You’ve Got Ismail!

Evolution of the Internet

January 19th, 2006

Bill Burnham has just published another must-read post.  It’s one of those where you think “damn I wish I’d written that”.

Rather than summarize his post, which you can more easily read yourself, I’ll talk about the history of the internet and what we see coming up.  There are some important points to be made about why it’s the right time to make the web ‘active’, or as the Searls clan (i.e. Doc and Allen) put it, “the world live web”.

Three Patterns of Information Exchange
First, a bit of background.   There are three fundamentally important patterns of exchanging information between people, or between applications, or on a network.   The first is called ‘Messaging‘, which is a one-way transmission of information from A to B.   Examples of this include e-mail, postal mail, text messages and so on.   You address a piece of information, send it and trust the system to deliver it for you.

The second is called ‘Request/Response‘.   Here, we are going up one level of complexity to one question followed by one answer (A to B and back to A).   Examples of this are “what time is the train?”, database queries, client-server architectures, etc.

The third pattern is called ‘Publish/Subscribe‘.    This is again one more level more complex and consists of one question, but many answers.   The form of the question is “tell me whenever…”.   If I ask you “Can you tell me whenever you’re free”, I’m ’subscribing’ to a condition that may occur in the future.   All alerting systems are based on publish/subscribe, including Google news alerts, eBay auction alerts and so on.

These three patterns of information exchange are age-old and exist as components in almost every computer system.   Databases are heavily based on request response, collaboration software is laced with messaging components and so on.   Even Outlook has a little publish/subscribe (or pubsub) pattern in it where it wants your mail server to tell it whenever there’s new mail so it can beep, flash and notify you.

Evolution of the Internet
Now let’s examine how the internet has evolved.   In the 70s, we spent most of a decade working out TCP and TCP/IP and connecting computers together into the worlds first distributed network through the work of folks like Vint Cerf and Bob Kahn.

In the 80s, we implemented messaging, email being the big application that came out of that effort.    And it took a full 10 years to roll it out and reliably get a morsel of information from A to B across the network.

In the 90s, we implemented request/response, and the big application there is the Web.   You ‘request’ a page by clicking on a link, and the server sends a page back as a ‘response’.   The web is just a generalized request/response application.

Note that after messaging and request/response became first class citizens on the network, thousands of applications have since been layered on top of them.   Why?  Because they’re both archetypal information exchange patterns.

(Gee, where am I going with this?)

The Publish/Subscribe Decade
After the rollout of messaging and request/response, we are now entering the third wave of the internet, the publish/subscribe decade.   The web has been phenomonally successful and the amount of information available on it is overwhelming.   However, (as Bill rightly points out), that information is largely passive – you must look it up with a browser.   Clearly the next step in that evolution is for the information to become active and tell you when something happens.

Blogs and RSS are the first general manifestation of publish/subscribe. The real reason for the explosion of RSS/blogging is the ability to ’subscribe’ to a blog or feed and be told ‘whenever’. We can expect this theme to dominate the next several years of the internet.

What Bill refers to as the ‘active web’, or Doc as the ‘live’ web, we refer to as the pubsub web (ok, so we’re not the greatest marketers).    Our whinge would be that it’s the implementation of publish/subscribe on the internet that will make the web active.   People often talk about PointCast as the first major effort to “live” up the web.   It was an effort, but not very well implemented.

Conclusion
If you think about the early days of email, we had different email systems (MS-mail, ccMail, etc), each doing messaging in different ways.   Then the industry accepted SMTP as a standard and email became ubiquitous.    In the early 90s, there were different request response protocols (Gopher, FTP) until we standardized on HTTP, at which point the use of the web exploded.   For a great discussion on this, see Bob’s post, (which as usual, is definitive).    Today, in the early days of publish/subscribe, we also have different systems doing it different ways.   Google News Alerts, eBay auction alerts, Weatherbug, etc. all have different interfaces to handle the publish/subscribe interaction.    We can surmise that this, too, will standardize (pstp:// ?)

Whew, ok, enough for today.   Tomorrow, I’ll examine the search problem and expand on Scott Moody’s thoughtful comments, which require some brow-furrowing.   I think I can cogently talk about why social networking will be a smaller part of the future internet experience than he thinks.

If you’re looking for an audio version of some of what I’ve written (and more), check out Tom Raftery’s podcast interview with me…

Entry Filed under: Internet & Technology