Why God Invented Quantum Mechanics Mar 1994

Gather round, sit down, and make yourselves comfortable. I want to tell you a story....

A long time ago, in a region of superspace not very far from our own, lived two multi dimensional, pan galactic, hyper intelligent beings; let us call them Fred and Jim. One long Sunday afternoonish sort of aeon, when they were bored with appearing in three space as white mice, they engaged in a little friendly banter. "Bet I can make a better universe than you can", said Fred. "Bet you can't", said Jim. Fred thought deeply for a while. "Bet I can make a universe which can evolve 3-creatures so smart they can figure out what the rules of the game are". Jim smiled hyper dimensionally, "no cheating - purely emergent from initial conditions?". "No cheating" said Fred. "OK", said Jim, "You're on - lets say a pint of ambrosia". "Deal", said Fred. He was a being of much thought, but few words.

Now Fred had some hard thinking to do. He knew how to create a universe easy enough, but how to ensure the right initial conditions ? Being hyper intelligent, he obviously knew all there was to know about chaos and complexity theory, and realised that this meant that even he couldn't figure out what the universe would do without creating it - and that was a Catch 22 if ever he knew one.....

Now, dear reader, I would like to digress for a while into the field of CAD, in which dwells the subject of Logic Simulation - or more generally, digital systems simulation. The aim is to simulate the behaviour of a digital system to an applied pattern of waveforms. Simulation programs model the system, and apply the patterns to produce event lists, which can trace what happens through the system as the 'clock' progresses. There are also a few fancy tricks such as using multiple values to represent indeterminate states in addition to logic 1 and 0. An extension of basic simulation is Fault Simulation. Here the aim is to find out if a particular test pattern exercises all the logic gates in a system, or at least to find out which gates it does exercise. Then a complete test can be built up to ensure that all parts of the circuit are tested - no mean feat when there are thousands (and more!) of gates. Logic simulation is bad enough, because there is just so much work for it to do; but fault simulation is worse. Every time an event can pass through a gate, the simulator has to cope with what happens if that gate is faulty (such as 'stuck at 0'). It tries to do this by 'splitting' the state of the system at that point into two variants - one with the fault, and one without. In fact, there can be more than two variants if more than one fault type is being simulated.

So far so good, but there is an explosion of data - as each scheduled event is examined, it can create more variants. Soon there can be a vast number of variants being processed. Practical necessity forces a limit on the number of variants being handled, so that many different 'runs' of a fault simulator are necessary to cover all possibilities. The only mitigating factor is that variants can 'converge' as well as 'diverge'. This is because a signal that may have two variants can be masked at a future stage by another gate (e.g. by being held down by some other signal), and hence return to a single value. The problem is that the program is being run on a sequential Von Neuman architecture, so each variant has to 'time share' the processor. Now what is really wanted is a parallel processor, and one with an indefinite number of parallel streams.....

Our friend Fred, being ineffably smart, soon realised all this stuff that I have been explaining, and rapidly came to the only possible conclusion - he had to have an infinite number of parallel versions of his universe. That way, although most of them would be incredibly boring, a few of them would be really interesting, and the winner would produce smart life of its own first! (pretty much like having an infinite 'Life' (a la Conway) plane, really). All he had to do was arrange that any event which took place would split into parallel versions. Being so smart (at least a smart as David Deutsch of Oxford) he realised that the variants would converge as well as diverge - just like in our good old fault simulator. This meant he could get away with a fairly limited number of infinite variants (Aleph null, I guess?), which was good because he was a dab hand at Hilbert spaces. The rest, as they say, is history....

And that, children, is why God invented Quantum Mechanics.

Along with many armchair philosophers, I've pondered on all the interpretations of QM. I didn't much like the Everett / De Witt version when I first came across it - so wasteful! Where was Occam's razor for God's sake? Recently I've noticed that the Quantum Cosmology boys more or less take it for granted. They have to - becasue there's nothing to 'collapse the waveform' when you write down the waveform for the whole universe. I'm not sure whether they really 'believe' it, or whether, like most physicists, they just do the maths and leave the interpretation to take care of itself. I wanted to ask Hawkins at a lecture of his I went to in Cambridge last year - but (understandably) he doesn't take questions at public lectures these days. As far as I can see, they also think it's the only interpretation which holds water - all the others have insurmountable inconsistencies; its just difficult to really believe (but then all quantum theory is screwy). You figure it out, what would you do if you were a God....

PPS: Of course there are a few snags in this argument - why not use a universe of infinite extent, rather than an infinite number of versions of a finite universe? Well, I guess the latter is more elegant in some ways; maybe there are other snags, I can't quite figure it out - but then I'm not as smart as Fred. Perhaps you don't need an infinite number - I'm still a little suspicious of infinities. Anyways, this is the closest I can come to thinking of a reason why something as weird as the quantum world should exist. If you want to read about quantum cosmology, I can recommend 'The Anthropic Cosmological Principle' by Tipler & Barrow, but be warned - its a biggg book.