How does an iPod decide which songs to play?

I normally have my iPod Touch in Shuffle mode. One thing I’ve noticed is that the same songs come up frequently in the rotation, while other songs seem never to be played. Is it just the breaks of randomness? Or are some songs weighted more heavily, such that they get played more and other songs that are not frequently played are not? (That seems strange, since it becomes a Catch 22.)

The human mind is great at finding patterns where none exist. Cf. superstitions.

Having songs repeat or omitted in a long stretch is pretty much what is supposed to happen in a truly random sequence.

There was a thread recently about possible iPod RNGs but nothing came of it.

Do you have any ratings on the songs? I don’t recall if the iPod takes that into account; it might, though, which could explain it. One thing that happens is that ‘unrated’ songs actually are considered 0-star if you have any other rated songs in the collection, at least for iTunes.

I’ve also noticed that with my IPod “shuffle” does not appear to be random. Just sayin.

It could be either a poor random number generator that has a tendency to produce values within a certain range, or it could be happenstance. It’s hard to say.

I know that letting iTunes on my computer run continuously will result in wildly disparate play counts for songs, but I haven’t run any math to determine whether this falls within the range expected by chance.

There was a short article in the latest issue of The American Statistician that investigated the claims made in Steven Levy’s 2005 Newsweek article that might suggest that the iPod shuffle prefers certain songs/albums/artists. None of the events described are at all unlikely under the assumption that songs really are shuffled at random. Rather, it appears that we have yet another case of people misunderstanding the nature of randomness.

The article is here, but you’ll either need to be an ASA member or buy access in order to get to it.

I don’t think it’s confirmation bias. I haven’t been keeping a spreadsheet, but certain songs do get played ‘all the time’ while others rarely get played.

That’s why I wrote this:


Maybe you should. Not trying to be snarky, but it would be a good experiment.

I only have I-tunes, not an Ipod, but it keeps stats on how often songs are played. If the Ipod has that also, it should be simple to dump the stats into a chi-squared test and see if you’re on the right track or not.

I can not stand the way my iPod shuffles song. I too have noticed that it seems to replay only a few and never play others.

From what I understand, I think that when you click shuffle, it generates a list of songs in random order and will go down the entire list. If you listen to the entire thing, it should play everything once.

If you turn it off, click shuffle again, skip songs, stop the random play to play a specific song etc, it generates a new list.

That’s how I remember it, I could be wrong. There’s a lot of discussion online about it.

Shuffles don’t keep track of the number of times songs get played? My two ipods do (nano and iPhone). Plug it in, open up iTunes, click on the ipod’s icon, check it’s library, and sort by “number of times played”.

Keeping a record of what’s been played is different than managing the shuffle order using number of times played as a basis.

This has been explored in the media a few times and Apple has sworn that it’s pseudo-random and makes no attempt to give preference to artists or songs. Google “ipod shuffle randomness” for lots of reading material.

**Fubaya **is correct that shuffle mode creates a playlist of all the songs on the device and plays the entire list (if you let it).

On mine, some songs I hear often, and some songs have never been played at all–and that is the nature of randomness. Sometimes randomness is not fair. :stuck_out_tongue:

Sure, but it’s close enough for government work. And we’re talking about a shuffle - that’s all it does!

Not at all.

How so? The OP said he normally has it in shuffle mode. What am I missing? No, it’s not going to be scientific. Yes, random number distribution is going to have groupings that don’t *seem *random. But it’s still going to give us a rough estimation of how “random” it is.

That would not be practical. Normally I use my Touch on the bus, and then take it off when I get to the office. Then, depending on what KEXP is playing, I’ll pop it into the radio/iPod player on my desk. At the end of the day I’ll listen to it as I’m going to the bus and riding to my car. To make a spreadsheet I’d have to go back to the beginning and see what’s played, and then enter the data. I’m usually very busy at work, and I usually don’t have time to enter a single line on my Prius mileage spreadsheet; let alone a large number of songs. Also, I power off before the weekend and before telecommuting days; so the random playlist is lost.

Or you could just reset all of your songs to “never been played” and exclusively shuffle (rather than play specific songs/albums).

A while back, Apple actually added a feature to iTunes that made it “more random.” The irony is that using that setting is actually LESS random because iTunes makes an effort to weight songs that haven’t been played for a while and to avoid playing songs from the same artist/album twice in a row.

It’s all just another example of the fact that what statisticians mean by “random” and what the average guy on the street means by “random” are two totally different things.

With about 6000 songs in my iPod I never made it through all of them – a direct consequence of wanting to re-sync and get the new songs I seem to regularly add to iTunes.

Combining that with the seemingly obvious poor randomization (I too heard some songs over-and-over again, even if I started at a different spot in the generated playlist), I got really frustrated.

So to fix both these problems, I created a smart playlist of about 300 songs. It consists of all new songs (under a certain playcount) because I want to hear them more frequently for awhile, and the 200 songs played least recently.

With only 300 songs I can go through the entire playlist before wanting to re-sync. Then when I do, I get a new set of songs to listen to.