Modeling The World: Part I
The problem I've been facing is that it looks as if everyone these days wants a hypothetical-predictive model of the world. This desire masquerades under many guises; I've heard it called the scientific method (despite sometimes not being particularly scientific or methodical). When pressed on the assumptions of the models produced, many scientists turn snarky and say, "Well, this is the best we have; it models reality, it predicts outcomes. What more do you want?"
There are a few difficulties with that. First of all, we can all agree that we use this model by simplification (the doctrine of parsimony, sometimes tagged as 'the Razor of Occam (or Ockham)'), and by induction (so far it's all been like this, why should it change), and by deduction (here is my rule, and if that is the rule, here is how it applies to everything).
The difficulty with parsimony is that it's a useless rule. It seems to be aesthetic (the idea of elegance being sparse) or convenient (the fewer entities we have, the simpler it is) or philosophical (nature tends to be fundamentally simple). But there is actually no reason to believe that the simplest hypothesis is the best, or to assume that what we're cutting away is non-essential. Of course, we should always try to cut away the non-essential, but the more complex the problem, the less we can identify what is non-essential.
The difficulty with induction is what most people call the black swan problem. Having seen that all swans so far are black (or that the sun rises everyday), we assume that all subsequent swans will be black (or that the sun will also appear as scheduled tomorrow). We know that this isn't necessarily true – it amounts to nothing more than an argument from experience until we figure out why this is true (swan genetics or celestial mechanics).
Which brings us to the third point. The difficulty with deduction is that you have to find a starting point from which things are deduced. This is called 'setting up the axioms'. Axioms are things that are self-evidently true. The problem is that there isn't much that is self-evidently true, on closer inspection. In fact, if we invoke parsimony, the best universal model begins with ONE self-evident truth and continues from there. If it began with NO self-evident truths, then the model would not exist and hence would not be 'best' – or anything else for that matter.
The tripos on which the scientific method is based is therefore aesthetically pleasing, convenient and useful, and philosophically satisfying to a lot of people. It is rightfully a majority point of view. Which does not make it right, but it's the best we have for simple problems. For complex problems, and for determination of what is true and what is axiomatic, it will run into difficulties.
And the most complex problem is this: we exist. It is the only truth that has to be axiomatic, for without it, there are no models and no reason to create models. If only one of us existed, ditto. But as a multiplicity of entities, we are actually intolerable to the scientific method, and some other kind of modeling philosophy must be used to handle the problems of our existence. The alternative is what philosophers call the reductio, and which Hitler called 'the final solution'.
There are a few difficulties with that. First of all, we can all agree that we use this model by simplification (the doctrine of parsimony, sometimes tagged as 'the Razor of Occam (or Ockham)'), and by induction (so far it's all been like this, why should it change), and by deduction (here is my rule, and if that is the rule, here is how it applies to everything).
The difficulty with parsimony is that it's a useless rule. It seems to be aesthetic (the idea of elegance being sparse) or convenient (the fewer entities we have, the simpler it is) or philosophical (nature tends to be fundamentally simple). But there is actually no reason to believe that the simplest hypothesis is the best, or to assume that what we're cutting away is non-essential. Of course, we should always try to cut away the non-essential, but the more complex the problem, the less we can identify what is non-essential.
The difficulty with induction is what most people call the black swan problem. Having seen that all swans so far are black (or that the sun rises everyday), we assume that all subsequent swans will be black (or that the sun will also appear as scheduled tomorrow). We know that this isn't necessarily true – it amounts to nothing more than an argument from experience until we figure out why this is true (swan genetics or celestial mechanics).
Which brings us to the third point. The difficulty with deduction is that you have to find a starting point from which things are deduced. This is called 'setting up the axioms'. Axioms are things that are self-evidently true. The problem is that there isn't much that is self-evidently true, on closer inspection. In fact, if we invoke parsimony, the best universal model begins with ONE self-evident truth and continues from there. If it began with NO self-evident truths, then the model would not exist and hence would not be 'best' – or anything else for that matter.
The tripos on which the scientific method is based is therefore aesthetically pleasing, convenient and useful, and philosophically satisfying to a lot of people. It is rightfully a majority point of view. Which does not make it right, but it's the best we have for simple problems. For complex problems, and for determination of what is true and what is axiomatic, it will run into difficulties.
And the most complex problem is this: we exist. It is the only truth that has to be axiomatic, for without it, there are no models and no reason to create models. If only one of us existed, ditto. But as a multiplicity of entities, we are actually intolerable to the scientific method, and some other kind of modeling philosophy must be used to handle the problems of our existence. The alternative is what philosophers call the reductio, and which Hitler called 'the final solution'.
Labels: Epistemology, History, Philosophy, Science
etched with acid at 9:54:00 am
4 Comments:
Ideally what we don't know won't affect us... Mmm. Seems as if we can't know anything accurately due to the inherent assumptions existing everywhere in and around us heh. So cynical!
Well, then if we don't know anything, nothing can affect us! But empirically, we think (or feel) that we are affected. So it is best (as in makes us feel most in control) to assume that we know at least that we perceive this and that there is some cause leading to the effect of perception.
Or else... haha!
But.. Why do we have to "assume" that we "know" we perceive something? Why not just "perceive" something? Haha. Seems like a sort of catch 22: We'll never know if we are perceiving with certain arbitrary assumptions because then our perception of said assumptions would be affected :O
The point really is that we shouldn't care if we know or not, since we cannot know. So we operate on the assumption that what we think we know is what really is. Otherwise, we should just all be nihilists and go away.
Post a Comment
<< Home