Slashdot Mirror
Is the iPod Shuffle Playing Favorites?
marksilverman writes "Steven Levy at Newsweek is reporting that his iPod Shuffle seems to favor certain songs. Is Apple receiving kickbacks to promote certain artists? Apple denies it, of course, and Levy had the good sense to ask a mathmatician and a cryptographer who explained that it's probably just humans finding patterns where there are none." Less neurotically, both CNet and PCWorld have discussions of the Shuffle's interior spaces.
... A bad random number generator. It happens.
From the article:
I explained this phenomenon to Temple University prof John Allen Paulos, an expert in applying mathematical theory to everyday life. His conclusion: it's entirely possible that nothing at all is amiss with the shuffle function
The slashdot article??
Is the iPod Shuffle Playing Favorites?
and
Apple denies it, of course
Enough with the inflammatory headlines!
Quidquid latine dictum sit, altum viditur
This post brought to you by Teenage Angst and Caffeine!
Elmo knows where you live!
The summary suggest that Apple may be playing favorites, citing an article that concludes pretty definitely that they are not...
I think the point of the story is not just to report "news" but open up the topic for discussion. I am sure making the front page will cause at least a few more people to investigate the claim, or at the least monitor the behavior more closely of their shuffle.
Saying Java is nice because it works on all OS's is like saying that anal sex is nice because it works on all genders.
Seriously. There have got to be better submissions in the queue than, "I think my iPod shuffle is preferring certain songs over others. It's a conspiracy!" Come on.
Go ahead and call me a dumbass. I deserve it for commenting without reading the article, which turns out to be very interesting. Please ignore my stupid post.
So, I know its us the loyal readership that is supposed to do the editing essentially (which makes me think we should have a moderation-style system for voting on which stories in the queue should go live (editors, you reading this?!!) but hey!) but when our favorite web tool has spell-checking built in, and you can search up to 32 words at a time.... I mean you might as well just cut and paste.
What's really interesting, that even though this is a geek site, we can't even get mathematician right. Even more silly is if you check the link above from google, you'll see that the first two results (of 3) are also from right here at the dot.
All I'm saying is: if it's power to the nerd masses, let's do it. Many posts so far are already complaining about the story. Not to mention it's from MSNBC. Not to mention I've already read it, because it's from almost four weeks ago.
I realize I am off-topic and complaining, but I wanted to see if we couldn't get a discussion going about a smarter, more democratic way to elect submissions to go live.
Those of you that think that sounds completely plausable, please step to the left.
Look up the word "payola" if you think that record execs wouldn't do this.
It doesn't need to be in the shuffle. It could be part of the AutoFill in iTunes.
Do I think it's really happening? Not really. Do I think it's plausible? Yeah... experience with radio tells me that record companies would do this if Apple let them.. and who is Apple to turn down money?
Sure, apple is spending R&D costs on creating a way to get their audio player to promote songs that users have already bought. Thats makes a ton of sense.
Think of it this way:
By supporting Apple, you're also supporting R & D for much of the entire personal computing industry.
It's not offtopic, dumbass. It's orthogonal.
You would know by analyzing the algorithm which produced the numbers, not by analyzing the output. Of course, the fact that the numbers are produced by an algorithm proves that they are not random. If you ran the same algorithm again, you'd get the same numbers. Since computation is deterministic by design, there is no purely computational way to generate truly random numbers.
If you're not a programmer, try to imagine writing out a mathematical equation which, when evaluated, comes out to a random number, different every time. It doesn't make sense, does it? How could the same equation have a different result every time, without changing the inputs?
This absolutely right. Also, using this method, inevitably some songs will occur multiple times in the playlist, since you're not keeping track of dupes (much like slashdot).
If you need to randomize an array, the easiest way to do it is to assign each item a random number, and then sort the array using the random numbers as a key. That way every item occurs only once in the randomized list.
You still need a decent pseudo random number generator of course, if you're using a pseudo random number generator that only produces 5 discrete values and repeats them over and over it doesn't help much.
On the other hand, for an application like a playlist shuffle, you don't need a cryptographically secure RNG, just a PRNG (such as a Mersenne Twister), that uses the current time in milliseconds as a seed would do nicely.
SCO employee? Check out the bounty
This is not a problem of a bad shuffle algorithm. The problem here is thousands of years old. It is human nature. People see patterns where there are none. People generate theories based on these non-existent patterns. This is how people concluded that the sun orbits the Earth: "Oh, look, the sun looks like it is orbiting the Earth! Therefore it must be!"
The problem is, if you just put together a playlist with a bunch of artists and play it, it is entirely likely that someone will be played three times in the first hour. And in this guy's case, that someone was Steely Dan. So what does that prove?
About 500 years ago, we invented something called the "scientific method". Although it is taught to most people in both science and history classes, few seem to understand it.
The scientific method says that you cannot use past observations to make a conclusion. You must develop a specific test to prove or disprove your hypothesis. You must show before you perform the test that the outcome of the test is relevant to your hypothesis. You can then perform the test and use the results to back your conclusion.
The scientific method could very easily be applied here. What this guy needs to do is start with the prediction that Steely Dan or whoever will be played three times in the first hour. He must use statistics to compute the probability of this happenning in a purely random shuffle, and should show that the chances are less than 1% (this is a pretty straight-forward use of standard statistical methods). Then he should run the experiment and see what happens.
My guess? Steely Dan will not play three times in the first hour.
Without a proper scientific experiment proving this guy's theory, there is no story here.
The iPod Shuffle does not randomly play songs. It shuffles the playlist...then plays the songs in the new order.
whereas a random play function could lead to a single track being played twice as often as another, shuffle precludes that.
hence the name, rather than "iPod Random".
--
While what you describe might be what you desire, it isn't random at all. It is rules based. Which is fine if that is what you want.
Lasers Controlled Games!