[Home]PredictivePower

ec2-54-80-10-56.compute-1.amazonaws.com | ToothyWiki | RecentChanges | Login | Webcomic

There is a theory that a theory is only scientific if it has PredictivePower.  That is to say that if from it at least one prediction can be made that is, in principle, falsifiable. I.E. one can state under what circumstances one would accept that one was wrong and that the theory was false.

Predictions of what will happen after your death, or after the death of the universe generally lack falsifiability.

However, even if such is a necessary condition, is it sufficient?  If I claim that "all wombles are green" and say I will be happy to declare myself wrong if shown in person an example of a womble that is pink, blue or any colour other than green, does that make it a scientific theory?

Yes, according to some. RichardDawkins, in fact, alleges that a number of statements made by Christianity are scientific theories in this sense; i.e. they make or can be used to make testable predictions about the state of the world; and that some of these tests fail; which incurs a lot of wrath both from people who try to argue that science and religion are completely separate things, and from those that try to argue that they are not but that Christianity can somehow be demonstrated to be true by observing the world.

Not really, in another sense. A scientific theory is something that explains existing observed facts, as well as making testable predictions. What facts does "all wombles are green" explain? To clarify, the problem here is not that it doesn't explain the reason wombles are green (a non-ridiculous analogy would be the theory "All cells are DNA-containing", which talks about observable properties of observed things, but doesn't explain why all cells contain DNA - this is a (simplistic) scientific theory in the sense above); the problem is that no wombles have been observed to exist in the real world, so theories about their properties are not scientific since they cannot be related to observable data.



PredictivePower is only one factor in deciding which of several competing theories to accept. OccamsRazor advises avoiding unnecessary complexity.  ReaysLemma goes further and actually asserts that, other things being equal, the simpler theory is in fact more likely to be true.  Either way simpler theories that fit more of the facts better with fewer exceptions generally supersede more complex ones that don't fit so many of the facts quite so well.

Is "all wombles are green" really the simplest theory that fits the largest number of known facts?  The fact "I have observed no blue wombles" is explained both by the theory "all wombles are green" or by the theory "there are no wombles".  The latter's simplicity could be said to give it greater predictive power - it is making a stricter assertion, and is falsifiable by a greater number of cases than the former (for instance a blue womble being observed would falsify it).
PeterTaylor has observed some Caucasian (i.e. pinkish) wombles. In fact, he shared a train corridor with them from Preston to Euston.




The text above uses the word "theory" throughout. Actually it mostly ends up talking about the concept of a hypothesis - a prediction. A theory is something that joins together several of these with observed facts in order to try and establish some underlying principle or explanation. A scientific theory also has to be ultimately based on observable facts, make falsifiable predictions, not contradict observed facts etc., but the falsification of a single underlying hypothesis does not necessarily kill a theory - theories can be adapted to fit observed facts better. Theory-making is an iterative process - the aim is to always have something that fits all known facts and can be used to interpolate / extrapolate into the unknown.




Relevant info:

Link: [interesting rant on the difference between exploratory and confirmatory factor analysis, by 3-toed-sloth]



(Attribution left off - feel free to directly edit words)
CategoryScience CategoryThoughtExperiment? DocumentMode?; see also CreationScience
See also [Hitherby on the subject of falsifiability].

ec2-54-80-10-56.compute-1.amazonaws.com | ToothyWiki | RecentChanges | Login | Webcomic
Edit this page | View other revisions | Recently used referrers
Last edited January 29, 2011 11:02 am (viewing revision 22, which is the newest) (diff)
Search: