And you'll note that I didn't say that he did. What I said was the Knuth gives evidence than randomizaton can be evidence for design.
No he didn't, he stated that certain algorithms perform better with randomisation. Things like QuickSort, for example, have worst-case behaviour on sorted input. By adding randomization to the input, you make it less probable that the input to the quicksort is presorted. This does not in any way imply that randomness is a sign of intelligence, merely that intelligence is capable of using randomness.
Suppose you were inside the computer, a la Tron, and were observing the behavior of quicksort. Would there be any scientific way to tell one way or another whether the process was designed or not?
Just curious, but is science the only guide to true knowledge?
What is 'true knowledge'? You're straying way away from science now and into philosophy.
These days it seems there's a fine line between the two.
Science doesn't claim to be true, it claims to be useful, which is far easier to test.
First, if that's true, then why do a number of those who reply make the claim that there is no evidence for God?
Because they are not scientists, they are neoathiests who use science as a surrogate religion, and choose to delegate their thinking to scientists rather than priests.
Then, circling around to the reason for this post, in order for science to be better accepted, scientists need to do a better job of denouncing the neoatheists, just like Christians need to do a better job of, say, denouncing the Westboro Baptist fruitcakes.
Second, it seems to me that you're in an epistemological catch-22. Using the example of randomization, it can be evidence for sight and blindness. Why, then, is evolution "blind"? This is a metaphysical statement, not a scientific one.
No, it is evidence of nothing more than the fact that God is not required for the theory to work. Evolution guided by random mutation, evolution guided by God and evolution guided by pixies all produce the same measurable results.
I'm not sure I buy that. Take Dawkin's Weasel program, for example. Yes, it's not an example of evolution, since it embeds information about the target in the search. Nevertheless, if the mutation rate, max number of children per generation, and max generations aren't chosen correctly, the target string isn't found. The parameters need to be "fine tuned" for it to work. Please correct me if I'm wrong, but I don't think biologists have address the issue of tuning.