Sunday, April 10, 2016

They Offer Nothing But Lies, 6


Once again, the creationists are telling fibs about information theory. Are they dishonest, or just stupid? In the case of Denyse O'Leary, I'm inclined to suspect the latter:

The belief that randomness produces information (central to Darwinism) is obviously false. It’s never been demonstrated because it can’t be. It is assumed.

No, it's not "assumed". It's proved. It's one of the most basic results in Kolmogorov information theory, demonstrated every year in the classes I teach. With high probability, a randomly-generated list of symbols will contain a lot of information. To understand this you can use one of Dembski's own metaphors: the combination lock. Which will be harder for someone to deduce, a combination that is your birthday in the form mmdd, or the first four digits of pi, or a randomly-generated 4-digit code?

This does not seem to penetrate the skull of the rather dense Ms. O'Leary, who then tries to weasel out of her claim by saying

by "information," one means here complex, specified information, produced in vast interlocking patterns on a regular basis.

Oh, so she's not talking about "information" in the way it is used by mathematicians and computer scientists. She's talking about creationist information, that vague, incoherent, and self-contradictory mess invented by Dembski and used by basically no one except creationists.

That mess was debunked years ago.

Here's an example: take any English text T, like the first 10 lines of a Shakespearean sonnet. Now apply any decent encryption function f to it that is not known to an adversary, getting U. To the adversary, U will look like random noise and hence be "unspecified", so it will not constitute creationist information. Now I come along and apply f-1 to U, getting T back again. Voilà! I have now magically created information deterministically, something Dembski claims is impossible.

No matter how many times you explain this, creationists offer nothing but lies in response.

4 comments:

Paul C. Anagnostopoulos said...

I think perhaps the creationist might say that U is specified:

"It is an encryption of the specified text T."

Of course, I've never really been able to grok what they mean by "specified." Sometimes it means "allowing a short description." Sometimes it means "prespecified." Perhaps my short description has to include the definition of f, in which case it's not so short.

~~ Paul

Jeffrey Shallit said...

The thing is that specification is always relative to an agent's knowledge base, according to Dembski. So all I need to do is choose an encryption function f that is not in your knowledge base.

JimV said...

At one point Dembski tried to abandon CSI but couldn't, I think because to his followers it was like abandoning the concept of the Trinity. Like the Trinity, CSI is theology which purports to make sense out of a senseless position and inspires warm feelings in those who would like to believe in something which doesn't make sense. Based on his actions - avoiding the Dover trial, and dropping out of ID involvement - I think Dembski knows CSI is bogus, but to the ID cult it is part of their revelatory scripture (and Dembski was its prophet).

Curt Cameron said...

I subscribe to the DI's podcast ID The Future, and recently they've had a couple of episodes hosted by Robert Marks. I've always noted how the creationists equivocate with the word "information", where they try to redefine it to suit their purposes but then still want to use the structure of the mathematically consistent definition.

However, in these recent podcasts, Marks says that creationist information is the difference between the Kolmogorov information and Shannon information of a data set. Something that has a lot of Kolmogorov information but not much Shannon (or is it the other way around?) would have a lot of CSI. IANAM, so I didn't really get the point they were trying to make with this. Something that's complex but not unlikely has information? Or the other way?

Jeff, have you seen them doing this?