Perhaps the software industry will eventually look a bit like a highway. The infrastructure (operating systems, networking technologies) will be largely a public good, while services (support, training) and specialised applications are for sale.
I find it alarming that this idea seems to become more and more popular. Why is it that Free Software should be delegated to infrastructure? If Free Software is better because of the peer review process and more moral because it encourages sharing, then this is true for all programs, not just for operating systems. It's just that historically, operating systems, networking and systems utilities were the first free programs. There is absolutely no reason why it should stop there. I for one sure hope that it does not.
The Turing Test is inadequate since it really only measures how well the machine simulates the input-output behavior of a human. That does not have much to do with intelligence; hyperintelligent aliens for example could not pass the Turing Test. Instead, we should try to build machines which can pass the Boldt Test: their input-ouput behavior should be as interesting as that of the average human. A conversation is defined to be more interesting than another if a human judge prefers to continue this conversation over the other.
I've written up this proposal in a bit more detail here.
To quote Bruce Schneier (from the book I have at hand): "A nondeterministic Turing machine [works by] trying all guesses in parallel and checking the guess in polynomial time" (Applied Cryptography, second edition).
Nondeterministic Turing machines have nothing to do with polynomial time. He was probably confused about NP problems, like pretty much every layman.
Guess I'm not at fault and shouldnt be punished for murder? Sillyness.
You are a highly complex probabilistic information processor. Modern science does not allow for anything as foggy as "free will".
That doesn't mean that punishments don't make sense however. A probabilistic information processor, if informed that certain actions may result in unpleasant consequences, will perform those actions with a lower probability. Just like an intelligent but deterministic chess computer, once it figures out that giving up the queen is usually detrimental, will refrain from doing so.
All the talk about "responsibility", "guilt and innocence", "free will" etc. is utterly meaningless; it's just a different way of telling those chess computers that they are going to lose if they make a mistake.
Slashdot Mirror
I find it alarming that this idea seems to become more and more popular. Why is it that Free Software should be delegated to infrastructure? If Free Software is better because of the peer review process and more moral because it encourages sharing, then this is true for all programs, not just for operating systems. It's just that historically, operating systems, networking and systems utilities were the first free programs. There is absolutely no reason why it should stop there. I for one sure hope that it does not.
--
--
--
I've written up this proposal in a bit more detail here.
--
Nondeterministic Turing machines have nothing to do with polynomial time. He was probably confused about NP problems, like pretty much every layman.
--
Guess I'm not at fault and shouldnt be punished for murder? Sillyness.
You are a highly complex probabilistic information processor. Modern science does not allow for anything as foggy as "free will".
That doesn't mean that punishments don't make sense however. A probabilistic information processor, if informed that certain actions may result in unpleasant consequences, will perform those actions with a lower probability. Just like an intelligent but deterministic chess computer, once it figures out that giving up the queen is usually detrimental, will refrain from doing so.
All the talk about "responsibility", "guilt and innocence", "free will" etc. is utterly meaningless; it's just a different way of telling those chess computers that they are going to lose if they make a mistake.
--