Re: Implausibility continued (in C)

On 18 Sep, 02:20, Garamond Lethe <cartographi...@xxxxxxxxx> wrote:
On Mon, 17 Sep 2007 05:31:41 -0700, someone3 wrote:

<snip>





Regarding the program I supplied:

-------------------------
variable x = random number between 1 and 1,000,000,000,000
variable y = random number between 1 and 10

while ( x > y)
{
printout(x + " is greater than " + y)
x = random number between 1 and 1,000,000,000,000
y = random number between 1 and 10
}

printout (x + " is not greater than " + y)
end.
-------------------------

It is deterministic, that was the point, as computers aren't truly
random.

Except when they take random outside events as inputs, of course. Compare
and contrast /dev/random and /dev/urandom.

Secondly the statement only talks about x not being greater
than y, thus will be true for when x = y (it will drop out of the
loop, and the printed statement will be correct).

Yep, you're right. I read that too quickly.



Regarding predicting the answers to the question you gave, it might
not stop for a million years, or might not stop, and as we don't have
an elegant means of predicting that answer and have to labouriously
work through what would happen, we have to give predictions in terms
of moves ahead.

And so the problem is as yet unexplained, and unpredictable, for the
question of infinite loops. I think we agree on this point and can move
on.

<snip>

Recall that we also can't derive a higher-level description of the Turing
program -- we've selected it because of its interesting behavior instead
of sitting down to code it to do something. Recall that neural networks
also have this property -- we're looking for some behavior for our ANN,
and need to know how to choose one out of the gazillion ones available.

How do you train your neural network? I'm not looking for the algorithm
(e.g. "backflow"). I'm looking for your fitness function. Given two
neural networks, how to you decide which is "better" for your robot
example?


No the behaviour of the computer when it runs the program is not
unexplained. As I said in the piece you snipped:
---------------
As said, at one iteration per second, we could predict 10 minutes
ahead (600 steps) for example, given our knowledge of how the program
works. The understanding (the communication of which is the
explanation) isn't specific for the 10 minute time frame, but is
the same that will allow a prediction for 30 minutes, a year etc. Are
you suggesting, that if you can make predictions in terms of a certain
amount of time ahead, even though these predictions are 100% accurate,
that you don't regard them as demonstrating an ability to predict
based on the explanation of how the thing works?
---------------

It would be useful if you didn't snip and avoid answering the
question, as it might clear up why you don't consider the program
explainable.

Also out of curiousity do you consider the behaviour of a computer
running the program I supplied unexplainable?

.

Relevant Pages

  • Re: Explaining Time Dilation
    ... > SNIP ... predictions are in better accord with the evidence. ... Relativity does not say space 'performs' any physical act - it says ... merely that they are in accord with experiment - which both SR and GR are. ...
    (sci.physics.relativity)
  • Re: Some fancy metalworking
    ... As an update for the people that were waiting for the "black hole ... Predictions of the world?s end are nothing new though. ... and wisdom and counsel shall not alter them to the better, ...
    (rec.crafts.metalworking)
  • Re: Acceptance and growth
    ... won't be expecting His return. ... is that in regard to modern "prophecy", hardly one of our self styled ... such predictions exist; considering how much ...
    (soc.religion.christian)
  • Re: Implausibility continued (in C)
    ... Regarding predicting the answers to the question you gave, ... Recall that we also can't derive a higher-level description of the Turing ... neural networks, how to you decide which is "better" for your robot ...
    (talk.origins)
  • Re: Upended quantum physics in the news
    ... You're the one in this thread making conclusions, ... predictions can be useful and they are certainly ... calculate from first principles all the possible phenomena ...
    (talk.origins)