Personal Data Stores

Personal data stores allow individuals to share online data with service providers. Facebook Connect users can give third-party web sites like Digg, Amazon, and YouTube access to information stored at Facebook, turning Facebook into a personal data store for over 500 million people.

What makes personal data stores special is the seamless sharing with websites for real-time personalization of the web. It’s more than just file back-up or synchronization.  It’s not just publishing “content” to our friends or the public. Personal data stores allow us to bring our information to websites when we want to. It’s a way to treat the user as the point of integration.

Personal data stores can be anywhere, shared with websites whenever we want. Consider giving FedexKinko‘s a link to a Flickr account so they can download photos to print a new calendar. Or giving a new doctor permission to access our personal health history rather than filling out a paper form while we sit in the waiting room. Or giving a website access to our Outlook contact list on our desktop computer so they can give us birthday reminders and gift suggestions. The key is user-managed access, wherever the data lives. Facebook Connect gives this kind of access control over all the data we store at Facebook, enabling web-wide personalization built around the individual.

Mash-ups

In recent years, mash-ups and real-time APIs have made it easier and easier for companies to combine information from different services into a single user experience. Instead of building bigger and more complicated proprietary data silos, companies take advantage of services like Google Maps and IP-address geolocation, using real-time information to enhance their websites.

Some service are even built around other companies’ data: Twitter clients like Seesmic and Tweetdeck, which access our Twitter data on our behalf; Trillian, which works with various instant messaging networks; and Mint, which pulls in our financial data from hundreds of websites. The “real-time web” is constructed on the fly, using linked data and real-time APIs to dynamically customize services for each of us.

Personal data stores let us bring our own data to the mash-up party. Not only do we have better control over who sees what, we can provide more timely, higher quality data than service providers can get from other sources. Effective integration with personal data stores means no more ads for that car we’ve already bought; no more recommendations based on false assumptions. Unfortunately, data in the wild is constantly becoming outdated, miscopied, and misconstrued, because that’s the best companies can do using the billions of dollars worth of proprietary data that’s gathered about us rather than provided by us. Personal data stores easily allow individuals to give the most relevant, most up-to-date information to just those companies we want to do business with. That means not just better data, but more intimate relationships with our favorite companies and organizations.

Perhaps the most liberating aspect of personal data stores is that everyone gets to have as many as we want. We all have our favorite websites for different online activities. As those sites open up their data with a user-driven permissions mechanism, they become personal data stores. So, whether it’s YouTube for videos, Flickr for Photos, Foursquare for location updates, TripIt for travel plans, or RunKeeper for exercise data, we get to bring our best data with us wherever we go. Savvy websites pull in this high quality data to personalize our visits, while those with unique data open it up for use elsewhere to maximize value to their users, which is exactly what Facebook is doing with Facebook Connect.

Facebook Connect

Facebook Connect makes this kind of access simple for everyone, with industry changing adoption rates. Over 66% of the top 100 websites and over 1 million total websites now integrate with Facebook in some way. Nearly 1/3 of Facebook users—over 150 million people—use Facebook Connect every month. Every time we do, we give websites access to information stored in our Facebook accounts, such as our name, gender, names of our friends, and all the posts currently on our wall or posted by us. It’s an archetypal personal data store, with highly credible and timely data in the form of our friend list and our status updates. Sure, Facebook Connect is still far too limited in the amount of information we can store and we lack control over how that information gets used… but architecturally, Facebook has changed the game for a vast portion of the World Wide Web.

To find out what information Facebook is sharing, I built a website called “I Shared What?!?“, an information sharing simulator for Facebook. The site uses javascript and Facebook Connect to display everything it can get from Facebook. Visitors see in specific detail exactly what they share when hitting the “allow” button in the Facebook Connect permissions dialog.

Facebook uses open standard technology to bring mash-ups to a new level, built on information provided directly by the user, in real-time, with minimal fuss or bother. There are shortcomings, of course. A lot of them, but I’ll save those for future posts. For now, think of Facebook as the 800 pound icebreaker of a new way for companies to connect with their customers.

To this veteran VRM evangelist, Facebook has done more in 2010 to usher in the era of the personal data store than anyone, ever. In one fell swoop, Facebook launched a World Wide Web built around the individual instead of websites, introducing the personal data store to 500 million people and over one million websites.

Unexpectedly, Facebook has moved VRM from a conversation about envisioning a future to one about deployed services with real users, being adopted by real companies, today. We still have a lot of work to do to figure out how to make this all work right—legally, financially, technically—but it’s illuminating and inspiring to see the successes and failures of real, widely-deployed services. Seeing what Amazon or Rotten Tomatos or Pandora do with information from a real personal data store moves the conversation forward in ways no theoretical argument can.

There remain significant privacy issues and far too much proprietary lock-in, but for the first time, we can point to a mainstream service and say “Like that!  That’s what we’ve been talking about. But different!”

I’m still haunted by hearing that users get a maximum number Twitter postings (tweets) before the old ones scroll off. If true, it means Twitter is a whiteboard, made to be erased after awhile. The fact that few know what the deal is, exactly, also makes my point. Not many people expect anybody, including themselves, to revisit old tweets.

The flow-by web is great for sampling the current pulse of selected friends, an ephemeral dipping of the news ladel into a current river of updates. Yet it is also a place where people share things they often don’t share elsewhere, which makes it a great fishing pond for lightweight pointers to interesting media.

I have often used my own tweet stream–or others’–as a reference point when looking for websites or YouTube videos I first saw in the update stream. This happens consistently with media that’s fun enough to tweet but not important enough to blog. Often, in ordinary conversations, I find myself referring to resources I linked to in a prior tweet. Sometimes I just piont them to my Twitter account. Other times, I look it up myself and email the link.

Perhaps that works for me more than most because I don’t tweet that frequently, so my history is relatively compact. However, everyone’s tweets stick around, see My First Tweet as a case in point. In fact, it is perhaps more problematic that people consider these tweets gone, when, in fact, they are not.   Even though it is possible to delete your tweets from you stream history, it doesn’t remove them from all the downstream syndicators and third-party clients.

People should have more control over the lifetime of our shared information. In particular, it seems to me that people should be able to share information in one of two modes: at-will and on-the-record.

At-will posts can be erased by the owner at will, whenever they want, thus avoiding those embarrassing photos–and the resulting oppression from the future that keeps us from living fully in the present.

On-the-record posts are taken to be additions to the permanent record, with all parties understanding that they will be (or should be) always accessible. This allows for statement by officials operating in their official function, statements of policy, contractual agreements, and similar permanent records.

Everyone should be able to decide in which mode they want to share information, just as we can select a Creative Commons license as an alternative to copyright.  A simple microformat-style tagging system would go a long way to enabling a self-asserted, voluntary compliance approach.  Even better would be a data sharing protocol that could actually assure that compliant parties erase the at-will data when we no longer want those hot tubbing photos shared publicly.

Instead, we currently have a lot of frustration and surprises when people share information in one context only to find it appearing in another, undesirable one, some unpredictable time in the future. SMS messages, emails and Facebook photos, all show up in the most inopportune places and times. Managing these contexts–and the information we share in each–is vital in a world where we fluidly flip contexts as quickly as kaleidescopes change color.

As a co-proposer of the User Managed Access working group and acting co-chair of the User Driven & Volunteered Personal Information workgroups at Kantara, I am hoping we can find a way to make this new model work. Because the current model is too fragmented to be managed reasonably, and it is only going to fragment further unless we start to unify around user-driven principles or something similar.

The idea is that we should coordinate a global conversation to develop a Standard Agreement that credibly represents the voice of the user. We’d start by clarifying the data rights for user contributions and eventually cover the entire TOS in a framework addressing the standard kinds of transactions most users do every day online. The immediate benefit for users: ridding the web of nearly all the annoying checkboxes we are currently forced to check, indicating agreement to legal contracts we never actually read.

Facebook has seen the same writing on the wall: users are already in charge. After the fiasco of trying to update their TOS last month, they have turned to a vote of their membership to select the new Facebook TOS to use.

Of course, members can only vote for a slate already approved by Facebook, but it’s a start.