do very similar things with the (GNOME) desktop. http://www.collabora.co.uk/projects/telepathy/



Scott W

How about something that involves client certs? I’d love to have these seldom-used gems more readily available and easy to use. It would let people manage multiple identities as they wished, etc.


Aza, It is a privilege to let in on the inner workings of your mind. This is some really amazing work.



    Yannick

    You mean, like in foaf+ssl [?http://esw.w3.org/Foaf+ssl] :p



      Yannick

      ooops what in reply to Scott W, sorry


I got another one that sums it down: Your data is too important to be owned by any one company.

Good interfaces guys, very intuitive, but still far away from a sustainable solution, which i believe is in extension to the infrastructure. I am playing with it, I’ll announce it when in alpha. Till then watching this closely!

Great stuff guys!


What about WebFinger?


Some quick notes:

-Making the user’s identity first in the location bar seems to map really well to the fact many URLs are really universal since they require the user to be logged with the same account in order to view the same content

-I really like how we are visually grouping the user’s identity and the site identity, which quickly displays an obvious representation of “you are giving your personal information to this site”

Notes on a few variations in my early mockup:

-I tried to highlight the notion of “you are giving your personal information to this site” with a visual reference of injection, the personal identity box is literally pushing into the site identity box.

-We need to be careful with how we interact with the DV/EV color scheme

-Since we’ve avoided using metallic to convey value for Site Identity (the old Firefox 2 gold bar), this is now available for conveying personal identity. This also seems to work well since we obviously want users to value their personal information (and valuable things are shiny).

-I’ve placed the personal identity and site identity grouping into the field instead of capped onto the end of the field since we are likely going to try to use “browser chrome level” items in the breadcrumbs trail for linking directly to browser chrome pages (like a bookmarks page), while the site information is content level.

-Aza’s use of language instead of icons “log in” is easier on first use, but I’m worried about it detracting from the messaging of the site identity in the long term. So I’m not really sure if iconic representation or language is best for these actions.


First off, I have to say that this is one of the most interesting identity mockups I’ve seen to date – kudos for that.

Having said that, I think we could or should, in fact, take it one step further: my identity and my friends are too important to be owned by a browser! At the end of the day, Mozilla Firefox is just another lens through which I consume the web.

Besides, I don’t consume everything through _one_ browser. In the course of a day, I touch at least 3-4 different browsers in any business setting.


Wow. Thank you Aza. Thank you so much. I have been talking about how this needs to happen for a couple of years now and it’s so wonderful to see it coming to life. It would seem that the days are numbered for login pages and a variety of share buttons online. I think that the next iteration is to solve for family computing. How to solve for multiple users without having multiple user accounts on the computer. Can a family use the same browser and benefit from personalization?
I have been waiting for the day when login would move to the browser. This will lead the way towards customization across brands and help users to carry preferences, favorites and histories across the web.
Again, thank you.


Had one question about weave. Why it does not accept open id login ? It can have login using openid and add passphrase which is required for encryption.

Secondly, About issue you mentioned for giving infomation to other site, I think open id solves this issue. ofcourse, there issue is only one trusted party we are storing lot of information. So should browser be an open id provider ? Its very unlikely.

Solution would be intresting for “No single entity stores all your information, but you still have single id”.

Probably there could be another use case where browser stores set of openIds for given user. So for each site i visit ican say which open id to use. This will minimize number of uname pwd but still won’t solve issue of having single sign on.

Can OpenId spec or extension to it can solve this issue ?
Like current extension to it allows sharing profile info once you login, It can further extend it to have ’selective information sharing’. So user can login with openid and then asked to share “some” information with given site. probably no information at all if he wants.


There are some paradoxes to consider when doing identity in the browser (or in a single-sign on system like openid.)

1) Make something easy to do, and it gets done more often. Make login be one click, and every site will have a login, whether it needs it or not. Think about the magstripe on your driver’s licence. It has good ease of use. Is it a positive feature that now anybody can read all your data in a swipe and you have no reason to say no?

2) Giving the user choice may take away their ability to negotiate. Contracts (such as the click to agree contracts done on account creation) are not really negotiated when it’s take it or leave it between unequal partners.

That said, I do have some identity in the browser proposals but they are quite different, and move away from the “login” model which tracks all you do to a more stateless “authenticated actions” model where you perform actions (like posting comments on a blog) with an identity, but still with one click.

Consider:
http://ideas.4brad.com/authenticated-actions-alternative-login

But also look at my “openid” tag http://ideas.4brad.com/tags/openid and my privacy topic for more details on this.



mk

(Summary: great looking alternative to the “auto-form-fill” currently used to solve the “signing up is painful” problem. Vulnerable to abuse, and accidental misclick personal info exposure. Doesn’t need to be there for 99.9% of sites. Needs either crypto or OpenID/Facebook Connect to solve the “single identity” problem.)

OpenID is the same thing as Facebook Connect. To use OpenID, the site that you’re logging in to has to check with the third-party OpenID provider.

There’s a difference between 1) having a tolerable way to log in to or sign up for sites, 2) using consistent registration details across sites, and 3) being able to use one single authenticated identity across multiple sites.

Yours is a beautiful solution to 1 and 2. RFC1738 specifies http://user:password@host:port/path, which is an early variant of exactly what you’re doing now. This was misguided because naive users could accidentally copy-paste their password to others: mixing your identity with the resource identity was a bad idea. Your proposal keeps them close but distinct, which is great.

I use throwaway accounts on most sites I visit. Different usernames, different passwords. This means that while in most cases I want 1 (easy login), I don’t want 2 (consistency). I don’t want to be given the option to register on some new site with 50 identities. For every identity, there is a set Q of sites that I use it on. Hide the identities where Q.size me@>Yahoo> http://yahoo.com.)



mk

(continued…)

To address 3 (verified single identity), you need either an authority which you check in with each time, or crypto. Please add crypto. Allow the user to either generate or use an existing public-private key pair for each identity. If the user clicks “(always) log on”, expose a javascript method that lets the site perform the appropriate crypto “yes it’s really me” handshake. Most sites won’t use it at first, but the few that do will love you forever.

You’re going to make a javascript function that I can call to have the login box pop up on top of the browser chrome. This will become annoying, and many sites will abuse it. If it opens up, and I misclick – well, now the site has my personal info. Oops. It will also become annoying to see “sign up” in the URL bar on every site I go to. I don’t want to sign up for or log in to 100 blogs. Find a way to minimize this. It looks great because I’m imagining that it’ll be there for when I want to sign up – I’ll be saved the usual horrors! But I’ll also see it when I don’t care at all about signing up, which is bad.

(While we’re on the subject: I use a proxy. Add this info in the bar as well, since I often am annoyed by having to turn the proxy on and off: [proxyname>me@>Yahoo> http://yahoo.com.)


Very nice mockups guys! Good job! I really want to see it in action soon so…I’ll try to involve as I can and do my best :)
I’m in!



SB

This looks brilliant, if a slightly unexpected aspect of navigation to factor into the browser chrome. I suppose it was only a matter of time. I’d like to see a separation of identity and friends network, as while the two go hand in hand with a lot of the identity providers at the moment, there’s no fundamental connection between the two, and it’d be a shame for the potential programmatic weight of friends functionality to impede the fantastic benefits of identity functionality.

Oh, and:

Putting verbs into the navigation bar isn’t new

That’s true, but can you make sure at the proofreading stage that you’re consistently using verbs and not nouns in place of verbs?

It looks like you’ve got the verb “log in” fine up there into the browser chrome, but then someone’s added a comment to design saying “Via Weave, you can login [sic] to the entire browser.” I don’t mind specific site designers being illiterate, but I’d hate to have to stare at a spelling mistake every time I open my browser.

“Login” is a noun. “Log in” is a verb, which is why we talk of “logging in” rather than “loginning”. I appreciate we have Anglo-Saxon roots but let’s not actually turn the language back into German without doing so for a good reason; say, something to do with getting free beers during Oktoberfest. That’d be a good reason.


Hi

Firefox is one best browser which can present better tools for webmasters.


There are many interesting aspects in this concept, and I think that’s were we NEED to go in order to improve the web.

Some points:
1.
Is a “login” feature really needed? I mean, once I’ve accepted to use that service I think that it should automatically link to it every time I visit it.
We should probably try to forget about the “login” idea. Yes, it’s similar to the “connect” idea.
To handle “anonymous” browsing we already have a good concept: the “stealth browsing session/window”. I think that anonymous navigation could use that UX flow.

2.
Even if we stick with “login” I think that the “Sugn Up” part should be avoided. It’s about identity: I can transmit my identity (after an initial handshake), or I could avoid it. There isn’t any registration involved: the first time I *agree* to transmit my personal data, it automatically register (and, the system could provide me hints and aids to support my navigation).
So: I will kill “Sign Up”.

3.
You’ve added a phishing warning to the “Sign Up” feature. Nice, but if you agree with me that once identified you should be always identified, I will be warned if I noticed that I’m not anymore logged in (and I’m not in “stealth” navigation).

4.
Converging again to an “Identity” solution against a “Login” solution: I think that the browser should handle my identity in a central point (where I could enable or disable it with one click).
There, I could use Profile 1, with my personal Gmail account and my personal Yahoo account, or Profile 2, with my work email and my Basecamp work account.
In synthesis: don’t think about “Logins” think about “Identities”. :)



Xa2

Great job!!
I’ve “different identities”, depending of where I am. It’s not the same to connect to a site for work purposes or for personal ones, so customising it, or having different profiles as Folleto Mallefico says would be great.



Eric

I like the idea of this, however, at least in Windows OS’s, there seems to be an overlap with CardSpace. Essentially CardSpace allows you to maintain your identity at the OS level. Granted, CardSpace isn’t being used anywhere of note, but would you try to develop some integration there?


Fantastic mock-up!

As mentioned earlier in the comments though, how does my identity transfer between browsers? Even between computers? If I go to work or school and use a computer there, how do I get my identity on that computer?

It also immediately reminded my of Chrome OS’s sign-in. You log on to your computer and then you are automatically signed in to all the Google services.


This is a manual pingback for the blog entry in which I respond to this:
http://nathanhammond.com/identity-management


Hi Raza,

I just published a detailed analysis of your plans which I like a lot: http://blogs.sun.com/bblfish/entry/identity_in_the_browser_firefox

This answers some question put above such as how this could be integrated with client certificates. It also puts forward a more generic synchronization mechanism, which I think you could integrate nicely into your work too.

There are some pieces where I am a bit critical too. But since this is an early stage project perhaps this can help spark a few ideas.



Steve

Aza, please tell me why the mac version of Firefox have changed the standard and more convenient shortcuts. I’m talking about the Command + 1 … 2 … 3

In the Safari and other browsers are used to open links from the bookmarks bar. But the biggest argument is that the way we fall into precisely this kind of behavior. Habit – a major asset in the interfaces.

You plan to revert this shortcuts back or add the possibility of change in the browser settings? Please…


This is very exciting work! I’d like this to do more, though. You should get this exact same “connect” experience however you log in with your browser; through HTML forms or with HTTP WWW-Authenticate.

This login behaviour should be standardized through an HTTP header or something similar, so it can be implemented regardless of how the credentials are submitted to the server. I love the UI and the idea of putting it into the address bar. This is definitely a huge step in the right direction regarding web authentication! Great work!


great concept!


IMHO any web authentication scheme should be based on the standard HTTP authentication mechanism: http://www.ietf.org/rfc/rfc2617.txt

I mean we’re in 2009 and still browser ask for user & pw using that cheesy modal dialog. If browsers used a more pleasant UI, just like what you are proposing now, many more sites would have used standard HTTP rather than implementing their home-brew authentication systems.


I find the project really interesting. I have an account in a lot of sites, sometimes sites I was going to use just once. Over time, we tend to change our user names and passwords and we forget to change every log in information we have, when the time comes. After a while, when you need to return to some site you don’t even remember what your user name was… Besides, we tend to use the same info in every site, creating a security issue. There’s a lot of future in this project and I’ll be sure using it!


Aza,

cool work. I agree with your premises that identity will be important, and is currently largely overlooked.

Three thoughts.

1) this is about identifying user to site. How about identifying the site to user? Should these experiences work together? Currently the site certificates and Extended Validation is confusing and, you might say, broken. CMU SOUPS did some work about this in 2008 to study certificates and EV, and people basically don’t understand it and ignore all the warnings.

2) Consider other authentication devices besides username/password. For example, some countries are issuing smartcards that can be used with a smartcard reader to authenticate with a client certificate. There are also all sorts of secureID tokens (e.g Paypal was at one point selling them to people for extra security) and who knows what else. These mockups don’t outrule any device, but wanted to mention it just to keep it in discussion.

3) what role should OS/device play? For example, my mobile knows who I am because I have billing relationship with my provider and they have my details. Should I be able to reuse this identity somehow in the apps running on the device? Safari has keychains, where they store credentials in keychains that get synced across devices if you wish to do so. In the era of netbooks and all sorts of ChromeOS-es and other browser-based things coming, is there a difference between OS/device identity and browser identity?


Identity on the internet is a tough thing. I would never want to dictate it, because there are such fine lines with digital communities. With that in mind, it would be cool to think about creating a way for the user to “own” that space a little bit. Drag and drop some of the links, features, users, etc. that matter to them. Because ultimately identity isn’t just about who I say I am, but more about who I interact with and what I do.
(Just some half-formed thoughts)


Just a note… I posted a comment some time ago with thoughts on identity in the browser and it still sits awaiting moderation, while many comments after it have been cleared. Wanted to check if you saw it.


It would be great to see you extending support information cards. A browser with a built in I-Card selector that works across all platforms will be extremely useful and enable secure identity online services linked to claims. Audit I-Cards would significantly raise the bar in the anti-phishing stakes. The US Gov and UK Gov are interested in them for online services and their OASIS and NIST have accredited them. We have I-Card desktop selectors,iphone selectors, a cloud selector but what is missing is a built in browser selector….


The anti-phishing warning is a nice touch. But is the word ‘phishing’ so well-known as to be used without further explanation? I would change the message to something like “Your personal data may be at risk” instead of “Are you being phished?”



NoneOfYourBusiness

After reading about Google’s “opt out” customized search results, I think that “identitiy management” should also include a way of having multiple identities with fully separate history, cookies etc, directly in the browser.



coco

“Time on site” ?

Have you ever think of a web browser history (CTRL+H) that is showed using charts ?

(I like also the Identifiyng idea, but that’s for tomorrow.)



dl

Have to say I had to go back and look at this today. I think your identity should be separate from the browser but your identity should be in a key just like a car. The browser should be a car in clothing form. Your id should be the a secure piece of separate technology that has to be put into the browser for you to go anywhere. Then you will be recognized by whole package.
So I think what I am saying is individual identity needs to be a new security software separate ..having said that..the browser identity build should try to be designing an “ignition” to fit various keys of identity. may seem complicated…but I think I am just saying that ignition should be flexible to need that ignition if user so chooses. does that make sense?

anyway had to return to this after thinking about it for a few weeks (as we are working on living room transmedia interoperability and this is ringing through my head with it)



DavidM

The user interface is brilliant and the concept of letting your computer take care of identity and registration, instead of the website doing that, is really relevant.

But I wonder if the browser is the proper place to manage the identity data, or that is something that the OS itself should do, so Firefox can grab my Twitter account ID from a (very protected) place within the OS, and a Twitter desktop client can sign on getting the same ID from the same place.


Hey,

Thank you for this, I will certainly look in to getting involved with this. Something very much alike of what you are talking about here is something I wanted to explore.

I will get in contact, find out some more information and post what I already have here. Will read this here again tomorrow, I am to tired now, most things will go past my mind. I even almost missed; ‘Get involved here.’


I am a little confused.

I have some things already written down. Although it’s in Dutch, I will translate it to English. Maybe combine them with what I found here.

How do I get involved? The page linked doesn’t give me a lot of guidelines. I am also one of the people that is scared to act at all. Will like to do something, only for me it’s not an easy thing.

Please, if you can, get in contact.



ben

Good idea!
Even just for computer for the whole family… We don’t change session every time we change users… and changing the identity in one click is a good solution… just in order to check mails, and gets his personal favorite



socialamigo

I was just wondering what the status of this project is at Mozilla and what you can tell us about innovations like these for 2010 – I’m currently working on a website that is aggregating social networks for subscribers to circumvent this kind of unsmart login/multiple site issue – but your ideas would be smarter across the range of sites one would visit.

check them at http://www.socialmadesimple.com


what will happen if the website generates a click event on that button? Will the website be able to log in the user at will? If you didn’t already – please make sure that only trusted events are accepted there. Unfortunately, it doesn’t quite solve the problem – since that button is put into the webpage and can be manipulated by the webpage, the webpage can make it transparent and move it under the mouse pointer, just to make sure that you will click it when you eventually click something on the page. That is somewhat concerning – in case you want to browse a site anonymously, and especially in the case that the website has been XSS’ed and tries to steal your login data.


Nice idea Aza. I discussed a similar issue a while ago with a couple of the Ubiquity guys. My angle was to look at Firefox’s security/password storage system from a ’service’ point of view, where rather than saving a password against a particular domain/URL the password would be saved against a service definition, which could contain multiple endpoints. There would be many different ways to build the ’service database’ both centralised and decentralised.

I put a few more details into a blog post (http://www.sample.org.uk/blog/post/firefox_security_overhaul)



Sheila Tombe

Sorry I don’t have any tech input to share — I’m just a dumb surfer; but this is an excellent idea. I would use such a feature on my browser the second it became available for download. Yahoo Messenger has a list of identities that can be used for instant messaging; however, the notion of one “central” bank of identities (not that I use more than a couple. . .) makes sense, adds security, and protects what little privacy we imagine we have left.


I’ve posted some thoughts here: http://blog.romeda.org/2010/04/identity.html, but the major issue I have with the Firefox Contacts / Login proposals, and have since discussing them with Michael in the fall, is that these proposals lead us further into identity silos, not away from them.

I can’t see how “identity in the browser” is supposed to work, if the identity can’t leave the browser, and the sites that tie themselves to the browser. Please, if you truly believe in the quotes above, incorporate some aspect of that into your solutions.

I won’t press on any one approach, but I’ll offer one constraint: I must be able to share who I am online with someone else on the bus in less than 30 seconds (i.e., communicating with a near-stranger in an environment unsuitable for laptops or complex operations on a phone).


tebrik ederim


Tebrikler.


Sevindim.


Hi brad;
“Just a note… I posted a comment some time ago with thoughts on identity in the browser and it still sits awaiting moderation, while many comments after it have been cleared. Wanted to check if you saw it.”

i dont understand?


Nice post. I will bookmark it.


The technical roadmap behind the links also looks really smart. The insight that what’s making Facebook Connect so popular is that the browser can recognize a mechanism and respond to it is brilliant. I agree that the user should be in full control of that transaction, and not have to farm it out to a third party site like Facebook. This has real potential to – through disintermediation and repatriation to the browser – put the power and ownership back in users’ hands.


v pills ile büyüme olsun. ereksiyon ve erken boşalma gibi sorunlardan da ömür boyu kurtulun. etkisini ilk kullanımdan itibaren görebileceğiniz v-pills sizlerin cinsel yaşamına büyük bir katkı sağlayacak.


Good article. Thanks.


Selülit kremi ile selülit tedavisi olmanıza gerek kalmadan bu sorundan kurtulabilirsiniz.


Leave a Comment