In my inaugural column I mentioned a last-century cloud technology called AT&T PersonaLink. AT&T had a lot of the right ideas. PersonaLink was a distributed service that ran on your devices and in the cloud. Its agents interacted directly with you on your devices, and negotiated for you with other agents in the cloud. They represented your identity and respected your privacy.
Unfortunately for AT&T, PersonaLink depended on a closed network. The cloud in which your agents and other agents met to do business was a walled garden owned and managed by AT&T. In 1994, as the web was awakening, that was a fatal misstep and the story ended badly:
PARSIPPANY, N.J.--(BUSINESS WIRE)--July 11, 1996--AT&T today informed AT&T PersonaLink(SM) Services customers that the service will be discontinued on August 30, 1996. This action is a continuation of AT&T's strategy to shift from proprietary network-based services to the Internet.
It was inevitable that the ideas embodied in PersonaLink would resurface on the open web. Mostly they haven't yet. The personal cloud that exists today is mainly about file synchronization across devices and the virtualization of client/server web apps and services. This is great stuff, mind you, and I'll have lots to say about it. But this week I want to focus on an ambitious next-generation idea that's being implemented by a company called Kynetx and that's described in Phil Windley's new book The Live Web.
First, some disclosure. Phil Windley and I have been good friends since I brought him onboard as an InfoWorld contributor nearly a decade ago. We have worked together at IT Conversations. And I gave a keynote address at the Kynetx Impact Conference in 2010. I hope Kynetx succeeds. That said, I'm certain that the underlying ideas and capabilities are broadly significant. Here are three of them.
Every high-level programming language is, among other things, a wrapper around a set of lower-level sublanguages, the most universal of which is called regular expressions or regex for short. A regular expression abstractly declares a class of character sequences -- for example, all alphanumeric sequences that do (or don't) include hyphens. A hyphen-exluding regex makes it trivial for a program written in any higher-level language to transform a phone number like 603-555-1212 into its hyphenless counterpart. Whenever a website forces me to manually remove hyphens from a phone number I've entered on a form, I curse the programmer who could easily have spared me the chore.
In the event-oriented system described in Phil's book, the analog to the regular expression is the event expression or eventex. An eventex abstractly declares a class of event sequences, for example: a phone call received (more than once) from a caller whose number was also (more than once) the source of an SMS message -- all of this occurring within the last hour. Who would use a pattern like that? Suppose you're a customer service center. This sequence of low-level events from the same number, arriving on multiple channels, might suggest an increasingly frustrated customer. KEA enables and encourages developers to define that as higher-level event, log it, and then "raise" that event in order to trigger an escalated response.
The eventex sublanguage is embedded within a new high-level language called the Kinetic Rule Language (KRL). A KRL developer would include the phone-calls-and-SMSs-from-same-number eventex in a rule that logs the synthetic event and does the escalation. That might entail querying customer history, modifying a web page being viewed by a customer service rep, and perhaps alerting a supervisor by email or SMS. When a rule fires it can perform actions in a variety of domains supported by the underlying Kinetic Rule Engine. It can modify web pages, invoke web services, send email or text messages, make phone calls, and trigger other KRL rules.
Today's personal data stores synchronize files. What's on your phone or computer shows up in your cloud account, and sometimes vice versa. What Phil envisions is something I've also long imagined. As Phil puts it:
You shouldn't think of the PDS as a single place that stores all the data, but a single interface to ... data, regardless of where it is held.
The common interface doesn't merely join disparate cloud data stores. It also enables us to define the rules that govern access to our data, and to delegate the enforcement of those rules to software agents that work for us.
In 2006 Doc Searls coined the phrase The Intention Economy and used this example to illustrate what he meant:
In The Intention Economy, a car rental customer should be able to say to the car rental market, "I'll be skiing in Park City from March 20-25. I want to rent a 4-wheel drive SUV. I belong to Avis Wizard, Budget FastBreak and Hertz 1 Club. I don't want to pay up front for gas or get any insurance. What can any of you companies do for me?" --and have the sellers compete for the buyer's business.
In The Live Web Phil Windley works through a detailed example of how to make that scenario real. The Kinetic architecture isn't the only way forward, but it's one way, it's real, it's open, and it's inspiring.