Shuffle randomness question

I have searched the FAQ for "shuffle" and "randomness", and I only found a thread talking about 2.0b3 still having problems. Latest thread is this.

Here are some behaviors I see during shuffling. The playlist I am shuffling has about 2000 songs with multiple artist/album playlist underneath. The number of songs played is less than 100. Software is 2.0b11.

1. It seems like certain albums always get songs played. It is not the same song everytime, but the album seems to span a range of songs that gets picked often.

2. Out of 2000 songs, I would not expect to see two songs from the same album to be placed close to each other. This is independent of #1. I have once heard two songs back to back from the same album. I haven't done that many shuffles (maybe 20-30), so I would think it is statistically low to get this shuffle in so few tries.

Has anyone else seen this in 2.0b7 or 2.0b11? Also, I would like to do a test of the randomization. I would let the player randomize the 2000 songs, and then for the first 100 get the song postion in the unshuffled playlist. Is there a way in emplode I can get the position of a song in a top level playlist? I really don't want to skip to each song on the player to get the position.

Thanks for any info,

This happens all the time to me when I've got the whole player shuffled. I'm absolutely sure that there is some bunching going on. I had the whole player shuffled the other day and there was a series of about 20 tracks and all of them had been selected from about three albums. I've got just over 2500 tracks in all so it isn't like I've only got that number of albums on there. It was really freaking me out I can tell you. I reshuffled and everything was OK again (well back to the usual semi shuffled/semi bunched pseudo random shuffle). I haven't noticed it being any worse in the 2.0 betas than in 1.02/1.03. Except of course for the aforementioned 20 tracks.

Everyone notices this, it's not just you.

I still don't know whether it is a bug in the randomness or if it's simply a statistical thing that LOOKS like nonrandom behavior to us.

The last discussion on this topic is in here.