HARD CELL

Dick Pountain (21/10/1999 11:12am): Idealog 63

Once or twice a year my old friend Tim and I get together to talk rubbish. I must immodestly point out though that it is a rather exalted kind of rubbish, as Tim used to study the mathematics of relativity (speciality, twistors). We sink a couple of drams of Highland Park and maybe a herbal chaser, and then just take flight: Tim holds forth on field theories in n dimensions and I parry with algorithmic complexity as it relates to the halting problem for Turing machines, and now and again we strike sparks. We had one such session quite recently and the only reason that I'm telling you about it is that it clarified a problem that I've been droning on about for years in this column, concerning the relationship between the human organism and computers.

We had been discussing the possibility that everything in the universe can be considered to be a computing device of some sort - at the very least a thing might store one bit by either existing or not, and most things store far more bits than that. That lead us to realise that we were indeed talking rubbish, as the existence or not of thing cannot be expressed in one bit: certainly you can identify, say, a particular penguin with the value 1 (or the Boolean value True) but you cannot meaningfully identify the absence of a penguin with 0 or False as this leads to absurdity. How many non-existent penguins are not there in your living room at this moment? Devise an algorithm for counting them. (You should be getting the flavour of our sessions by now).

That lead us on to consider how the concepts of truth and falsity can be represented by the halting of Turing machines: if you identify truth with a program that halts and falsity with a program that doesn't halt, then the deep oddness of these concepts becomes clear, since truth is something that you might witness (if you are lucky) but you can never witness falsity. If the program has already halted, OK, but if it hasn't, perhaps you haven't waited long enough yet. It's rather like living in a remote place and waiting for the one bus of the day - does the fact that it hasn't turned up yet mean that a) it's on its way, b) that you'd just missed it or c) that the company has gone bust and it is never coming.

Setting off at a slight tangent (as if that mattered!) I then raised the issue of the human organism, and which parts of it can be considered to be computing devices. I'll digress here to say that this is rather a sore point at PC Pro, because I offered to write a feature on this very topic last year, but failed to deliver it after I discovered that the scads of scientific papers I'd expected to uncover turned out not to exist: all I could find were pretentious and largely meaningless post-modernist ramblings produced at the Santa Fe Institute. In our marathon ramble, we observed that most commentators have so far dwelt on the similarity between the human brain and a computer (a subject I've returned to several times in this column) but we decided that far more interesting is the human genetic system. Every single cell in our bodies contains the whole blueprint for making a new copy of our body (Jurassic Park style), stored as a digital code (DNA base triplets) and provided with a highly efficient processing system that can duplicate the information and actively correct errors. The ribosomes then turn an RNA copy of the DNA into protein, and they certainly resemble little computing devices (though rather less general than Turing machines), far closer to the silicon-based digital computer than our analogue neurons are.

That was when the thunderbolt struck me. At one level the ribosome may look as though it demonstrates a similarity between human body and computer, but at a different level it is actually a pointer to the gaping, fundamental, perhaps unbridgeable, difference between them. What the ribosome does is turn a cell's information into a cell's fabric, protein. Computers only work on digitised representations of the world - in Nicholas Negroponte's famous dictum they 'move bits, not atoms' - but living things not only distribute that representation throughout their fabric (since every cell contains DNA) but are capable of turning it directly into new fabric (protein). Living beings are the beings that build themselves, not merely represent themselves.

In contrast interfacing a digital computer to the real world is a nightmare task, as anyone who has ever indulged in robotics, or indeed any sort of real-time programming will tell you. Once your signals leave the cosy world of RAM for the hard (and soft, and viscous, and elastic) world of objects and actuators, things get messy and complicated.
Those wildly optimistic proponents of nanotechnology who persist in prophesying that we will soon be sending miniature submarines through our arteries to diagnose and cure all diseases completely overlook one huge fact. Certainly micro-engineering is making great strides in building miniature motors and actuators, but they are built using silicon fabrication technology: a bloody great factory full of UV lamps and etching chambers and furnaces, billions of times larger than the device it manufactures. The fatal weakness of silicon-based computing is that a computer is not a factory, and a factory is not a computer. The humble cell on the other hand not only contains its own blueprint, but the whole factory too.