“All models are wrong. Some models are useful.” — George Box


smoothingsplines
(Image courtesy of the Damien Garcia.)

As a statistician and quant, I’ve thought hard about that oft-cited Boxism. I’m not sure I agree. It’s not that there is such a thing as a perfect model, or correct model, whatever in the world that would mean, but it is that whatever is appealed to in our heads or “gut” when we look at a model and find it wanting is, well, just another model. Frankly, after a lot of practice, I think models get a bad rap: I don’t think it makes any sense at all to look at observations without having at least an informal model in mind. Observations are necessary but rarely can they stand on their own; not and have a chance of being generalized, anyway. That’s because we really can’t see something, let alone understand, unless it’s abstracted away, beyond details.

bspline_cubicle_smoother
(Image courtesy of the Open Motion Planning Library.)

I like this other quote, from physicist Arthur Eddington:

It is also a good rule not to put overmuch confidence in the observational results that are put forward until they are confirmed by theory.

About hypergeometric

See http://www.linkedin.com/in/deepdevelopment/ and http://667-per-cm.net
This entry was posted in abstraction, American Association for the Advancement of Science, astronomy, astrophysics, mathematics, model-free forecasting, numerics, perceptions, physical materialism, physics, rationality, reason, reasonableness, science, spatial statistics, splines, statistics, the right to know, theoretical physics, time series. Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s