At the recent very silly Engineering and Metaphysics conference, there were at least three different talks with mathematical content.

But the funniest by far was by Eric Holloway, whose masterful work I admired once before. You can watch Eric

- claim that intelligent design will revolutionize all human thought
- shamelessly promote Dembski's "complex specified information", without mentioning that this bogus concept has been debunked in detail
- claim that physics deals with only two kinds of "agents"
- claim that a "search process" is not an "algorithm" - and then later talk about "search algorithms" (!)
- confuse the complexity classes NPC and EXPTIME
- claim that all polynomial-time algorithms depend on "incrementally find[ing] better solutions"
- confuse finding solutions with maximizing a function
- claim that the travelling salesman problem can be solved in polynomial time (at 19:14)
- claim that humans can solve the travelling salesman problem in linear time
- repeat Stephen Meyer's lie that "only intelligent agents create information"
- claim that "clouds" have "no information" (and hence imply that weather prediction can be done without any information at all!)
- claim that a
*specific instance*of a maze can be changed, by removing a wall, into an NP-complete problem - thus making two fundamental errors in one sentence

All good stuff! You can see why creationists have to set up their own parallel pseudoscience conferences, because junk like this would be laughed out of any real scientific or mathematics conference.

## 9 comments:

I wonder if he got the "Salesmen solving the problem in linear time" from Stan Kelly-Bootle's "Devils DP Dictionary"

"Travelling Salesman Problem: A classical scheduling problem that has baffled linear programmers for 30 years, but which, in a more complex formulation, is solved daily by travelling salespersons."

I think it comes from papers like this, but the problem is that if you read the paper, there are definitely at least two jumps in logic from the data in the paper and the conclusion that people can solve TSP in linear time.

intelligent design will revolutionize all human thoughtYes, it is quite revolting.

physics deals with only two kinds of "agents"Let me guess...FBI and CIA?

a "search process" is not an "algorithm" - and then later talk about "search algorithms"So what is this "Al Gore-ism" anyway?

the travelling salesman problem can be solved in polynomial timeYes, the travelling sales

manproblem. But the travelling salespersonproblem...well, that involves women (and children too), so is at least three times as complex. Besides, real men(TM)never stop and ask for directions, they just keep driving. Hence linear!!!111one!!!humans can solve the travelling salesman problem in linear timeI doubt that...no successful salesperson thinks linearly.

"only intelligent agents create information"Exhibit A: Ray Comfort.

Exhibit B: Kirk Cameron.

No information or intelligence between the two of them.

QED.

"clouds" have "no information"To the cloud!!!!

weather prediction can be done without any information at allSeems to work for a lot of TV stations.

You may be entertained by Eric Holloway's recent

mathematical proof that intelligent design is necessarily supernatural.

Which of course means that ID is religious, contra Casey Luskin's "I believe it's God but ID only says it's the Intelligent Designer."

Jeff, could you give us some more detail as to what Holloway gets so wrong? Yes, I know the whole "information can only be created by intelligence" is bogus, but I am not so up on computational theory that I know off the top of my head how difficult TPS is.

Thanks, good feedback. I'll respond per point at a later time.

One point, the claim is not that people solve the TSP in linear time, but that time taken to find non-optimal solutions comparable to what the best algorithms can find scales linearly or near linearly as problem size increases. However, the time taken by the algorithms scales at a polynomial or greater rate.

time taken to find non-optimal solutions comparable to what the best algorithms can find scales linearly or near linearly as problem size increasesBut the crucial thing is, over what kinds of problem instances, and what range of sizes?

Randomly-chosen TSP's are not likely to be very hard. You also need to study TSP's that are likely to be challenging, such as those that arise from reductions from known hard instances of factoring.

And you also need to study problem sizes with thousands or tens of thousands (or larger) vertices. Human performance on small instances is not likely to scale well, in my opinion.

I would like Eric to defend his claim that only intelligence can create information. That is the crucial falsehood.

Also since he thinks intelligent design is supernatural, which laws of physics are being violated in our brains and in which region. Descartes thought it was the pineal gland. If first law of thermo is violated I say we make an array of brain batteries like the Matrix.

What score would this presentation receive from the Crackpot Index?

http://math.ucr.edu/home/baez/crackpot.html

Post a Comment