There are times when I'm sitting in some product launch or technology briefing and I wonder what on earth possessed me to come along. Last week was completely different. Not only did I understand why it was important to attend the briefing I was at, I also suddenly realised why it had been worth attending an even more obscure event back in March.
On that earlier occasion, the only immediate consolation had been that it took place at Bletchley Park, which I'd never visited before. Once the home of British codebreakers in the second world war, this is now the site of a museum which includes a working replica of Colossus, the world's first programmable digital calculating machine.
Otherwise, it had seemed a total waste of time. The iconoclastic analyst firm Bloor Research had flown in a Canadian academic to reveal a product which it claimed could render all modern software obsolete. Professor Bernard Hodson's contention was that the father of modern computing, Von Neumann, had set the industry on a false path when his team built the world's first digital computer, ENIAC, back in 1948.
Hodson, now in his sixties, had never gone along with what he saw as Von Neumann's excessively mathematical view of computing. He had spent more than thirty years single-handedly developing Genetix, which he believed held faith with Alan Turing's landmark theory of the Universal Turing Machine. The result is a computing environment that builds applications out of small software 'genes' without needing compilers, languages or operating systems.
Sadly, Hodson's premises seemed suspect, his presentation was nigh on impenetrable, his demonstration software kept falling over, and temporary deafness had rendered him incapable of responding to detailed questioning.
Self-aware storage
Last week's briefing dealt with another academic exercise, this time sponsored by Seagate Technology along with other big names in storage such as IBM and HP. Their exciting new concept is the storage object. Researchers at Carnegie Mellon University in the US are experimenting with adding a small dose of self-awareness to bits of data stored on hard disks.
Instead of simply storing meaningless rows of bits as commanded by host computer's file system, their disks store clusters of bits which have certain attributes associated with them. The attributes tell drives things like which bits go with each other, how often they should be backed up, and so on. The model is attractive because it enables storage to improve its performance and manageability, while laying down a foundation upon which true multiplatform, universal storage could be developed.
The light that went on in my head was that here was Seagate talking about the same type of componentisation of function that Professor Hodson had been proposing. Without going to the same extremes, it was nevertheless breaking functionality down to a very low level.
The implication is that Hodson's vision is correct, albeit probably a hundred years or so ahead of its time. Gradually, step by step, all of computing will ultimately get boiled down to simple, reusable software components that deliver functionality precisely where it's needed.
Hodson's mistake is to try and do the whole job now, all at once. The components can only work when it's absolutely clear exactly what we need them to do. We can only achieve that degree of perfection through many decades of trial and error.
more columns