What is Digg?Sponsored by FUNimation Enertainment
Watch the First 8 Minutes of Evangelion 1.01 Movie -- Anime view!
funimation.com - Available now on DVD. The rebuild of the ground breaking anime Neon Genesis Evangelion.
59 Comments
- revmitcz, on 10/12/2007, -1/+14Well, no, that's actually the title of the article.
It's just a shame the description doesn't explain that the article is more about "what is 'random' anyway?" as it relates to computers and a couple of references to the iPod (more because it's a nice, ubiquitous, modern-day example) than it is an actual answer to the question posed in the title.
That having been said - I enjoyed the article. Informative and short. I hope to see his findings but I read something like this about an experiment done years ago. So, I'd like to compare the old to the new and also see how this number-cruncher's experiments vary. - crammaz, on 10/12/2007, -0/+10I have always been under the impression that by definition a machine cannot be programmed to be random...
- richbradshaw, on 10/12/2007, -0/+10Yeah, but over an infinite time, you would expect each song to have been played an equal amount of times - if over a very large amount of plays, track one was played twice as much as any other track, you would conclude that the series isn't random, but biased towards track 1.
If each track is played the same amount of times over a huge number of plays then it is impossible to guess which track will come next - hence it is random.
If track 1is played more than any other over a huge number of plays, we can guess that track 1 will be next, with a higher than 1/10 probablilty - hence it is not completly random. - MrBabyMan, on 10/12/2007, -1/+11It would seem that it would be too complex and pricey if the iPod had a true RNG incorporated into it.
from Wikipedia
"A physical random number generator is based on an essentially random atomic or subatomic physical phenomenon which are aspects of quantum mechanics. Examples of such phenomena include radioactive decay, thermal noise, shot noise and clock drift."
It's probably some fraction of a percent shy from 100% random, which is, honestly, close enough for a gym workout or a long car trip. - gargantuan, on 10/12/2007, -1/+11don't be an *****. If computer science is your thing then yes, it's relevant, but if you collect miniature figurines of dancing kittens for a hobby, then I suspect it's not relevant.
- noGoodNamesLeft, on 10/12/2007, -2/+11ARTICLE: "But first, it's worth pinning down just what makes a string of numbers random. The strings we're talking about generally consist of whole numbers covering a certain range -- say, if you have 10 songs in your iPod, numbered one to 10. A random sequence must contain each number in equal frequency"
Now, I'm not a mathematician, but this sounds like BS to me. Correct me if I'm wrong but...
A *truly* random sequence should be totally unpredictable. That's the only criteria. "1 2 3 4 5" should be equally as likely as "5 4 2 1 3" (although in general an unordered result is more likely simply because there are more possible unordered combinations).
Of course, we have to define *what* we want to be random; e.g. we could say "Pick any song at random, then pick another song FROM THOSE THAT HAVEN'T BEEN PLAYED ALREADY at random, then pick....". But this itself isn't part of randomness. If your algorithm is
while (true) { play_any_random_selection_from_list(); }
then the "random" choice is BY DEFINITION *not* required to have a particular frequency; else it's not truly random.
Really, regardless of whether the iPod has a good random number generator or not, the fact is that people *do not* want a truly random sequence of songs, although they think they do. Most truly randomly-generated sequences will contain what *appear* to be patterns of numbers/choices, but this is purely because out of all possible sequences there are more that contain pseudo-"patterns" than don't; laws of probability. It's human nature to recognise "patterns", where there aren't any (i.e. they aren't predictable).
In short, people don't want a random song selection; they want what *sounds* like a random selection that avoids these "patterns", and THAT is problem. - losboccacc, on 10/12/2007, -0/+8it's called the cd player random alghoritm (subtitled: put the h in place)
you do not want a truly random alghoritm, but one that plays all songs once in random order, avoiding repetition, wich is not fully random (or it is random, but not the way the user wants it). - wastern, on 10/12/2007, -0/+8the shuffle feature in iTunes was truly random, but then people said it "wasn't random enough" so they added a slider to make it "more random" to the end user by mixing up the randomness even more so now you can select if you want it more or less random...if that concept even make sense
is this going to have to happen with the iPod now too? - rodrigo74, on 10/12/2007, -0/+6Well, I guess it depends on what you define as a computer, for example if you just happen to have a radioactive piece of rock inside it, and some sensor to measure its decay and then send it to your random generator code, then yeah you could say your computer can come up with truly random numbers.
- giantrobot, on 10/12/2007, -2/+7Re: noGoodNamesLeft:
"ARTICLE: 'But first, it's worth pinning down just what makes a string of numbers random. The strings we're talking about generally consist of whole numbers covering a certain range -- say, if you have 10 songs in your iPod, numbered one to 10. A random sequence must contain each number in equal frequency'
"Now, I'm not a mathematician, but this sounds like BS to me."
I am a mathematician, and I say that you are right. A random sequence does NOT need to have each option in equal frequency.
Get a coin out of your pocket. Flip it 5 times. If you get H,H,H,H,H, does that mean it is not a random sequence because it doesn't contain as many heads as tails? Absolutely not. Does it mean that the next time you flip it five times you will get T,T,T,T,T just to assure "randomness"? No.
The "frequency" of each number is NOT the "probability" of getting each number on a flip.
Mathematical ignorance drives me nuts sometimes. - Otto, on 10/12/2007, -0/+5This sort of thing has been debated for a long time. Example: http://msnbc.msn.com/id/6854309/site/newsweek/
The short of it is that the iPod actually has a pretty good pseudo-RNG in it. It's just that people are not capable of recognizing random when they see it.
People are very, very good pattern recognition devices. They see patterns in noise. They also interpret patterns as "not random", and see patterns without also seeing the lack of patterns. In other words, when you notice a pattern in the ordering of songs, you remember that. But when there is no pattern that you notice, you fail to remember that. So if it plays three songs by the same artist in a row, you remember that, but don't remember the 50-60 times it doesn't do that in the same playlist.
This sort of thing also explains all religion, BTW. - inkswamp, on 10/12/2007, -0/+5You are correct. Traditional computing devices (i.e., PCs) can only generate pseudo-random numbers. The article touches on that fact.
"...though I did learn that any computer algorithm for randomizing information can never be truly random." - paulmdx, on 10/12/2007, -1/+6It's not whether it's true RNG or not, it's whether it's a good or bad pseudo-RNG. Of course we're not expecting true RNG.
You can get very good pseudo-RNGs (google twister rng), but judging from the randomness of my iPod Video, I don't think they use a very good one. :-) - Orbatos, on 10/12/2007, -0/+4Your concept of "truly random" with regards to computing needs work. The most than can be hoped for is "highly random" numbers seeded by available data that changes in a manner unconnected to the algorythm being used. (ex. time, entropy pools). Essentially your slider is limiting this sort of action.
This type of analysis is possibly exactly because of the type of system I have outlined, so it is indeed possible to determine the quality of randomness on a such a simple device. - manageMyRights, on 10/12/2007, -0/+4Think about a fair die. If you throw it enough times each side will come up the same amount. Of course Bayesians have problems with this Frequentest definition because it could take an infinite number of throws to actually see this behavior and you can't really know when you've thrown it enough times.
- Tonamel, on 10/12/2007, -0/+41) Open iTunes
2) Turn on shuffle
3) Start a song, then click "next" a few times, and note each song
4) Repeat step 3, starting with the same song
5) Notice that you get the same songs in the same order
6) Deduce that shuffle isn't terribly random
No, I don't have the randomness turned down. - starsky51, on 10/12/2007, -0/+3Intel solved the random number issue by using a naturally occurring, unpredictable phenomenon.
http://www.microsoft.com/mind/0399/flux/flux0399.asp
"The random number generator integrated in Pentium III-based systems uses thermal noise from a resistor to produce a random number that actually is quite random." - BitwiseMcgee, on 10/12/2007, -0/+3losboccacc is right, audio players do NOT want a real random sequence, because if it is truly random you most likely will get repetitions of songs.
Of course, when a person hear's a song again when on shuffle they immediatly assume the player's rng sucks. This is not the case. When you select 'shuffle' a good player will randomize the playlist first, then play through, thus avoiding repeats - bluesdealer, on 10/12/2007, -0/+3"The general plan would be to record brief sound files consisting of a single tone -- A minor, G sharp."
Doh! A minor is a chord, not a tone. *hits head on keyboard*
Very interesting article, though. - bluesdealer, on 10/12/2007, -0/+3Relevance doesn't matter. This is just interesting science, pure and simple. No one is going to say, "OMG I can't buy an iPod because its RNG sucks!" but it's still an interesting test subject for this experiment.
- losboccacc, on 10/12/2007, -1/+4they cannot be random as they're programmed with predictable random alghoritm, but computer can read from random sources...
- Otto, on 10/12/2007, -0/+2It wouldn't help. People would still complain that it was not random enough. The problem is that people don't want random, they want to not notice any sort of patterns. The problem with this is it's so far from random that you can't actually call it random.
Here's one that would satisfy most people:
-Never play the same artist/album/song twice in a row.
-Always pick one of the least played songs, +/- 10%.
-Never play anything with a BPM close to the last song by 10%.
-Satisfy the conditions in that order, since there will be times that not all conditions can be satisfied.
Most people would say that resulting playlist was pretty random. Except it really isn't, is it? - vlurk, on 10/12/2007, -0/+2Or you could also buy a Quantis: http://www.idquantique.com/products/quantis.htm
This is a PCI card with a quantum random number generator, with drivers for Windows, Linux, BSD and Solaris. - Otto, on 10/12/2007, -0/+2My biggest problem with you and people life you is that you all seemingly fail to understand that:
a) You're wrong and it really doesn't do what you think it does (and this can easily be *proven*),
b) There are psychological reasons why you think it's behaving in ways that it actually is not, and
c) You think that being told you're wrong is somehow condescending.
You're not stupid. But you are still wrong. The two are not equal.
Nobody can design a RNG that behaves the way you want it to behave, because you cannot define how you want it to behave in a way that fits the definition of the word "random". If you don't want it repeating songs or having the same artist back to back, well, that's not *RANDOM* at all. - richbradshaw, on 10/12/2007, -1/+3This is the most random thing I have seen on digg all day.
- hongy_r, on 10/12/2007, -0/+1i think most of you are idiots. obviously, nobody is going to draw lottery numbers using an ipod, so is such a minute level of accuracy really all that necessary? sure, it is impossible to draw a *completely* random number out of a computer, but just so long as i dont keep hearing the same songs in a row all the time, im happy with my ipod.
p.s check this out. it is quite interesting.
http://computer.howstuffworks.com/question697.htm - matrixneo, on 10/12/2007, -0/+1No. Machines can be programmed to be random. All you have to do is, enable the randomize feature.
http://technogroove.blogspot.com/2006/11/ipod-shuffle-features-that-makes-it.html - SweetChuck, on 10/12/2007, -0/+1It's not. the A-minor chord above middle C is a combination of tones at 440 Hz, 523.251 Hz, and 659.255 Hz. In general the ratio of frequencies for a minor chord are around 10:12:15
- shadownight, on 10/12/2007, -1/+2I think the random feature of iTunes is perfectly random, but when I shuffle on my iPod, I do get consistently a weird "random" order.
- cbiz, on 10/12/2007, -0/+1As far as I'm concerned if you put Steely Dan on an iPod it will shuffle those into a play mode more often. Oh, maybe it's because I have EVERY Steely Dan Song made on my iPod + dupes...
- bluesdealer, on 10/12/2007, -0/+1Oops. I wish I had seen your post before posting my frustrations below ;-) Yeah, for all his knowledge of math theory, he doesn't understand the difference between a note and a chord. :-D
Oh, and Bach would disagree with you, Chuck. The note "A" hasn't always been 440 Hz. ;-) - CarzorStelatis, on 10/12/2007, -1/+2Does anyone actually care? A shuffle feature that produces a good, listenable playlist is perfectly fine for 99.99% of users who want their iPod for listening to music, not testing theories on the accuracy of computer random number generators.
- bluesdealer, on 10/12/2007, -0/+1You're missing the point. It's about the science, not the product.
- shockingbird, on 10/12/2007, -1/+2Dear people who make music players,
How about this: Instead of random. Never play the same artist or album within about 10, 20, or 30 songs of each other. You just played some Journey? Great. Don't play any more Journey, especially from the same album for a predetermined number of songs.
Bam.
The music is shuffled properly. - jvicinanza, on 10/12/2007, -0/+1Now thats a cool mod id like to see. An ipod containing a small piece of Krypton-85 installed in order to drive the shuffle function. That would be cool... any takers???
- davestar, on 10/12/2007, -0/+1@giantrobot
ok, yeah - the author of the article is not writing with mathematical exactness, but you should realize what he's trying to say. the implication is that if infinite trials of a random event are carried out, the possible outcomes should occur with equal frequency.
"Get a coin out of your pocket. Flip it 5 times. If you get H,H,H,H,H, does that mean it is not a random sequence because it doesn't contain as many heads as tails? Absolutely not. Does it mean that the next time you flip it five times you will get T,T,T,T,T just to assure "randomness"? No.
The "frequency" of each number is NOT the "probability" of getting each number on a flip.
Mathematical ignorance drives me nuts sometimes." - shockingbird, on 10/12/2007, -0/+1My results are exactly the same as yours.
- TheDeuce, on 10/12/2007, -1/+1note this function is called "shuffle" and not "random", so wouldn't expect it to be truly random.
- telero, on 10/12/2007, -0/+0The iPod Shuffle isn't supposed to be random, if it was, they'd call it the iPod Random.
It's kind of like single deck blackjack. They shuffle cards and deal them out one at a time until they're gone (ignore the fact that the deck gets reshuffled before you actually run out of cards). Then they shuffle them and start dealing again. Once you've seen the Ace of Spades once, you don't see it until the decks been shuffled again.
All the iPod is supposed to do is take your playlist, shuffle it, and play the songs one at a time until the list is done. Then it shuffles them again and continues playing. It does exactly what the name implies. - ctiptonk, on 10/12/2007, -0/+0This is why iTunes has a "more random/less random" slider in the settings... Steve announced it with great fanfare along with iTunes 6.
- CoolWind, on 10/12/2007, -1/+1It's headline news when somebody (actually a nobody) announces they intend to do something?
I mean couldn't they wait a little while until the experiment was actually conducted and the results were comprehended before they wrote the article? - BigEazy, on 10/12/2007, -1/+1ALmost fell asleep reading that article.
- emallove, on 10/14/2008, -0/+0I viewed my iPod songs by "Play Count" and noticed that many songs were getting skipped over by iPod's so-called random shuffle mode. I then created a Smart Playlist titled "Seldom Played", which contains songs that are below a certain Play Count threshold. I queue up the "Seldom Played" playlist to hear all the songs that shuffle mode has been skipping over.
- bluesdealer, on 10/12/2007, -1/+1"This sort of thing also explains all religion, BTW."
Not really. Humans forgetting randomness aside, organic life and an ecosystem are pretty significant patterns to chalk up to randomness. Also, even though the subatomic world may appear fully random, it becomes order in the macro-sense. Maybe you're not the religious type, but that doesn't mean that there wasn't some creative force involved in the forming of our universe. (BTW, the universe's age cannot be infinite because of the law of conservation of matter and energy. If the universe had always existed, therefore allowing for infite randomness, it would be stagnant and no energy could be transfered any longer.) - jimmmm, on 10/11/2007, -0/+0While on shuffle my ipod, which has a couple thousand songs, played the entire album, "Some Girls" by the Rolling Stones, in order. That's definitely not random
- marker7799, on 10/12/2007, -1/+0yeah who really cares how random it is...
- grow, on 10/12/2007, -1/+0It's random enough for me to not know what's coming up. That's good enough for me.
- bobalien, on 10/12/2007, -1/+0what a waste of time
- guinsu, on 10/12/2007, -2/+1My biggest problem with the Ipod (and every other mp3 player I have owned) and its random is that the same songs come up constantly. And really, its pretty condescending when tons of users complain that they are unhappy with the random and they are told to STFU, its really random and they are too stupid to see it. Why don't the people at Apple or elsewhere design a RNG that behaves the way the users EXPECT it to?
- CDRaff, on 10/12/2007, -7/+5Computers CANNOT Do random
the can only emmulate Random
See Steve Gibsons Password Page on GRC -
Show 51 - 59 of 59 discussions
© Digg Inc. 2009
— Content created and posted by Digg users is