Sometimes really smart people, perhaps because they are harried or busy, help perpetuate badly flawed models of important ideas. Memes that get traction because they are easy to repeat, not because they are right. One that I’ve noticed a lot in recent times is what I call the sequential fallacy in talking about the Internet. This fallacy is at the heart of the story that goes ARPANET…Web 1.0…Web 2.0…Web 3.0/SemWeb. Here is how to get away from the fallacy and think more accurately about how the Internet is really evolving.
The Sequential Fallacy
Here is a picture of the sequential fallacy in the classical domain of evolution: biology. It is an “ascent of life” type of visual model that starts with single-celled primordial life and makes it way through increasingly complex stages to the magnum opus of the Blind Watchmaker — homo sapiens! No biologist or educated amateur thinks this way anymore, but a surprisingly large number of otherwise educated people still do (Daniel Quinn’s rather clever parable, Ishmael, has a neat episode where a starfish tells the story of evolution, ending with a triumphant “and then there was the starfish!“).
The more accurate model to keep in your head is the one at the bottom. Evolution is a branching process of exploration, constantly trying out various semi-randomly-constructed selfish-gene package designs for survivability, against a changing environment.
Occasionally bursts of branching speciation occur (such as the well-known Cambrian explosion) followed by bouts of extinction-pruning. There are relatively calm periods where much new differentiation focuses on minor variations on “trunk” branches, and more explosive periods, with a lot of major forking. The exact morphology of this branching, creeping process may involve subtle debates like the one around punctuated equilibria, but no serious biologist believes any sort of sequential model. At best you could say that sequencing the tree by means of one arbitrary metric (brain complexity) might put humans at the front of the line.
The Sequential Fallacy and the Internet
Now, computer scientists are smart people and possibly understand evolution better than biologists themselves do. After all, they invented genetic algorithms, and if you’ve ever seen a Subversion tree, you know they really do get it. But when it comes to introspection about their own technology, a curious sort of oversimplification seems to creep in, perhaps because it is easier to draw linear-sequential diagrams in Powerpoint (or maybe MBAs force them to dumb things down). Here are the sequential-error view and the more accurate branching view of the evolution of the Internet:
Even if many smart people actually believe the second kind of picture, the fact that they communicate using the sequential model helps perpetuate the fallacy.
A consequence of the fallacy is that some of the common critiques heard in “Web evolution” discussions are misguided. For example, “Web 2.0 is a distraction from the real goal of Web research, the Semantic Web.” This is not misguided in terms of content; the Web 2.0 branch really could be a dead-end that is sucking resources away from a more deserving SemWeb front. What is misguided is canonizing any particular vision of the Internet as the goal. The Internet is, and always will be, the outermost fringe of an exploration tree. Not a march towards a Promised Land.
How to Think and Talk about Web Evolution
Here is a worked example. With the tree picture in mind, zooming in on just the Web 2.0 and SemWeb parts, reveals many possible futures — the colorful lines. Either or both might be a dead end, or they might both successfully go different or convergent ways. That corner of the Internet might become the SemWeb, the SocWeb (my term for one potential asymptotic manifestation for the Web 2.0 line), a mix, or something else entirely.
Of the other major and relatively independent lines of development, the 3d Web that could take the Web from external-symbolic-metaphoric entity to submersive-sensory entity, is suffering the consequences of its own inflated-expectations phase in its hype cycle. But I still have hopes for it. Second Life is not a dumb idea.
The other, which I call Sensor Web, is a line of development that could potentially transform the Internet by flooding it with ubiquitous distributed sensor-network and RFID data. This line is currently under the stewardship of electrical and control engineers rather than computer scientists. I like to think of it as an Asimov-like Second Foundation. Perhaps I am biased, since by training I am a control engineer, but I find the vision of a SensorWeb the most exciting. More exciting than Web 2.0 or SemWeb.
But back to the model itself. There are legitimate reasons to use the sequential simplification. If you want to talk about the broad signature design themes of a particular “burst-extinction-calm” evolutionary epoch, and use one flourishing branch to represent the whole bush (“Web 2.0” for the whole diagram above) then the sequential model is fine. So long as you qualify what you are saying with a “but within these major phases, Web technology is evolving as a constantly branching, Darwinian exploratory process.”
So to summarize today’s lesson, think and talk in terms of the tree to the extent possible.