Evolutionary computation includes the non-deterministic element of random mutation, but that is not all there is to it. There are populations for example.
Non-determinism only acts beneficially when aided by the program. Basically, the difference between a blind and an assisted search.
If by "beneficially" you mean move closer to the target, or reach the target then even blind searches can potentially do that. So that isn't the difference between blind and assisted searches.
If you will note most "evolutionary" algorithms are very, very assisted -- at minimum in their semantic constraints on the possible datasets.
They are assisted, but what does that mean? For example recently I had to make a program to solve suduko puzzles. I was lazy so I used a genetic algorithm. I did not hard code how to solve them, in fact I did not even contemplate the logic of solving a suduko puzzle. I simply assigned a fitness based on how many numbers were in the right squares. Okay it didn't find the exact solution, but it did find highly optimum configurations where only two or three numbers were in the wrong place.
How would you say my suduko example was "very" assisted? The fitness function was simply a count of numbers in the right square. The mutation operator I used was simply to swap two random numbers. I used no crossover.
The point that ID makes is not that stochastic processes don't operate, its that stochastic processes are useless unless assisted in weeding out a large portion of the solution domain. Evolutionists completely ignore this when they point to "evolutionary" algorithms
How am I ignoring this in my above example?
"So random change and selection are capable of design" No, they aren't. Only when guided by an overseeing process that limits the choice available drastically (i.e. it only works when designed).
So what was the overseeing process that "limits the choice available drastically" in the example I gave above?
This is an ID position. The funny thing is that you still want to hang on to RM/NS as the primary mechanism of change when there is no reason
There is a reason to "hang on" to the mechanisms of RM and NS. First they are known to occur, second the algorithm is known to design.
ID'ers are not so constrained.
I disagree. They seem contrained against a random mechanism being a key player in evolution. I suspect this is because they are not content with a designer that sets the first lifeform up and lets the ball roll from there.
In addition, if you grant the special creation of an individual life form, there is no experimental reason not to grant multiple special creations, even if you believe that they all occurred at the single-celled level. In fact, such an idea is precisely what Doolittle and others are working on, and even they are working from an entirely atheistic origin-of-life scenario.
Sure the possibilities are endless
In your soduku example, you were exactly as constrained as I indicated. Now, I am very ignorant of how Soduku works, but let me point out the following:
1) All of your changes followed the rules of Soduku. Did you also consider changes that violated the rules of Soduku? This greatly limits the set of possible changes.
2) None of the fitness jumps required a search of more than 500 bits.
Searches do work for limited search steps. They do not work when a large step is required to maintain fitness. An organism cannot survive the few million years it would take to find the appropriate adaptation.
"I disagree. They seem contrained against a random mechanism being a key player in evolution. I suspect this is because they are not content with a designer that sets the first lifeform up and lets the ball roll from there."
Wells would not be content, but Dembski certainly would, as would Denton.
"There is a reason to "hang on" to the mechanisms of RM and NS. First they are known to occur, second the algorithm is known to design."
Are the mutations really random? We now know that many mutations in bacteria are caused because the bacteria decided it needed the mutation. It even has special DNA polymerases to cause specific kinds of mutations, whose use is regulated.
Please point me to a research article that shows (a) a beneficial mutation, and (b) shows that the mutation is in fact random, for any common-sense or mathematical definition of the word.