The Internet was not Built Wrong

Posted on October 29, 2008

Alex Payne writes in an article entitled, The Internet’s on Shaky Ground:

The Internet is built wrong.

Well, OK, not so much “wrong” as “not the best it could be.” But for all its problems, we can’t rationalize a radical departure from the technologies that currently comprise the Internet. Why?

Evolution.

Radical change is very rare in the evolutionary process. In biology, it is theorized that radical leaps are so rare that we can only find two possible instances where it might have occurred: the leap to eucaryotic cells and the emergence of consciousness. Everything else has been the result of natural selection. The Internet has evolved like biology, only on a much faster scale. However, the elegant lattice of the process is still the same and I would venture that the Internet has not encountered any radical changes in its evolution.

There’s a software engineering philosophy called Worse Is Better that’s been around for a good 18 years. In that time, it’s come to mean that an inferiorly designed system or piece of software may be more successful than its better-designed competitor, based on subjective or unexpected market criteria. A perfect modern example is that, while most Web applications are poor facsimiles of their desktop counterparts, they’re successful because there’s nothing to install and collaboration is a snap.

“Worse is Better,” appears to me to be a great way to start a partisan argument. The truth is that web applications have features that a group of developers and users believe to be advantageous enough to warrant their attention and efforts. Whether one is better or worse than the other is subjective and matters little. The Internet is a large complex system made of many smaller systems. It’s more feasible to think of it as an evolving ecosystem of software. There is not a single dominant email client any more than there is a single dominant browser, web server, database, and so on.

The Internet has developed incrementally much in the way that Darwinian evolution has. As users demand new features and developers think of more efficient ways to do things, the system as a whole changes and improves. Sometimes flaws develop when features once seen useful find themselves in competition with changing demands upon the system. Eventually new features will simply surface and supplant these flaws and everything will continue on.

In contrast, radical change should be abhorred. It can introduce radical failures as easily as it can introduce radical benefits. Resiliency is a side effect of avoiding extremes. As flawed as SMTP is in today’s environment; it’s still useful as a system and too much of the Internet depends on it to operate. A better protocol can take its place, but it must take an incremental approach or else it poses too much risk in disrupting the balanced ecosystem that exists.

So I say: No, the Internet was not Built Wrong. To misappropriate a Lisp meme: it just smells funny.