Yeah seems like you have to train it. I happened to have a lot of Pearl Jam in digital format on my computer (they like to release stuff in digital format) and iTunes decided that’s all I ever wanted to hear. I got it trained off that and now it’s on to thinking that ska music is all I want to hear (at least it went from a band to a genre). I teased it a bit to giving me more variety but if I don’t keep up with skipping songs, it sure does get in to a rut.
I’ve been trying to use WMP more and rate more. That seems to help a bit.
I don’t rate any of my music. I always figured that if I want it on my iTunes, I must like it. (Although I have no idea how I ended up with a Miley Cyrus song on my iTunes, so I guess this is not necessarily true.)
Does rating stuff change the likelihood that shuffle will play the song?
I personally find the shuffle to actually be pretty random.
There seems to be a lot of debate on whether iTunes is random (well, or techincally “pseudorandom.”) I’m not convinced one way or another, but until I see a good statistical analysis, I’m leaning towards random. People are good at finding supposed patterns in random data, and hearing complaints like a song hasn’t been played in ages, and then shows up three times in a day or so doesn’t necessarily compel me to believe the algorithm isn’t trying to be random.
One thing that I have noticed since switching to random is how incredibly annoying it is when a song comes on with more than a second of dead air for an intro. It makes me want to stab the artist in the back of the head with a tea spoon.
I haven’t used the random setting on my iTunes in a while, but when I did, it played “Fitter Happier” by Radiohead pretty often. Conclusion: my iTunes has a crush on the Stephen Hawking[ish] computer voice.
The apple random algorithm, if I recall, included bias for song popularity and the number of times it’s been played…there’s also the recognition that, if you have 100 U2 songs, and 10 Julio Iglesias songs, you’ll hear ten times as many U2 songs and STILL be random.
I haven’t seen a statistical analysis, but there was an article in The American Statistician within the past couple years that did some relevant probability calculations. Their conclusion was that everything that people complain about is what you would expect to see given a truly random shuffle method.
I have a Droid, and I thought that also, until some songs started repeating every two or three selections. I discovered that for some reason the playlist it was selecting from got very short. When I get too annoyed by this phenomenon, I reset the playlist and it gets better. (I don’t actually use playlists, so there shouldn’t be any.)
I have noticed, after a recent OS update, that it is favoring different tracks from the same album, though it shuffles albums fairly well. It goes through 3 or 4 tracks, selected at random, and then moves on to new ones.
It has always liked the second Traveling Wilburries album way too much, though.
#88 appears 4 times in a span of 32 random numbers (I’m reading left to right from top to bottom), and even appears consecutively. It’s played 3 times in a span of 14.
#38 appears three times.
If you say each group of 10 is an album (to keep it simple, let’s put “100” in with #1-9, and then #10-19, #20-29, #30-39, etc., are each one album), the random generator seems to favor album #8. Of the first 30 plays, 6 are from this album. In a stretch from plays 46 to 57, four songs from album #8 are picked. (#88,#82,#85,#89), including 3 of 4 in the final stretch of that slice of the playlist (#82,#58,#85,#89).
I’m sure you can find other interesting patterns if you look.
I’m not saying that Apple doesn’t perhaps include something more to its random algorithm–I don’t know–but I inherently the human perception of what is random and what isn’t.
The thing is, what people expect random to be should be what they give us rather than what random actually is. At this stage in the game, there’s no excuse for not keeping track of what songs have been played and not repeating any until you’ve went through them all. That’s what “shuffle” means, after all. It’s not like you can shuffle a deck of cards and somehow get two Aces of Spades.
I don’t care what the sciency types say, there are microcircuit elves that live in the damn Ipod and develop bizarre tastes which they feel they must inflict on me via shuffle. What I figure is happening is that a song comes up and the elves think “Oooh, that’s a good one! Let’s play it again, and again, and again!!!”
For instance, in an evil moment I added Ray Manzarek’s “Carmina Burana” to the playlist. Before I finally had enough and unchecked the album, the elves had played selections from it 65 times, compared to 46 playings of selections from an orchestral standard version that had been on the Pod longer and had many more selections to chose from. Both of these stats are way out of proportion.
And I don’t even particularly like “Carmina Burana”.
I’ve had to uncheck/delete a bunch of Mott the Hoople, because the elves were making it come up over and over again beyond the point of nausea.
There is also the damning evidence that the elves are playing stuff I don’t even get to hear. According to the most-played stats, one of my albums (a compilation of blues/swing numbers from the 30s-40s) has been played 21 times. I sure didn’t hear all those playings. I figure the elves have been grooving to “Red Bank Boogie” while I was off doing something else.
I’ve noticed this, not with iPod (which I have one somewhere, but damned if I know where) but with beading.
If I have a pile of beads of various colors, evenly distributed and evenly mixed, it’s going to happen sooner or later that, at random, I’ll blindly pick out 6 reds in a row. I have to actually pick and choose and design it if I want a strand to LOOK random, because random doesn’t look like what it really is.