27th October 2003 10:36 UTC
Just a mindproblem
How in the world can a computer pick a completely random number, since they only work with 1's and 0's? I can't get it out of my head the last few days :p.
Archive: Just a mindproblem
Deamon
27th October 2003 10:36 UTC
Just a mindproblem
How in the world can a computer pick a completely random number, since they only work with 1's and 0's? I can't get it out of my head the last few days :p.
UnConeD
27th October 2003 10:55 UTC
They don't.
All random number generators start with a seed, and the seed is supposed to be completely random. They then generate a series of numbers that is as unpredictable as possible based on this seed.
Some generators work with an entropy pool, where they need a constant supply of random input and generate a series of unpredictable, random numbers out of it. When the pool is empty, it is deemed that the numbers produced are not random enough, and the generator will stop until more entropy is introduced.
Some common examples of seeds are the current time, a collection of mouse movements, timings between keystrokes, etc.
There are basically two classes of generators: simple and secure ones. For everyday stuff (games, AVS, ...) you don't need real randomness, just apparent randomness. For cryptographical applications, much stronger generators are required with proven mathematical qualities.
Using the current time as the seed is fine for common programs, but I remember reading about a security application getting cracked because it was run at the same time every day automatically (and was using the time as a seed), and thus the cracker could predict what 'random' numbers it would generate.
If you're interested in this sort of stuff, I recommend the book 'Applied Cryptography' by Bruce Schneier which you can get at any scientific book store (.tar.gz).
Magic.X
27th October 2003 10:56 UTC
Simple: they can't.
They pick a start value (somehow use the current cpu time) and use a weird algorithm to change this value each time.
This is not really close to the point nor technically detailed, but it works this way.
[Edit] Wow, UnConeD was faster than me. :p so forget about this and read above.
Deamon
27th October 2003 12:29 UTC
Thanks, that clears things up :)
BigBob
31st October 2003 04:37 UTC
If go here: http://computer.howstuffworks.com/question697.htm,
you can see how the seed is computated and stuff. It's pretty cool.
UnConeD
31st October 2003 11:59 UTC
What a horrible explanation. It doesn't even cover cryptographical random number generators, and claims date/time is a good seed :weird:.
Oh and check your links when posting.
jonasc1
31st October 2003 13:38 UTC
I read somwhere in a magazine that the only way to generate complete randomness is to have a radioactive atom and have some devices that is monitoring the `breakdown` of it cus a radioactivity is random.
anubis2003
31st October 2003 20:36 UTC
radioactivity is apparently random, but it still follows laws of physics(we just can't figure out what these laws are) - I don't believe that anything is perfectly random, personally.
UnConeD
31st October 2003 21:33 UTC
Err you've heard of the Heisenberg uncertainty principle right? Electron density around atoms?
Deamon
31st October 2003 22:47 UTC
No I haven't :), but I would be delighted to hear it. It's some time again since I read some stuff with some depth.
Deamon
4th November 2003 21:02 UTC
UnconeD, I can't open the file. First winzip opens it, after that it's a distorted text file. You have it in any format?
UnConeD
4th November 2003 23:12 UTC
Save as, ungzip, untar (winzip should do this).
dirkdeftly
5th November 2003 01:12 UTC
the heisenberg uncertainty principle just states that we can't predict activity, doesn't it? doesn't mean it's completely random, just that we can't observe what it does without altering it.
anubis2003
5th November 2003 02:15 UTC
I don't believe the damn heisenberg uncertainty principle. I believe that things have to follow some sort of physics - uncertainty doesn't work, because physics are things that you can be certain of. I am waiting for someone to prove it wrong - it'll happen.
shreyas_potnis
5th November 2003 12:02 UTC
Can our brain generate random numbers? Now suppoose 98, I just cant figure any way how I typed the number. Or blindly punching keys on the keyboard would also generate random numbers.
Deamon
5th November 2003 12:13 UTC
I guess the brain can, also by punching keys on the keyboard blindly like "3475408674087590827". Though we don't know, we don't know how the brain works.
UnConeD
5th November 2003 13:00 UTC
We do know how the brain works. It's a bunch of neurons, which have been studied and modelled extensively.
We don't know how consciousness arises from it, though in fact I personally think it's a non-issue. People seem to have a problem with the fact that our brain is just a big pattern recognition machine, with some pre-programmed functions through evolution.
There's an easy test to show that a human child does not have self-consciousness before a certain age (I think it's about 1.5 years). Draw a red dot on their forehead and put them in front of a mirror.
Before a certain age, children will look behind the mirror, looking for that 'other kid'. From a certain age, they will realize it's their reflection, and begin to touch the red dot on their forehead.
However, humans are a lot less unique than you may think. Our dog for example has extensive knowledge of our family's behaviour. When my parents are going on a vacation, our dog will be sad and depressed at least a day in advance (before any actual packing has been done). Simply because my parents will doing different, specific things and will be running around more nervously.
I remember a backwards, ultra-religious teacher in highschool who was positive animals couldn't 'think'. I'm pretty sure they can, just that humans have more efficient and more advanced structures formed in their brain.
Modern science has shown that intelligence should be seen in a much broader scope. For example a single ant is not very smart. But the whole ant-colony is an efficient food-searching agent, showing adaptability and optimization behaviour.
When a person types in a random number, I think the process should not be seen as any more different from a computer. Except that a human has a whole lifetime of experience-entropy to use as a seed ;).
Tuggummi
6th November 2003 07:03 UTC
My brain can't create random numbers... when ever i think about random/irregular numbers i always use 1,3,7 & 9 :igor: and "regular" numbers for me are 2,4,5,6 & 8.
Deamon
6th November 2003 09:08 UTC
And how do you explain blindly punching numbers on a keyboard? Is there something in our brains which calculates every punch, and therefor might know what number will exactly be on the screen, though our conscious being doesn't, and therefor the number seems random?
Tuggummi
6th November 2003 09:56 UTC
It depends, how well you know your keyboard? I know that closing my eyes doesn't make me forget the position of the keys. But sure if you just punch the keys like a monkey will result in random text.
But i think i didn't express myself quite clearly in the first post, i'll take an avs example to show what i mean :)
xts = (getosc(0.371,0.1973,1.37)*3791%200-100) * speed
Now what i ment that this piece of a beat detection code looks/seems more random to me than for example:
xts = (getosc(0.2,0.4,0.8)*4000%200-100) * speed
I just ment that how i look at the numbers and think about what's random and what's not. Even numbers seem more regular to me than odd numbers :igor:
It's just a matter of perspective.
Deamon
6th November 2003 11:23 UTC
true. I know the keyboard very well, I won't forget the position of keys when closing my eyes. Though I indeed ment punching like a monkey ;).
shreyas_potnis
6th November 2003 13:54 UTC
Ok, today I conducted short survey to study whether someone could generate seemngly random numbers without halting to think - and i think most of them cant. Almost all of them started with some and the first four to five very very random bu they were followed by consecutive numbers or multiples of 5 or 3.
eg - 5,9,5,8,9,10,12...
98,34,95,100,105,1000...
You too can try this.
sidd
6th November 2003 14:54 UTC
when you punch stuf in a keyboard, your brain is probably going something like this:
"*punch* ok, ive punch once.. now, the next one should be randomish, so ill punch a bit over too the left *punch*, and now over to the right *punch* but now if i punch to the left again, it will seem like a pattern, so ill punch right again, *punch* etc etc"
This is just the same as when you try to cosnciously think of random numbers (as shreyas demonstrated), most people will conciously try to make the selections seem random. The actual selections made by a person depends on their experiences and their genetics.
Because the combination of genetic complexion and the experiences a person has throughout life, it is almost impossible to predict which numbers a person would select. But, if you could somehow know everything about this person, then you could predict.
So its not random.
dirkdeftly
6th November 2003 15:40 UTC
and even if you are typing "completely blindly", your brain is still conciously moving your fingers to a specific place all the time. blind typing is still non-random, it's just a much more complex system. input goes in to your body through four external senses, is processed by four internal senses, and whatever is at the front of your mind at any particular moment affects where your fingers and arms go.
anubis2003
6th November 2003 16:17 UTC
Which is why someone who figures out the physics of the mind would be insanely powerful...
13373571
8th November 2003 05:46 UTC
uhh... they would?
That's like saying that einstein could time travel because he figured out relativity, isn't it?
anubis2003
8th November 2003 06:20 UTC
no...
If someone figured out how the mind worked to a pinpoint, then he'd be able to figure out what they would do given certain cirumstances - and with that knowledge they could easily discover anything they wanted...
sidd
8th November 2003 23:31 UTC
That arguement, like the "mary" knowledge arguemnt, relies on assumptions of what "all the physical knowledge involving the human mind" would include. For example, the human mind might have some mechanism that prevents itself from accessing certain sencial elements, so the person wouldnt have the full picture.
Also, the brain can undoubtably only process a given amount of information at once. It is safe to assume that if we are processing that information, and then try to think about the information, thus two-folding all your thinkingness, your brain would system crash in a similar way to the old-school avs-variable-limit-death..
UnConeD
8th November 2003 23:58 UTC
If someone figured out how the mind worked to a pinpoint, then he'd be able to figure out what they would do given certain cirumstances - and with that knowledge they could easily discover anything they wanted...Err no. Just because someone knows how the mind works doesn't mean we can 'scan' someone's brain and construct an identical, working model of it to predict thought.
Fork me on GitHub