Friday, 1 May 2009

Drowned in Sound Stats!

Drowned In Sound’s 3 Month Statistics: January - March

Finally after weeks of suspense (and the most prolonged drum-roll in history) comes the 3-Month listening charts from Drowned In Sound User’s Last.Fms!

This has taken a bit longer than I imagined it would when I set myself the task of collating it all. After I started logging all the responses I started thinking of too many possibilities to rotate all this data, so in the end I pretty much did every variable I could think of. It might not all be 100% statistically sound but it was all too interesting for its own good. Bit of a shame there were less than 100 respondents though. But next time we run this we’ll have everyone on DiS with a Last.FM doing this for us. I’ll have this database already set up for next time too, which will save 2 weeks of hacking about and preparation.

These stats are all far more detailed than any you get straight from Last.FM. Prettier too. Click on the tables and figures to view them in full size hi-res

Hey ho, let’s go:

Here are the basic introductory demographics:

Number of users surveyed: 91
Number of bands/artists: 915
Total plays in survey: 126762
Average plays per user: 1393
Average plays per band: 139
Range of plays per user: 6427 – 117
Range of plays per band/artist: 6273 - 06

I think it’s best if we go straight in with what you really want to see: The bands that have been listened to the most by the most people:

In the top 20 most listened to bands Animal Collective have 262% of the next most listened to artist, Radiohead (or 162% more!). Everyone is well excited about Merriweather Post Pavillion then? I see a lot of comfort listening in this list – is that what you’d call it? Pavement, Yo La Tengo, Mountain Goats, Casiotone – DiS users really love their Americana! 16 of those 20 bands are American too, with the other 4 being made of 2 English, 1 Scottish and 1 Swedish. Across this top 20 there’s a range of 5208 plays but so significantly greater is the popularity of Animal Collective that without including them there is only a range of 1326 plays. Whopping. Check out that spike:

The Mean average plays of the top 20 is 1712.4, at which only the top 5 artists are above. The Median average is 1364 (at which Animal Collective are 4.6 times more listened to).

In the tables 1a and 1b below, it is interesting to note that the positions of bands in the order of most listened do not correspond with the average plays per listener.

82 unique listeners contribute to the 20 most listened to artists, which means that 9 of you don’t listen to any of this populist shit. Table 2 shows the list of users contributing the most artists to this list.

The 7 artists each of the top 3 users have in the Artists Ranked By Total Plays are shown in Tables 3a – 3c, with the count of plays and position in their own charts:

What do we have here? Nothing very cohesive except for Animal Collective. Again. If you listen to music, you're more likely to have listened to Merriweather in the last 3 months than anything else.

In Figure 2 below, there is more illustration of this dominance. Representative of their position in the total plays ranking, Animal Collective are listened to by a considerably greater number of individuals than the second placed artists – more than twice as many people have them in their charts than the second placed Radiohead and Yeah Yeah Yeahs.

15 of the 24 artists in Figure 2 appear in the list of 20 most played artists in Figure 1.

The Mean average number of users listening to each artist is just under 2 (1.98). Table 4 shows that 621 of the artists listened to are only listened to by 1 user each. From there, there is a huge jump to only 137 artists being listened to by 2 people, from which there is a gradual increase and fluctuation of artists per user up to 17 listeners. The big jump from 17 listeners per artist to 37 again shows the unique and significant popularity of Animal Collective.

This next table spins the data in a completely different way and averages the positions that the artists appear in user’s charts. These are all the joint highest appearing artists (in alphabetical order). It’s a way of making stats work for you and getting A Place to Bury Strangers to the top of a chart. How else are we going to do that? Here it is along with 5b, showing how many artists are in each position:

If this is all getting a bit far-fetched and airy these next numbers will bring us back to solid ground:

Figure 3 shows the people with the most unique artists in their playlists, with jimitheexploder having 18 – only 2 of the artists he listens to are listened to by another user. These were Ricky Gervais, Steve Merchant and Karl Pilkington (also listened to by kittensgotclaws); and Boxcutter (also listened to by fragrant_vagrant). Interestingly, these two artists that he shares are his two top listens with 95 and 89 plays respectively.

I don’t really know what else to say with this. Are these people the coolest? Maybe I should have done that figure in ice blue.

Every user has at least 1 artist that is unique to them. The bottom of this list is represented by these joint-top-5 most populist mother fuckers with 1 unique artist each:

GMMiles (Ra Ra Riot)
Pleasantpeasant (Buck 65)
Shuffles (The Delgados)
Wolflikeme (Simon Joyner)
XYZIA (The Flaming Lips)
YouRecluse (Codeine)

Be safe in the knowledge that they are the most likely to know what you’re listening to. And like it.

Table 6 below shows the number of listens the most played uniquely appearing artists has received. The average number of Plays for unique artists is 79. Victory_Rose has listened to Coldplay over 8.5 times more than the average lone artist listener (and 0.5% of the total listening count!)

Now my plausibility might start to fall apart - I am undecided as to which side of fallible these fall on – I am not sure exactly what they are supposed to be showing, but it’s a stat, so it’s going in! It is possible they need a bit of explaining too, I don’t think my snappy little title does them justice: The following figures show users ranked against the sum of the positions of all users across their own top 20 artists. Are you with me? So, what is a ‘good’ score and what is ‘bad’ here?

A score of 210 would mean that each of your 20 artists did not appear in anyone else’s lists but your own (the sum of their positions being 1+2+3+4+5+6…etc). As the person with the most unique artists jimitheexploder comes the closes to this perfect score. It makes this chart look pretty too because it looks like jimi is the apex from which everyone else fans out of.

This sister-figure shows the bottom end of the same data and is way more parallel, but that might be more to do with the scales I used. What does this one mean? It means that Shuffles listens to artists that are ranked around the lower ends of people’s top 20’s. As indicated by having a higher score than those users in the figure above, the artists these users listen to either feature lower down people's lists or feature across more lists. Most likely, a combination of both, as all the other tables above will give clues to.

From the least represented to the most representing – Figure 6 illustrates the users who listen to their top 20 artists the most. Pure and simple no messing numbers to finish.

DangerousSnails is pretty far out there in the lead as the most active listener of us all. So what does he do with all this listening time? This:

So what have we learned, and what do all these numbers and pictures tell us? That Animal Collective are by far and away the most popular band of the first 3 months of 2009? That their monopoly on the most played charts does not alter the fact that most people listen to more artists less often? That jimitheexploder is the go-to guy for music the rest of us are unlikely to have heard before? That DangerousSnails does nothing else with his day other than listen to music?
All very valuable information.

Do we do this again at the end of June, September and December?

Download a nice pdf version for posterity: HERE
Download the full Total Play Count list: Here

Responses on Drowned in Sound board: HERE



  1. hi bigtweet, this is pretty useful. did you capture the data automatically or was there a painful manual element to all this?

  2. Incredible stuff mate, well done. For future collation, it might save you some time to use the XML feeds produces, e.g.

    Some way to parse that quicker than copy pasting the data from the DiS messages maybe? You'd just need a list of participating usernames that way.

    Look forward to seeing this with a bigger sample size, and then making a "The ultimate DiS lowest common denominator party album in the world ever vol. 2" CD.


  3. This is interesting! & you should def. do this in another 3 months!