Independent Identity
Good name in man and woman, dear my lord,
Is the immediate jewel of their souls:
Who steals my purse steals trash; ’tis something, nothing;
’Twas mine, ’tis his, and has been slave to thousands;
But he that filches from me my good name
Robs me of that which not enriches him
And makes me poor indeed. --Othello, by Wm. Shakespeare
This column isn't just about identity. It's about the need to free the good names of countless customers whose souls' jewels are held captive in vendor silos. These are not silos of proprietary technology, so familiar to us here in the free software and open source communities. These are silos defined by CRM (customer relationship management) and transaction processing systems at all the large companies we do business with every day.
I've written about identity before, in Grounds for Identity (issue date) and What's Your i-Name (issue date). In the second piece I outlined a number of market scenarios that
Let's say you want to rent a car after you arrive at an airport. You grab your luggage off the carousel, and you head over to the wall of booths: Hertz, Avis, Alamo/National, Budget, Thrifty and so on. Now let's say you plan to go skiing, so you want a 4-wheel-drive car. You have a couple of club memberships: say, Avis Wizard and Budget FastLane; but you're also not sure it matters. What are your choices?
If you have lots of time, you can go through a series of lines (all slow, usually), and hope for the best. If you're short on time, you can go to the shortest line, and hope for the best. If you're pushy, you might run from one booth to the next, interrupting agents and saying, "Hey, got any 4-wheel drives?"
Whatever your choice, it's among silos. Every car rental agency is a closed habitat. The market category is open only in the sense that you're free to choose among competing habitats.
Ah, but how about on the Web? That's an open marketplace, right?
No, not yet. Ten years after Netscape hit 1.0, most big vendors (including all the car rental agencies, all the hotel chains, and most of the big retailers) still see the Web as a place to put up silos for the customers they've "acquired", "captured", "own", and wish to "retain" as if they were possessions. Slaves, basically. You see the bottom ends of their silos when you look at the credit and membership cards in your wallet, or when you talk to the"customer relationship" professionals on the phone. Remember where the card came from. It may have your name on it, but it's not yours. It's theirs.
Sure, you can go from one website to another -- from Hertz to Avis to Thrifty to Budget -- and see what kinds of deals you can get. In every case, your preferences, credentials and choices as a customer are isolated inside the agency's silo.
Airline websites appear to aggregate the silos of their car rental agency "partners" by putting all the agencies on one page. At United Airlines' site, for example, there's a tabbed subsite for renting cars from any of the airline's seven car rental partners. But there's also a hidden hitch: you can only interact with one agency at a time -- just like you do at the airport. If you click on another agency's link, it replaces the content in the open agency window. There are no right-click workarounds. All those are disabled by United.
Of course there are the travel sites: Orbitz, Travelocity, Expedia and so on. But these are silos too, even if they let you search across the silos of muliple hotel chains, airlines and car rental agencies. What Orbitz knows about your travel history and preferences isn't transferrable to Travelocity or Expedia. Your reputation with each of them is not yours. It's theirs.
What we have here is a personal problem: we aren't able to present our credentials and preferences to multiple vendors, on our terms, so they can compete for our business in an open marketplace. Our power as customers is severely limited, so long as our relationships with vendors are on their terms and not ours, in their habitats and not in the free world. Our market power will also remain deficient as long as data about ourselves -- our identities, our transaction histories, our choices and preferences -- belongs only to vendors.
This problem isn't one vendors can solve, any more than the lack of an Internet was a problem network makers could solve, or the lack of an open operating system was a problem computer makers could solve. This is a problem that can only be solved where it's best perceived: on the customers' side of the marketplace. And it can only be solved by independent developers -- especially those in the free and open source communities, writing code and making sure implementations are truly open and interoperable. It cannot be solved from the top down. It has to be solved from the bottom up. That is, from the individual to the institution.
The field we call "digital identity" has mostly been an institutional one, so far -- even though leaders in that field have believed from the beginning that an identity system grounded in personal control and autonomy is essential. This has certainly been the case with Andre Durand and Phil Becker, two of the leading figures in the category. Andre is the founder of Ping Identity Corp. (on the advisory board of which I serve). Phil Becker is the founder and publisher Digital ID World magazine, and also of the Digital ID World conference (DIDW). It was Andre who abstracted digital identity into three tiers (1. Assumed, or Personal; 2. Assigned, or Corporate; and 3. Abstracted, or Marketing), and placed the primary one at the center, with the individual. It was Phil who said that we would each be "managing by identity" -- meaning, "identity is what we use to manage what we do with computers".
Both Andre and Phil have said on many occasions that the fist moves would happen at Tier 2, at the corporate level. And they were right. Every DIDW conference, starting in 2002, has been preoccupied with largely corporate issues: identification and authentication, trusted computing and rights management, RFID implementation, PKI, standards and compliance, combating fraud, and so on. But no buzzword has been bigger than "federation". In an IT Conversations interview, Eric Norlin, Director of Marketing at Ping Identity Corp., said federation "leaves the distributed environment as it is, but seeks to let the end user link together those pieces and still have control over their privacy, what gets shared, and how." For example, the Liberty Alliance spec, "is purposefully designed to be opaque. (It) tries to accomplish one thing... to allow one end user to link one account to another account. But in order to protect the privacy of the end user... neither side of that account -- either Company A or Company B --would know who the end user is that is being passed between them. In other words, in order to find that it was me that was linking between Companies A and B, they would actually have to go outside of the specification through a backchannel. So the specification puposefully makes it hard to violate someone's privacy."
I've summarized this as "large companies having safe sex using customer data". This is less fair than funny, but it accurately characterizes federation as a big company concern. While federation might protect customers and provide them with certain benefits, it is essentially a silo-to-silo "solution." So far.
Liberty Alliance isn't the only consortium devoted to federation. The other big one is WS-I (where WS stands for Web Services). WS-I was founded by Microsoft, IBM and Verisign, partly in response to the Liberty Alliance, which was initiated by Sun, partly in response to Microsoft's Passport, which was part of a scary-sounding centralized identity initiative called "Hailstorm". Liberty gathered pretty much everybody not in the WS-I -- though many companies belong to both. For most of DIDW's history, the Liberty Alliance has had the higher profile.
In DIDW/2005, held this past April in San Francisco, Liberty's profile dropped almost to zero. I'm not sure why, or if it matters, beyond the degree to which the grass roots movement -- which I'll call Independent Identity -- adopts standards gleaned from the WS-I library. This appears likely, since it was in the WS library that Kim Cameron, chief Identity Architect at Microsoft, found what he needed to structure what he calls the "identity metasystem."
This is a system that, by definition, is not owned by Microsoft. The "meta" prefix in this case refers to inclusion. Above or around would be better synonymous prepositions than about (which is the approximate meaning of meta in metadata). While it may seem odd that Microsoft would suggest the need for a market solution they don't own, it's quite sensible to expect such a solution to come from Kim Cameron.
Kim came to Microsoft by way of acquisition, when Microsoft bought Zoomit, a Toronto company specializing in the "metadirectory" field. I came to know Kim through Craig Burton of The Burton Group, back in the early 1990s. Craig and his colleagues at The Burton Group coined the term "metadirectory", and called for vendors to fill the need for one. Kim Cameron led development of a metadirectory product, along with a lot of deep thinking about the subject, at Zoomit. It was during that time that I became good friends with Kim, an an occasional consultant to his company.
As it happened, the Zoomit acquisition came right after it was clear that Microsoft was failing miserably with its Passport initiative. That failure gave Kim room to begin working on an approach to identity that could be architected at the "meta" level, and good for the whole marketplace -- rather than just for Microsoft.
Starting in October 2004, Kim began publishing his Identity Weblog (at identityblog.com), gradually posting each of his Seven Laws of Identity. He did this on the installment plan, to give everybody time to talk about each one before moving on to the next. His Seventh Law was published in late March, 2005.
The Laws had two obvious puproses. The first was to guide product development for everybody. The second (and one Kim avoided bringing up, for obvious reaons), was to make sure his employer didn't do something that would hurt the marketplace.
None of the points I've made so far are new. In fact, I've made them in my closing keynotes at every Digital Identity World (DIDW) since the first one in 2002. My most recent DIDW keynote was in May, 2005, and I said it again there too.
But this time there was a difference. This time we had signs of progress. Big ones. Interested individuals and organizations were now meeting, publishing, working on products and protocols, finding and implementing useful standards, and adjusting work already in progress. An informal group had gathered around a shared belief that the world needed a new class of identity services that carried such adjectives as "grass roots", "lightweight", "user-centric", "next-generation", "independent", "human origin", "bottom-up", "distributed" and "individual".
The first Independent Identity meetings took place at the DIDW/2004, which was held last October in Denver.
At an Identity Commons BoF (Birds of a Feather) session, Drummond Reed of Cordance (and a winner of DIDW's first award, in 2002, for the XRI/XDI standard) suggested the need for CoRM, or "Company Relationship Managment": an ability by customers to manage relationships with multiple companies, and to support the safe and selective sharing of identity-related data with those companies for mutually agreeable purposes. This, we agreed, would be far better for the marketplace than the silo'd marketing machinery in every vendor's Customer Relationship Management (CRM) system.
We also agreed that CoRM-enabled customers in open markets would be a Good Thing for vendors as well. Customer relationships would be two way, not one-way. And they would provide vendors with much more useful data than silo'd CRM systems would ever allow -- much as the open Internet is better for everybody's business than the closed LANs and online services we had back in the 80s and early 90s. Many rich, interesting and useful customer relationships would be possible. Better products and services would follow. Vendor differentiation and specialization would increase, as resourceful sources of supply began to follow independent forms of demand.
In addition to Identity Commons, several other parties joined a series of subsequent conversations at DIDW, a In part this was a reaction to heavy conversation around
So instead I'll
To some degree all these adjectives were equal-and-opposite reactions to domination of DIDW's stage by large technology companies and preoccupation with identity problems shared by the customers of those companies. Issues included the usual: trust, privacy, authentication, single-sign-on, PKI, standards and compliance, combating fraud, distributed directories, RFID, and managing all the above. But the biggest buzzword at the conference was "federation". In an IT Conversations interview, Eric Norlin of Ping Identity Corp. (disclosure: I'm on the company's advisory board) called federation a "concept" or an "approach" that "leaves the distributed environment as it is, but seeks to let the end user link together those pieces and still have control over their privacy, what gets shared, and how. The Liberty Alliance spec, he says, "is purposefully designed to be opaque. (It) tries to accomplish one thing... to allow one end user to link one account to another account. But in order to protect the privacy of the end user... neither side of that account -- either Company A or Company B --would know who the end user is that is being passed between them. In other words, in oeder to find that it was me that was linking between Companies A and B, they would actually have to go outside of the specification through a backchannel. So the specification puposefully makes it hard to violate someone's privacy."
I've summarized this as "large companies having safe sex using customer data". This is less fair than funny, but it accurately characterizes federation as a big company concern. While federation might protect customers and provide them with certain benefits, it is essentially a silo-to-silo "solution." So far.
Liberty Alliance isn't the only consortium devoted to federation. The other big one is WS-I (where WS stands for Web Services). WS-I was founded by Microsoft, IBM and Verisign. Liberty Alliance was initiated by Sun, partly in response to Microsoft's Passport, which was part of a scary-sounding initiative called "Hailstorm". Liberty gathered pretty much everybody not in the WS-I -- though many companies belong to both. For most of DIDW's history, the Liberty Alliance has had the higher profile.
In DIDW/2005, held this past April in San Francisco, Liberty's profile dropped almost to zero. I'm not sure why, or if it matters, beyond the degree to which the grass roots movement -- which I'll call Independent Identity -- adopts standards gleaned from the WS-I library. This appears likely, since it was in that library that Kim Cameron, chief Identity Architect at Microsoft, found what he needed to structure what he calls the "identity metasystem."
This is a system that, by definition, is not owned in any way by Microsoft. The "meta" prefix in this case refers to inclusion. Above or around would be better synonymous prepositions than about (which is the approximate meaning of meta in metadata).
In fact, little or none of your customer data belongs to you, or is in your control. It belongs to the silo owners. To suppliers. It's data about you as Thrifty, Hertz, or Avis customer. Not as a rental car customer.
In other words, there is no such thing as a customer in the market for rental cars. Unless you understand that market to consist of a silo cluster surrounded by no open space at all.
Which is exactly what we do. We've been defining markets (actually, categories) as collections of vendor offerings for a dozen decades or more. We regard our silo memberships as a grace at best and an annoyance at worst. But in either case we frame our understanding in terms provided by the supply side. Our demand is not sovereign. Our choices are not free, because they between silos, and the offerings inside each silo.
Nothing clarifies the problem here more than a customer-side look at what we euphemistically call a Customer Relationship Management (CRM) system. Wikipedia says CRMs "enable organizations to better serve their customers through the introduction of reliable processes and procedures for interacting with those customers". But what kind of interaction are we talking about? Wikipedia explains,
The operational part of CRM typically involves three general areas of business: Enterprise Marketing Automation (EMA), Sales Force Automation (SFA) and Customer Service and Support (CSS). EMA provides information about the business environment, including competitors, industry trends, and macroenviromental variables. SFA automates some of the company's sales and sales force management functions; for example, keeping track of customer preferences, buying habits, and demographics, as well as sales staff performance. CSS automates some service requests, complaints, product returns, and information requests.
Many call centers use CRM software to store all of their customer's details. When a customer calls, the system can be used to retrieve and store information relevant to the customer. By serving the customer quickly and efficiently, and also keeping all information on a customer in one place, a company aims to make cost savings, and also encourage new customers.
Whatever its virtues, CRM is, at its heart, silo maintenance. Walled gardening.
Then imagine you want to buy something. Say it's a piece of . You make your attentions known to the set of mer
Characters
- Conversation on phone
Organizations
Vendors/Products
Meetings
Timelines
Kim's Laws:
(These) define the set of "objective" dynamics that constrain the definition of an identity system capable of being widely enough accepted that it can enable distributed computing on a universal scale. It is essential that we change the identity conversation enough that its laws are no longer argued as "moral imperatives", but rather as explanations of dynamics which must be mastered to craft such a universal system.
Craig's posts
Kim conversation:
When do you ever want identity information given out about you, when you're not there? What is the use case where I want my identity informatino to be given out by a robot? It's when I'm doing somethig that I come into a context where in return for some benefit I want to give some personal information out.
Can you give me some examples of cases when informatoin is given out when
Where does this type of thinking come from? It comes from people who do servers. It's an "If all you have is a hammer, then everything looks like a nail" thing.
With this Laws thing, I'm putting people at the center of the framework. People who don't believe that individuals belong in the center of identity can't really assail the position.
I'm not saying that their technology isn't useful. but it has to be the person's, the individual's decision, about what identity information is right in a given context.
It's a quetion of what serves what.
Nothing wrong with that, until you consider the absence of something: Your ability to present yourself as a customer to all the agencies at once, so they can compete for your business.
We all have a bunch of identities... Most of us don't usually think about it. They could be anything from your record at the ss administration... to a credit card number to Microsoft passport to your account at Amazon.com. So digital identity is becoming widely recognized as a distributed environment with a bunch of little pieces that end up identifying an individual.
The distributed nature (of identity) leads to certain problems. How is it we allow the end user -- be they an employee or an investor or a partner or a customer -- to actually have some sort of virtualized control over these distributed bits of digital identity? There are two broad solutions to that. One is to build something centralized: a great database in the sky. Larry Ellison's dream, right? We get a national ID program and everything goes into an Oracle database. The second approach, and the one that seems to be gaining in popularity, is federation. This approach leaves the distributed environment as it is, but seeks to let the end user link together those pieces and still have control over their privacy, what gets shared, and how.
The obviouis example of this is the whole Passport (vs.) Liberty Alliance thing that happened. Micrsoft came out with Passport, a centralized system... Then Sun formed the Liberty Alliance (which has) grown up around this concept of federation, mostly because companies generally don't like the idea of sharing their customer data with other companies, even if it's companies providing an identity service to them, like a Passport.
So, federation seeks to leave the distrubuted environment as it is and still attain the advantages and benefits of what would normally be a centralized database. The Liberty technology... is purposefully designed to be opaque. So the Liberty Alliance spec tries to accomplish one thing... to allow one end user to link one account to another account. But in order to protect the privacy of the end user... neither side of that account -- either Company A or Company B --would know who the end user is that is being passed between them. In other words, in oerder to find that it was me that was linking between Companies A and B, they would actually have to go outside of the specification through a backchannel. So the specification puposefully makes it hard to violate someone's privacy.
Grounds for Identity, November 2002: http://www.linuxjournal.com/article/6382
What's our i-Name? November 2002: http://www.linuxjournal.com/article/7888