Marcomony is the replacement of parsimony in science, where the burden of proof would normally be shifted away from a solution deemed as "simpler", by simply *not* shifting the burden of proof.
Thus competing hypotheses given the same evidence are on essentially a level playing field. A multitude of makhO's disposable razors rather than Okham' s individual razor.
I have listed in my head every case I could think of where parsimony is used in science, and I cannot think of a single one (yet), where I believe parsimony has benefitted science.
I believe models *must* necessarily be simpler than reality - eg Newtonian mechanics doesn't take into account relativistic effects. It is not true that the simplest solution is the most likely to be true. The point of simpler models is to fit better into our (perhaps prejudiced) world view, and to be able to be explained to a lay or more naive public.
5 comments:
I think your example lays bare the heart of the problem. You cannot really wish that we had ignored Newton and blundered along making bridges by trial and error until a perfect "theory of everything" emerged like Athena from the head of Zeus. It is *not* that the model that involves the least multiplication of entities is most likely to be true; it is preferred instead because it is useful. If I may appeal to your philoprogenitiveness rather than your autism-spectra-ness to look favourably upon the good rather than the perfect, Ockham's razor implies we should pick the simpler explanation because it is more *fruitful*.
Truth is not something we have, but something we asymptote towards. Not a delicate baby bird in the palm of our hand, but a terrible icy black mountain on the horizon that has never been climbed.
The point with Newtonian physics is that it can cope with the burden of proof. It does not need to be protected by Okham's razor from it, because repeatable observations back it up anyway. It is important to realise that it is "bounded", in this case, it is virtually 100% true for speeds< 1% of the speed of light. Similarly, it would be prudent to suppose that the uniformity of the universe "may" not be applicable inside black holes or in the first googlth of a second of the Big Bang. That is not really what I'm talking about. If you can demonstrate simplifications that do not shift the burden of proof, I may still be with you on this. However, simplifications where there simply is no repeatable observations may allow us to believe 6 impossible things before breakfast.
I should probably review the instances you complain about where hypotheses are prematurely ruled-out on the grounds of 'simplicity'. I suspect in each individual case I probably agree with you.
Did you need me to list them out again? I was planning to list my top six before breakfast tomorrow as a symbolic gesture. Anyhow, it is not quite about "ruling out" hypotheses, some of them being my pets which I obviously am going to favour, but it is about "protecting" hypotheses from the burden of having to be proven before we make a chain of hypotheses having to assume these protected ones true, which could lead us to believing a billion impossible things before we even properly wake up.
Here is a nice example to get your blood pressure up. There are some fine false dichotomies and unexamined assumptions imposed by a dodgy understanding of parsimony.
Note how early on the prediction of the standard inflationary model of an 'island universe' a trillion years hence where only our galaxy is observable to its inhabitants is made, but nobody thinks to ask the obvious question: "What if the evidence we need to explain the universe is unobservable to *us*?"
Post a Comment