The Definitive “Is the Shuffle Random?” Thread

[mtnagel]

There are quite a few of these threads that talk about whether the iPod is truly random in how it plays the songs or whether it could repeat some songs while others had not played. Well, hopefully this will end the speculation and confirm that the Shuffle is infact random and doesn’t repeat songs before all the songs have played.

My Test
The first test I ran was flawed because I accidentally put songs that already had a playcount greater than 0, which led me to believe that it played some songs while others weren’t played because I had playcounts of greater than 1, while others were still 0 (so if you want to test it using playcounts, make sure you don't do that).

I realized my mistake and then put on only songs that had a playcount of 0. After playing 154 songs out of the roughly 240 songs on my 1 gig Shuffle over a couple sessions (turning it off and letting it sit in between), I synced it up to iTunes and those 154 songs now had a playcount of 1 and the rest still were 0. While not 100% definitive, it seemed pretty likely to me that it wasn’t going to repeat a song before playing through all of them. I think this also shows that it's okay to turn off your Shuffle and it will still be random (and save your place) when you turn it back on.

Another Test
Then niknah posted a test in this thread (http://forums.ipodlounge.com/showthr...5&pagenumber=1) which I think illustrates an important point. If you use the track advance button to skip past a song, even if there is only a second left in the song, the Shuffle may repeat that song before all the other songs are played. Here are the results of a test using only 4 songs when using the track advance to quickly go through the songs:

song 2, song 4, song 1, song 2, 3, 4, 2, 3, 1, 4, 1, 2, 3, 4, 1, 3, 2, 4, 3, 1

This obviously had song 2 repeat before song 3 was ever played. But then I asked about letting the songs finish and they repeated the test, but this time not using the track advance button and just letting the songs end. Here is the order the songs were played:

{1, 4, 3, 2}, {1, 3, 2, 4}, {1, 3, 4, 2}, {1, 2, 4, 3}, 1, 3

As you can see, for every time all four songs were played (I highlighted this by using brackets), it never repeated a song before all the others had been played. I’d say this definitely proves that the Shuffle is truly random and won’t repeat a song until it has at least played every other song.

Interesting to note that in the test above, there is a different random order for each time it finishes all the songs on the Shuffle. I believe others have suggested that it would use the same random order when it finished, but this is not the case. By “the same random order,” I mean if the results would have been {1, 4, 3, 2}, {1, 4, 3, 2}, {1, 4, 3, 2}, etc…, then that would have been the same random order, but obviously that’s not what happened.

Newsweek Article
http://msnbc.msn.com/id/6854309/site/newsweek/ - Here’s a Newsweek article that also talks about the “randomness” of the iPod in general. It’s a pretty interesting read on the topic.

Randomness After Syncing
I haven’t run the test yet, but I’m pretty sure based on the behavior or my 4G 40 gig iPod, that if you were to sync and replace some songs after listening to some, that you may get repeats on the songs that you didn’t delete off the Shuffle. Of course, if you always replace all the songs on the Shuffle, then you will be fine.

On my 4G 40 gig iPod, if I were in the middle of listening through a playlist (obviously not the same on the Shuffle since it only has one big playlist) and then synced, it would lose my place in that playlist and then it would repeat songs while others would go unplayed. So, I believe that if you, for example, listened to song 2 on your Shuffle and then synced up with iTunes, but didn’t replace song 2 on your Shuffle, that it would be entirely possible to have song 2 repeated before you hear all the other songs at least once. Like I said, I didn’t run this test, because I always put songs with 0 playcounts on mine, so when I sync, I can tell which ones have played already (the ones with a playcount of 1), so I delete those off and add new songs with 0 playcounts and then start over. So, the situation I described with song 2 above would never happen to me, but it may happen to others. It’s just something to be aware of and maybe that’s the issue some were having.

Randomness After Charging
When I say charging, I’m strictly speaking of using a computer without iTunes or using a USB wall charger. If you want to charge your Shuffle on the computer with iTunes and it automatically loads and starts updating, then I believe you may have a situation mentioned above in the “Randomness After Syncing” section. This might apply if you use your computer with iTunes, but have it set to not open when you insert the Shuffle, but I’m not sure.

Again, I haven’t run this test, but I’m basing this on the behavior of my 4G 40 gig, which when using the wall charger, it did not lose its place in a playlist and then wouldn’t repeat songs before all of the songs were played. I’m assuming that to be true in the case of the Shuffle also. I believe that you could use a computer without iTunes or a USB wall charger to charge the Shuffle and it would still be random and not repeat songs before all of the songs were heard.

Summary
Unless you use the track advance to skip past a song, the iPod Shuffle will play every song only once before picking a new random order to play all the songs again, which will not play a song a third time until you have heard all the songs a second time.

Please let me know if you have done any other tests that show that it is truly and I will add them here.

[jonathan_ive]

wow, now that is a random test

yeah i also have noticed that when i skip tracks i get repeat songs on my ipod and if you reshuffle it.

[Kurashima]

Good, but to really be definitive, you should call it shuffle instead of random as the whole point of the test is to prove it's not random. I guess the main thing that determines whether a song repeats during regular listening then is whether you skipped the song or played it fully. That's good and bad, as if I skipped it it may have been a bad song and the shuffle will almost always play it again before all other songs have been played, making it haunt me throughout my playlist. However, if I just wasn't in the mood for it, then it's probably a good thing that I get to hear it again. I think this will only end up annoying people though as most would skip the song for a reason. Not a big flaw certainly, especially if you enjoy all of your music (after adding my own stuff, I autofill from 3 stars and up).

[edit] In fact, expanding on that, I wonder how it would work if you - for example - continuously skip only one song while playing the rest fully. If the shuffle does go by playcount, then the closer you get to playing all of your songs the more often that song shows up, making you handle the shuffle all the time when you just want to leave it alone and listen to music. I wonder if there's some kind of safety-catch for this.

[edit3] I'm thinking that instead of going by playcount, then if the song is skipped, it is inserted into a "random" place in the playlist (it's not put at the end as proven by examining your initial test results, as song 3 would then have played before the other songs were repeated). This also begs the useless but interesting question that, if you go backwards after skipping the song and listen to it fully, does the repeat it created dissapear? Hmmm, I think Apple should really have explained the exact nature of the shuffle to us. These points might seem trivial but when the whole point of the shuffle is to play your music shuffled and you're very limited using play-in-order mode, then you probably want to know exactly how it works.

[mtnagel]

Originally posted by Kurashima
Good, but to really be definitive, you should call it shuffle instead of random as the whole point of the test is to prove it's not random.

I still random is the right term, but you could call it "random-without-repeat." That's accurate

[edit] In fact, expanding on that, I wonder how it would work if you - for example - continuously skip only one song while playing the rest fully. If the shuffle does go by playcount, then the closer you get to playing all of your songs the more often that song shows up, making you handle the shuffle all the time when you just want to leave it alone and listen to music. I wonder if there's some kind of safety-catch for this.

For me, that's not an issue since I doubt I ever would listen to it for 15 hours without being near my computer to reload.

[kloan]

That's interesting that it will repeat a song if it's been skipped. The whole point of skipping a track is that I don't want to hear it, but it's going to play it again before it plays other songs?? I don't like that.

[jonathan_ive]

Originally posted by kloan
That's interesting that it will repeat a song if it's been skipped. The whole point of skipping a track is that I don't want to hear it, but it's going to play it again before it plays other songs?? I don't like that.

some times i like a track but i am just not in the mood to listen to it at that time.

[mtnagel]

Originally posted by kloan
That's interesting that it will repeat a song if it's been skipped. The whole point of skipping a track is that I don't want to hear it, but it's going to play it again before it plays other songs?? I don't like that.

It would only repeat a track you skipped if you actually listened to all the other songs. I guess that would be easier with the 512 mb one, but with my 1 gig one with 15.5 hours of music, I would never do that.

[Kurashima]

Not necessarily. Going by your first test, you haven't even gotten to song 3 yet but song 2 already repeated itself.

[mtnagel]

Originally posted by Kurashima
Not necessarily. Going by your first test, you haven't even gotten to song 3 yet but song 2 already repeated itself.

I know it's long, but you should make sure to read it all. That test was based on using the track advance, which will allow them to be repeated. It's not a valid test. You need to let the song play all the way through or will get repeats.

[Kurashima]

I did read it. Track advance or not, the third track never even had a chance to play before the songs you skipped repeated. Therefore, if you skip even just one track, then it may (and I think most likely will) repeat before all of the rest of the songs play, which was what you told koun it would not do.

By the way, koun asked if a track he skipped (aka. used track advance) would repeat, so using the track advance test is not only logical, but necessary.

[mtnagel]

Originally posted by Kurashima
I did read it. Track advance or not, the third track never even had a chance to play before the songs you skipped repeated. Therefore, if you skip even just one track, then it may (and I think most likely will) repeat before all of the rest of the songs play, which was what you told koun it would not do.

By the way, koun asked if a track he skipped (aka. used track advance) would repeat, so using the track advance test is not only logical, but necessary.

Ahhh. Now I see what you are saying and you are correct. What I said to kloan was wrong.

[Dance-Rock.com]

Originally posted by kloan
That's interesting that it will repeat a song if it's been skipped. The whole point of skipping a track is that I don't want to hear it, but it's going to play it again before it plays other songs?? I don't like that.

I don't like it either. ####es me off when I use my shuffle at the gym.

[mtnagel]

Originally posted by Dance-Rock.com
I don't like it either. ####es me off when I use my shuffle at the gym.

Yeah, I'm sure that can be annoying, but that's why I only try to add songs to my library that I really want to hear. Plus, I only go to the gym an hour a day for 4-5 days, and I have my 1 gig filled to the brim with around 240 songs, and I usually only get through 15 a day at the gym, so if I do skip one, it's not that likely to get played again before I sync it (usually every week).

[shenkin]

The fact that it plays all songs before it repeats any doesn't make it "more truly random" than if it didn't -- it just makes it "random with replacement" (in the high-school probability teaching metaphor of urns and beads); "random without replacement" would make it play each song with equal probability whether it had been played before or not.

The sequence "{1, 4, 3, 2}, {1, 3, 2, 4}, {1, 3, 4, 2}, {1, 2, 4, 3}, 1, 3" looks pretty non-random to me. Why does it always start with 1? I would have expected it to select a single random sequence and then repeat that sequence, or else to completely randomize every song, rather than making Tune 1 the first tune in each cycle.

I'm also curious about the randomness of the selection from iTunes when the shuffle is filled. I had it filled randomly without regard to "higher-rated" songs. Indeed, I don't have ratings on my songs. Out of 10000 tunes, I have 250 by one artist, or 2.5%. However, 15 of the 219 tunes on my iPod are by this artist -- about 7%. I haven't calculated the likelihood of this occurring due to a random fluctuation, but it looks like a rather anomalous enrichment offhand. Could there be a bias in favor of songs that one has listened to most recently or most frequently? This is an artist that I like a lot and have listened to in iTunes more frequently than average.

[shenkin]

Whoops, I got it backwards: random with replacement would would give equal probabilities regardless of whether it had been played before. Random without replacement would play all songs before repeating. Sorry about that.... But they're both random.