What is Digg?Sponsored by HowLifeWorks
How Private Online Shopping Clubs Work view!
howlifeworks.com - How to become a member and get discounts of up to 80% on must-have luxury goods
124 Comments
- knowall, on 10/12/2007, -5/+138Sigh... I know this is going to get front-paged... and will spawn all sorts of conspiracy theories.... but this is a ridiculous "study" if you can even call it that. This is hardly conclusive proof. They need to do it again and again to see if it's reproducible.
I can flip a coin 10 times and get tails 8 times. Does that mean the coin is broken? No. it means there's high variations in small data sets. Flip a coin 100,000 times and it will be much closer to 50:50. - inkbomb, on 10/12/2007, -25/+84"John Mayer (Sony) was iTunes' least favourite artist."
sweet, one down. now if only someone could make james blunt disappear from playlists around the world... - avalys, on 10/12/2007, -3/+54Nimrods. Random means random - it does not mean "automatically perfectly evenly distributed".
- joeshlub, on 10/12/2007, -1/+42This article is just ***** stupid. Anything that attempts to detemine whether or not something is random without any statistical analysis is such a huge waste of time it sickens me. Instead of dealing with this question by taking a statistically significant sample and analyzing it, they took a small one and then failed to actually see if there was any real correlation at play.
- CrankyMcGuy, on 10/12/2007, -3/+43"John Mayer (Sony) was iTunes' least favourite artist."
This fact alone should prove that there is no conspiracy going on here. Apple has featured John Mayer at Stevenotes a couple of times in the past few years. They like him. If anything you should expect his music to be at the top of the list if iTunes skews the results.
As another poster commented on, this isn't a very good study of iTunes random play and has been done better before. The sample was much too small to give reliable results and needed to be repeated again and again. - bSimms, on 10/12/2007, -6/+43John Mayer may still have that teenage girl scene image, but he is a god on the guitar on Continuum.
- jefdub, on 10/12/2007, -0/+36Exactly. If you read the 3 comments appended to the CNET story, their brief analysis is more insightful than all 3 pages of the article (including pretty excel graphs):
--Comment 1---
There are no calculations to show whether these findings were statistically significant over the dataset, there is merely conjecture while showing raw results. http://en.wikipedia.org/wiki/Statistical_significance
--Comment 2---
They should analyze it using the Chai squared analysis to determine if it is truly outside statistical variation or just chance.
--Comment 3---
From just glancing at this, it looks consistent with most types of LCGs. In fact, these are pretty close, in terms of frequency, for all categories. A 3% difference is negligible. Random does not mean a bell curve, sorry C-NET. This is amateur statistics at best. - HalBSure, on 10/12/2007, -0/+24I agree, the data set is way too small. Here's a semi-old article on the same topic that it is much better.
http://www.omninerd.com/2005/08/25/articles/34
It deals mainly with how weighted the ratings are, but it certainly gives a better glimpse into the whole thing. - Otto, on 10/12/2007, -2/+19This doesn't show that it's not "random". It shows that people don't understand WTF "random" really means.
The iPod and iTunes are both highly random. It's been proven, over and over again.
People, on the other hand, are very good at seeing patterns in chaos. It's hard-wired right into our brains. It's that part of the brain that helped our ancestors see the predators in the fields of grass and such. The downside is that we're really good at seeing patterns that are not actually there. We assign significance to these patterns where no such significance exists. - cds0528, on 10/12/2007, -1/+18they did the test one time. they should've done the same test, exact same setting, same files and everything all over again 2 or 3 more times... then i'll buy the conspiracy
Edit: looks like someone else said the same thing, sorry - sideshowRAHEEM, on 10/12/2007, -2/+18"Through payment of a fee, the theory goes, a record label could increase the rotation frequency of their own music by tweaking Apple's randomisation formula."
A good theory execpt for one thing if I already have a song on my iPod I have already bought/"stollen" it so why should the record lable give a ***** how many times I play it. - SurrealDream, on 10/12/2007, -14/+29I actually used to listen to John Mayer!
Thank the Lord for iTunes, for telling me what not to like. - Mootabolife, on 10/12/2007, -7/+20Makes you wonder, how else are they telling you what to like/dislike?
- MyNameIsJoe, on 10/12/2007, -0/+10Next up. Is a deck of cards random? I played cards last night and got pocket aces on 3 times and kings twice. I must have been cheating. Now all I have to do is figure out how I did it, so I can do it again next time I play.
- Santabot, on 10/12/2007, -0/+9lol @ cnet bashing via comments after the article, not a single positive statement.
- erikerikerik, on 10/12/2007, -8/+17eh, itunes is in fact "too random" and its a flaw with reality. I believe this was covered in an issue of wired some time last year.
what itunes was said be working on was a song flag. If a song was played too soon it was tossed back into the mix and not played.
But any one knows that order fallows chaos fallows order fallows chaos fallows order (get the point?)... How much or how little of a picture (or algorithm) you wish to look at will determent if "it" looks like order or chaos.
And skipping a few steps, this is very related to how we observe an action such as light. Study light as a wave and your results will back you that light is a wave. Study light as a partial and you're results will back your findings that way.
/end ranting - ezweave, on 10/12/2007, -2/+10I think they also fail to realize that random number generation (LCGs, in particular) don't always have a "bell curve" (normal distribution). That's the thing about random numbers (or statistics in general)... as you mention, even if you flip a coin 100 times, and it comes up heads (say) seventy times... that doesn't mean that the coin is broken. Statistics are useful for studying expected behavior, but it does not dictate outcome. So many people fail to realize this.
A bigger data set would make this more interesting and tell us more about it. I also notice that they never use any independent artists, which makes this seem more like a result of too many pints and not a well thought out study. - nihilette, on 10/12/2007, -2/+10iPods have feelings. you just have to leave it on shuffle to figure them out
- kapsar, on 10/12/2007, -2/+10I'd like to see this analyzed and shown to be statistically significant.
But personally i have noticed the same artists/songs coming up over and over on both my iPod and iTunes. One other thing i've noticed is that i'll hear a song on my computer then go to my car and then hear the exact same song on my iPod... i have 30 gigs of music and with constantly reshuffling i should not be hearing the same songs over and over again. - UberC, on 10/12/2007, -0/+8I wonder if they had the smart shuffle option on. That would explain why artist with multiple songs were played the same as artists with one song. Smart shuffle makes it less likely to hear multiple songs from the same artist.
- SultanTravi, on 10/12/2007, -2/+9Agreed. With modern computers, larger sample sizes and numerous trials are easy to do. They should have done much more.
- phronko, on 10/12/2007, -1/+7While I agree that those comments give more information than the article: It's "Chi Square", not "Chai". Chai is a type of tea.
- doctechnical, on 10/12/2007, -2/+7Feelings, perhaps, taste, no... based on it's putting Kate Bush and ABW&H at the bottom.
- goodoldharris, on 10/12/2007, -0/+5Seriously, did not a single person at CNET take statistics ***** 101? The ***** they write is getting worse by the day.
- flap, on 10/12/2007, -1/+6rip songs from c-d's, don't connect to the internet and strip all the metadata. Then do it 1000 times.
I would care about the outcome if they did that, then it would be easier to tell if itunes just sucks at shuffling. - mattsydoz, on 10/12/2007, -0/+5I don't let the music selection be random. Usually I play music in iTunes using Party Shuffle based on a playlist that is made up from several playlists:
1. The 40 least played songs and not played in the last 2 days.
2. The 40 least recently played songs.
3. 40 random rated songs not played in the last 2 days
4. 40 songs added in the last 3 months and not played in the last 2 days.
I have over 17 continuous days of music which has all been played in the last 6 months.
Of course at times I will add into the party shuffle stream music that I specifically want to hear right then. - SpeedyG, on 10/12/2007, -1/+6If you put about 70 people in a room, the odds are EXTREMELY likely that two people in that room will share a birthday. Conversely, you could put 700 people in that room, and it's still somewhat likely that not every day of the year will be covered.
Same applies here. Some are prevalent, some aren't, but when you've only drawn enough times to (ideally) cover everyone three or four times, you shouldn't be surprised at wild variations. - kp3469, on 10/12/2007, -0/+4knowall: thank you for taking the words right out of my mouth! also sadly lacking was any sort of probability analysis, i.e., in a group of 1300 slots, what is the probability that artist X gets played 60 times? 59 times? 4 times?
the statistics in the article are WEAK. - DaDiggCode, on 10/12/2007, -0/+4I'm taking a class of the study of economic statistics and I'd say this is BS.... this is what CNET and other news publications want you to think - that there is some unknown force controlling your music playlist. Their isn't... listen to Knowall
;) - rollem, on 10/12/2007, -1/+5Buried. This article is a perfect example of improper statistic approaches. In order to make any argument about the randomness of the playlist, one would have to do a simple chi-squred test to measure if the expected frequencies based on random drawings were actually different from the actual results found. Since just some raw percentages were given, there is no way to find out if this articale is actually reporting an interesting finding or not. Actually, they may have provided enough data to proceed with your own measurements, but this simple step should ahve been done by any self-respecting tester.
- laplacian, on 10/12/2007, -0/+4@jefdub
the best comment was next:
"Hey, CNET, how about a follow-up article with an actual statistician analyzing the results rather than someone who conducted this study like a high school intern." - Bigcat1021, on 10/12/2007, -0/+4Random is a process not a result.
- troysteinbauer, on 10/12/2007, -1/+5No matter what you guys think, it is random. Take a high school level statistics class and you will find out that even thought it may look NOT random, it very well is.
- acomj, on 10/12/2007, -2/+5@doctech
Your ipod should have your tastes, since you put the music on it (hopefully its music you like , isn't that the point?) - DaDiggCode, on 10/12/2007, -1/+4Joy - lets learn about how statistics really works :) http://en.wikipedia.org/wiki/Sampling_(statistics)
- DelMonte, on 10/12/2007, -0/+3The iPod shuffle creates a "random playlist" in advance each time you synch with iTunes.
If you want to change the shuffle playlist order without synching to iTunes, you can press the Play button on the shuffle three times (quickly) while in shuffle mode, this will generate a new "random playlist". If you don't synch with iTunes or don't press play three times, the random list will stay the same.
In linear mode, pressing play three times will make it jump to the first song on the normal playlist. - Raz4Life, on 10/12/2007, -0/+3...or party shuffle? It let's you select an option to play higher rated songs more often.
- dagospelman, on 10/12/2007, -0/+3awful piece of analysis. Not one recognised statistical test was performed to see if the results were statistically significant or not
- CorporateFelon, on 10/12/2007, -0/+3@Cleanlyness
One Thing I've noticed, and this isnt proven by any amount, but that the random playlists created have every song in the play list. So it knows what songs are coming up. The next song isnt chosen at random when the current one ends. Once you get to the end of the playlist, assuming that you have repeat on, a new random playlist will be generated. So when you hit that end/beginning mark and try to go forward/back continuously it'll keep being different songs.
One easy way to test this, but I'm too lazy to test, reset all the play counts, turn repeat off, and just hit shuffle and play. When it gets to the end, I think that each song should have played once. - inactive, on 10/12/2007, -2/+5Watch out for Paul Anka, my ipod can't get enough of him
- Bistromath, on 10/12/2007, -0/+3Didn't CNET have anyone on staff that could do a proper statistical analysis of the data? Like even some undergrad intern with 1 semester of Stats 101? Absurd conclusions being drawn here.
- jonahan52, on 10/12/2007, -0/+2There is also a preference under playback that allows you to change the likelihood of hear the same artist or same album. Apple did this a few versions ago because someone else did about the exact same article before. Jobs even talked about it in one of his keynotes.
- InfinitySnatch, on 10/12/2007, -1/+3iTunes: Just How Random is Random? The results may or may not surprise you.
- pHr3ak3r, on 10/12/2007, -2/+4Jeeezus ***** christ...how often do we need to hear about how random freaking itunes is? Just chillax...if you dont like the way it does it make it more or less random or...you could even skip a song :o Time is much better spent watching grass grow than running distributions on a random algorithm.
- GuyHersh, on 10/12/2007, -0/+2Dude, thats the most genius thing I've heard of ever.
I never really played with the power of smart playlists, and never thought of doing things that way, and then making a 5th playlist to combine the other 4, AND then using a party shuffle to play them so you can see whats coming.
God damn, you are my hero.. I just set up playlists as you described, and will tweak them as I want, but great thinking! any other tips? - ericsemail, on 10/12/2007, -0/+2As many have said, the data presented is more than likely statistically insignificant. 1000 times on a 1000 different machines plus a control group and maybe they'd be closer. This is why we have advanced statistics--which the article doesn't recognize at all. Is the iPod shuffle truly random--NO. No computer we have today can ever be truly random. Is the iPod shuffle feature genuine--I'd assume so--but the data presented to us tells us nothing to prove or disprove this.
- pruppert, on 10/12/2007, -1/+3Yo, CNET. Use a much bigger sample size. And show me some chi square results. Seriously, I'm curious.
- mwsherman, on 10/12/2007, -1/+3I don't believe anything without a p value.
- DelMonte, on 10/12/2007, -0/+2"The algorithm would be to assign a random number to each song in the entire library, then sort the songs by that number. The resulting list will have every song once, but in shuffled order."
That's actually how it works... Unless you make an iPod or iTunes re-generate its shuffle list you will never get the same song twice in the sequence.
What most people don't realize is that when they choose the "shuffle songs" command on the main menu of an iPod, it generates a new shuffle list. To never get a song twice you must avoid using this command and use the "now playing" feature instead to get back to where you were in the current random playlist. - foolfromhell, on 10/12/2007, -3/+5Edit time ran out: Anyone who wants to, pick up Stadium Arcadium. The best $30 you will spend in a year.
-
Show 51 - 100 of 125 discussions
© Digg Inc. 2009
— Content created and posted by Digg users is