Slashdot Mirror
Paterson's Worms Solved by Number-Crunching
An anonymous reader writes "Thirty years ago, Martin Gardner described Paterson's Worms to the world. Just recently, Benjamin Chaffin, one of the designers of the Pentium 4 chip, managed to trace a couple trillion steps of the 'unsolved' worms, and has pretty much solved all but two of them."
Did somebody say worms?
The brute force solving of problems can be very useful. The scientific method relies on theories, and having ample data to look at helps people understand complex systems, sparking the intuition that leads to more theories, and hopefully, more elegant solutions. I've worked on the optimization of large systems, and nothing helped me understand the processes involved as much as "brute force" simulations.
Brute force, aka trial-and-error, is what drives evolution. Brute force created the human brain, your mind, and thought.
Computers are useless. They can only give you answers.
-- Pablo Picasso
nah...
evidence indicates that it is not brute force that the mind uses, but rather hueristic pattern matching, followed by brute force. There is a huge difference.
It also allows for some rather incredible pattern matching and unbelievably stupid mistakes on the part of humans.
One of the more interesting things is that humans don't search for an exact fit when doing pattern recognition, they go for a "good enough" condition. (Rembeber teh atrilce on raeidng?) This actually allows for more rapid processing, but opens the door for some pretty stupid mistakes.
On the whole, though, the human mind is an incredible processor. It is also non-binary, since many nerves can exist in many different states, some of which are qualitative, and it is non-linear, and parrallel! Branches, forks, etc., are quite common, and each nerve connects to a LOT of other nerves.
"We don't know what we are doing, but we are doing it very carefully,..." Wherry, R.J. Personnel Psychology (1995)
I devoured his columns as a boy. His simple, clear writing style made it easy to understand very sophisticated concepts. Today, I aspire to write like he did.
He is getting on in years and it's been awhile since I've seen anything new from him (either on math or junk science, his other favorite topic). His collection, The Night is Large is a great overview of his work.
Anway, it's a pleasure just to see his name and know that people are still pursuing the topics he wrote about.
-- Brian
The most rabid believers in American Exceptionalism are the exact same people whose policies are destroying it.
Brute force is taking all the possible combinations (e.g. all the base pair combinations in DNA) and test them *once*.
Evolution takes a small sample (the current instances of gene combinations, i.e. the current generation) and creates another small pool (the next generation) depending on a selection algorithm (survival of the fittest). Most combinations are never ever tested.
And unlike a traditional brute force approach, the same gene combination may be tested many times (in theory at least), and the selection is not deterministic (that is, the "best" individual can e.g. die by chance).
Another thing, brute force may only find the best selection if there is one such combination of genes. Contrary to that, it is likely that there'll be specializations in the gene pool (e.g. at some point, many species specialized into male and female forms, some into worker/queen etc.)
Kjella (have moderated thread, so ACing)
Strange Idea, but, what about using this in encryption for pseudo-random number generation?
It's obviously simple to implement, but requires exponential processor/mem usage to generate each successive generation of pattern's. Would this be effective? would the reduced keyspace be better or worse than the computational requirements?
#!/bin/csh cat $0