Sticky labels

Posted in Indulgence, music stats on July 22nd, 2010 by Alex

Metalheads do like to label stuff. The proliferation of metal subgenres is nothing sort of staggering. According to the Encyclopedia Metallum – The Discordance (basically a big old database of metal based information mined from the Encyclopedia Metallum – thanks to Cosmo at Invisible Oranges for publicizing this fascinating site):

There are almost 5500 “genres” attributed in [Encyclopedia Metallum]. 5000+ of these have fewer than 10 bands each. Almost 4000 of them are assigned to exactly 1 band. But this chaos of descriptors is actually constructed of a fairly small set of significant terms, rearranged in more or less every possible permutation and assembly.

Metal band count by genre

When describing music it’s essential to have a lexicon of descriptive terms to work with. Conveying the essence of a song, sound or style in words is not easy, so saying “X sounds like Y” is an essential mechanism. When certain styles repeat often enough we create classifications for them, usually described as genres. Most types of music have a pretty discrete set of genres supported by a range of other related terms to help qualify and further refine one’s description. So if we take an Indie band, for example Elbow, we could describe them as “British prog inflected melancholic indie rock”. This not a genre in itself – most people, on hearing the music would simply call it “Indie”, or “Alternative” or maybe just “Rock”, those words are just one person’s description. However, in the metal world, Tombs are not simply described as “blackened hardcore”, to metal fans the are Blackened Hardcore. It’s not enough to say that At the Gates are Death Metal, they are Melodic Death Metal. Melodic Death Metal may be a sub-genre of Death Metal, but they are not the same thing. Most other genres don’t suffer from this genre proliferation. We have shoegaze indie (and shoegaze metal lately!) and Happy Hardcore is a sub-genre of dance, old-school Hip-hop, Alt-Country, Prog Rock, but it usually stops at a single sub-strata rather than branching endlessly as sometimes happens in metal. So we have:

Metal ->

Death ->

Progressive Death

Brutal Death

Gore Metal

Deathcore

Melodic Death

Death/Doom

Technical Death

Blackened Death

We see this to an extent in other genres, but it it somehow seldom manages to stick in the same way (eg. Big Beat, Nu-rave, and Acid Jazz were all pretty short lived. That said NWOAHM and Pirate Metal are unlikely to stand the test of time).

In metal, it gets more baffling still when you look at the lexical topography of the metal genre spectrum. At first glance it would appear to be broadly taxonomical – that is, there is a rigid hierarchy of genres and sub-genres, and any given band fits somewhere. We start at the top with Metal (itself a subset of Rock), and below we have the mainline sub-genres, eg. Thrash, Death, Black, Hardcore…hold on, isn’t Hardcore a sub-genre of Punk? But its influence on metal (not mention the influence of Punk in general on metal) is huge. So here we hit our first problem – Metal isn’t pure, it’s an amalgam, and strictly taxonomical structures will not easily apply. If we gloss over that we hit other issues. For example, Deathcore would probably considered a subgenre of Death Metal, whereas Mathcore is most definitely a subgenre of hardcore. So although the *cores are (somewhat dubiously) afforded a common heritage, they are taxonomically fragmented. Then we have the issue of genre bleedthrough – Alcest are both Black Metal and Shoegaze. Neurosis could variously be described as Post-hardcore, Sludge, Doom, Post-rock etc. etc. And what of Sludge? A sub-genre of Doom or Punk? And try even classifying Baroness, SYL, Cobalt, Om, Earth

(Incidentally an attempt has been made to establish a taxonomy of musical genres in an attempt to make sense of the burgeoning availability of music in the digital era. Read about it here.)

The ‘structure’ is perhaps more ontological (in the more modern sense of the word), where genres are related to each other in a more semantic sense. So Symphonic Metal, a sub-genre of Power Metal, is like Symphonic Black Metal even though they’re from a pretty distinct lineage. Some genres are grouped geographically – Swedish Death metal was very much a style of its own, NWOBHM (and the various other perversions of that acronym that we’ve been subjected to since). NWOBHM could also be taken to denote a period in time and only bands from that period (late 70′s to early 80′s) could belong to this genre – bands practising this type of music these days would probably be described as Classic Metal. Death Metal derives (to a significant extent) from Thrash, but is not a subgenre of Thrash. Thrash in turn borrows from hardcore although it largely derives from NWOBHM and proto-BM.

Some genres are stylistically ridiculously broad – Folk Metal (which encompasses some pretty disparate regional influences), Post-Hardcore (this one pretty much defies any meaningful definition and probably doesn’t exist at all). On the other hand some descriptors and sub-strata are pretty definitive, and a genre of 1 (or very few) is a very real concept. The Black Metal fraternity are particularly proficient at this, being so pathologically tribal. So there’s really only a couple of Orthodox Black Metal bands, and post-BM and Depressive-BM are pretty sparse. I think this gives us a suggestion of what all this genre chaos is about – loyalty, belonging, pride.

The bewildering array of sub-genres is as constraining as it is helpful, and many bands balk at the being pigeon-holed, but still journalists, record labels, bands and fans alike create and perpetuate these labels. Metallers are often outsiders, and tend to cluster in cliques of similar minds. The different genres attract different types of metalheads, and the genre label acts like a badge, drawing people to it and each other – something more tangible to take pride in and remain loyal to. I also think metalheads are often shameless geeks – like stamp collectors they collect and group and organise, like train spotters they note sightings.

Interestingly, genre labels are often synonymous with feelings of quality. So labelling something Nu-Metal is generally derogatory, whereas calling something ‘Progressive’ is generally considered good (despite the link with the overtly indulgent and un-cool  70′s prog scene). These days the *cores are often seen as bad, mainly by the metal snoberatti and hardcore purists. Black Metal kvltists are notoriously defensive of their beloved genre to the extent of attempting to ‘unclassify’ bands that don’t suit their view of what the genre represents, despite clear lineage (Cradle of Filth, Dimmu Bogir). There’s actually a report in The Discordance that claims to quantify this (see it here).

Perhaps the most compelling reason for metal genre proliferation is simply the diversity of the metal genre. When both Steel Panther and Sunn O))) coexist under the metal banner which derives influence from Hip-hop, Folk, Classical, Dance and even Country some serious descriptors are needed just so we can get a feel for what stuff might suit our particular taste. The ‘metal’ genre itself comes with much baggage and preconceptions particularly with outsiders whom I find often have a narrow view of what that tag represents. Simply calling a band ‘metal’ does little convey what the bands represent (unless perhaps you’re talking about archetypes like Judas Priest or Iron Maiden) so a more specific definition would seem appropriate – shame they often mean nothing to anyone not pretty well acquainted with the genre!

It’s a fascinating subject, and you could probably write a whole book on the variations, nuances, idiocies and inconsistencies contained within the metal subgenre spectrum. At the end of the day, call it what you want – if it sounds good then stop talking about it and concentrate of listening to it!

Tags: , , ,

Numbers of the Beast #2: A skew loose?

Posted in Indulgence, Uncategorized, music stats on June 24th, 2010 by Alex
Lemmy - A skew loose?

Lemmy - a skew loose?

So we’ve learned that some bands have pretty obsessive fans. It’s nice that they’re listening to a lot of Opeth’s music, but are they just getting gooey over one album or even a single track? Let me demonstrate what I mean at its most extreme – the one hit wonder. This is not a phenomenon that’s particularly prevalent in the metal genre, so we’ll look to the genre most susceptible: pop. When I think of one hit wonders, one track always seems to spring to mind: Deep Blue Something – Breakfast at Tiffany’s. It’s a nauseating ditty that seems to blight the airwaves still, even 17 years after its release. Let’s see what Deep Blue Something’s last.fm listener profile looks like shall we?

Deep Blue Something last.fm

Breakfast at Tiffany's and some other tracks

Oh dear, a staggering 83% of Deep Blue Something’s overall listens were from that one track that you doubtlessly find occasionally looping round in your brain, eating away at your soul. In statistical speak this effect is broadly referred to as Skew or Skewness. According to Wikipedia “In probability theory and statistics, skewness is a measure of the asymmetry of the probability distribution of a real-valued random variable.” Ummm, yes. Put more simply, and in the current context, if folks are listening to 1 or 2 songs from a bands catalogue lots, and hardly anything else, then that band’s playcount would be considered to be skewed. In last.fm circles, the calculation for this is often referred to as AEP (I won’t bore you with what it means, other than it’s a fairly arbitrary statistical calculation) which gives an indication of skew across a band’s top 50 tracks. The AEP is a value of between 0 and 5 that indicate skewness, where 5 is not skewed at all (all tracks listened to exactly that same amount of times) and 0 (or less) is very skewed. Thanks to that wretched track, Deep Blue Something’s AEP is -13 (yes minus 13), compared to, to pick another more successful pop act, Michael Jackson, whose AEP is a respectable 2.7.

So, how does my list of metal acts fare in the AEP stakes? Let’s have a look:

Artist AEP
Devin Townsend 4.43
Neurosis 4.41
Blind Guardian 4.37
Napalm Death 4.35
Opeth 4.32
System of a Down 4.3
Cathedral 4.29
In Flames 4.28
Tool 4.27
Children of Bodom 4.25

So here we see a different picture again. These are very high AEP’s, which indicates that the bands’ top 50 most listened to tracks are listened to a comparable amount of times. What this suggests about a band is that they’re not just a 1 trick pony – their fans love a wide variety of their tracks rather than listening to just a couple before moving on. Predictably, In Flames make a reappearance, Opeth remain strong, and we all knew that Devin Townsend fans were an obsessive bunch (this incidentally, doesn’t include all the numerous variations on Townsend’s solo band names, or SYL, who are #32 in this list). This is a respectable list – Neurosis, Tool, Cathedral, all at the top of their game and widely respected, and there’s a real mix of genres here. Perhaps this is a demonstration of a quality all round band, no filler. Albums bands, career artists.

This may all be true, but things get a lot more interesting, and confusing, if we consider the other end of my (far from exhaustive) list:

Artist AEP
Nirvana 3.03
Ozzy Osbourne 2.53
Black Sabbath 1.84
Soundgarden 1.69
Mötorhead -1.28

Yeah, a bunch of flash in the pan, one hit wonders – non-players. Oh wait…those would actually be some of the most revered and respected bands in the rock/metal arena! What went wrong? I’ll give you 5 reasons: Smells Like Teen Spirit, Crazy Train, Paranoid, Black Hole Sun and last, but by no means least, Ace of Spades. If you don’t know exactly what those 5 labels refer to, then you must have been living in a cave for the past 40 years. Now, for many of these bands, these tracks are the worst, but not the only, offender (Come as You Are is a close second for Nirvana for example), but each bands have significant skew thanks to these BIG hits in their back catalogue, something that few of our least skewed acts have. So we’re still missing a dimension here…

Let’s take Mötorhead, who are skewed into minus numbers by their ‘classic’ (quoted as Lemmy doesn’t reckon it’s their best track) Ace of Spades. Now, as we saw earlier, Breakfast at Tiffiny’s accounts for 83% of DBS’s overall listens, so what of Ace of Spades? Well, it clocks in at a modest 10% of Mötorhead overall listens. So where are all the other listens going? Well, remember that AEP is calculated across a band’s top 50 tracks, so the majority of listens of Mötorhead’s tracks must be happening outside of their top 50. Given Mötorhead’s rich and voluminous back catalogue this is hardly surprising.

So, there’s another calculation that will tell us which bands benefit from this sort of listener attention, it’s called the Long Tail and we’ll discuss this in the next article.

Related articles:

A long tail of a critical discrepancy

Numbers of the Beast #1: Love you long time

Tags: , , , , , , , , , , , , , ,

Numbers of the Beast #1: Love you long time

Posted in music stats on May 3rd, 2010 by Alex

Since my bout of rampant music stats geekery I’ve been getting a tad OCD over the last.fm listening habits of you metal heads. So much so that I wrote a PHP application that queries the API and pulls back lots of lovely data (I’ll unleash this on the public when I’ve had the chance to debug it and tidy it up a bit). I’ve been munging and caressing the data to see what wonders it yields, some of which I’m going to share with you over a few articles.

Previously I uncovered In Flames fans’ obsessive behaviour by observing the interaction between their playcount and listeners. This is interesting, but only tells us part of the story. Just because a band’s fans listen to lots of their tracks, doesn’t mean that they spend that much time listening to them. Napalm Death fans listening to their debut album Scum will consume 28 tracks in 33 minutes, whereas Earth fans would only squeeze 1 track off of Earth 2 in that time. So we need to look at another dimension: track length. To do this I calculate the average track length from the band’s 50 most listened to tracks then multiply that by the overall playcount. This is what we get:

Band Average Song Length Total Minutes Listened
1 Metallica 5:56 769,675,558
2 Nine Inch Nails 4:26 436,266,693
3 Iron Maiden 5:43 396,650,321
4 Linkin Park 3:19 382,221,816
5 Tool 5:45 367,679,593

(Just in case you were wondering, that amounts to Metallica fans having collectively spent 1.5 centuries listening to their beloved band)

This doesn’t really change the picture that much. Metallica are still way out front, and the big boys still dominate. So let’s look at a slightly different stat – Average Time per Listener. We get this by multiplying the Plays per Listener with the Average Song Length. A pretty fuzzy calculation I know, but indicative none the less. Here’s what we get:

Band Average Listener time Average Song Length Plays per Listener
1 Opeth 9:35 7:26 77.41
2 Metallica 7:55 5:56 80.1
3 In Flames 7:38 3.50 119.55
4 Nightwish 6:39 5:01 79.35
5 Deathspell Omega 6:11 6:53 53.9

So here we see that Opeth fans spend quite a lot more time lost in the meanderings of Mr. Akerfeldt than any other band. In Flames, perhaps unsurprisingly, make a reappearance, and bringing up the rear, the real underdog, the dark lords of Orthodox Black Metal – Deathspell Omega.

Interestingly this does even the score somewhat with the critics list, as we now have Neurosis and Tool in the top 15, however, Black Sabbath suffer even greater humilation in this list as they drop down to number 49, just behind Lostprophets…ouch!

So, what does this tell us about the listening habits of the metal head? Well, I think it demonstrates that we’re and obsessive bunch, and we like what we like, and lots of it. Also, stats like this go some way to filtering out the noise of the casual listener and indicate where the real heart of the metal community lies.

Tags: , , , , , , ,

A long tail of a critical discrepancy

Posted in Indulgence, Music Industry, music stats on April 23rd, 2010 by Alex

last.fm's top metal artists

In this digital age of seemingly infinite choice of music that’s easy to access and at low cost (or free) it would easy to assume that the music listening masses would broaden their horizons a little. The Long Tail economy surely applies to music as much if not more than any other popular media. Not so according to various sales and download figures which suggest that in the current market, music sales are showing a higher propensity to the more popular tracks/artists than they did 5 years ago. There are various possible reasons for this (the paradox of choice, pack mentality, an explosion in poor taste) and it would be easy to dismiss this as this as mere statistics, but with recording artists struggling to cover costs and labels slashing rosters the harsh reality is only too obvious for some. What’s even more depressing is that there’s evidence in the movie and book worlds higher selling items often get better reviews, even when they suck. This is a known phenomenon:

In “Formal Theories of Mass Behaviour”, William McPhee noted that a disproportionate share of the audience for a hit was made up of people who consumed few products of that type. (Many other studies have since reached the same conclusion.) A lot of the people who read a bestselling novel, for example, do not read much other fiction. By contrast, the audience for an obscure novel is largely composed of people who read a lot. That means the least popular books are judged by people who have the highest standards, while the most popular are judged by people who literally do not know any better. An American who read just one book this year was disproportionately likely to have read “The Lost Symbol”, by Dan Brown. He almost certainly liked it.

Could this be the same for music and more specifically metal? My apparent snobbishness leads me to automatically assume that it doesn’t. Metal is a hugely diverse genre, and metal fans, despite being somewhat cliquey, are a diverse and contrary bunch who positively revel in the lonely outposts of the long tail. But is this really the case? I decided to do a little research on the matter. This lead me on a bit of an odyssey. I’m a closet stats junkie, and I’m easily distracted by trivia, so this is more of a journey than a destination, but I thought I’d share what I found anyway.

So firstly, do metallers show a propensity to rate popular music more highly? I headed off the social review site rateyourmusic.com and pulled the charts for the highest rated metal album of all time.

  1. Black Sabbath – Paranoid
  2. Metallica – Master of Puppets
  3. Black Sabbath – Master of Reality
  4. Black Sabbath – Black Sabbath
  5. Metallica – Ride the Lightning
  6. Megadeth – Rust in Peace
  7. Kyuss – Welcome to Sky Valley
  8. Tool – Ænima
  9. Slayer – Reign in Blood
  10. Death – Symbolic
  11. Opeth – Still Life
  12. Opeth – Blackwater Park
  13. Tool – Lateralus
  14. Iron Maiden – Powerslave
  15. Judas Priest – Painkiller

(full chart here)

Now, I’m not going to get into the relative merits or shortcomings of this list, I don’t really agree with it, however it seems reasonable and everything there certainly warrants a mention in the context of the best ever metal albums – there are no real stinkers here. The caveat here is that folks who write reviews on this site will tend to be self styled ‘critics’ rather than the casual listeners.

So, how does that compare to what people actually listen to. I pulled a (somewhat patchy) chart of overall listens of metal bands from last.fm.

Band Listens Listeners
1 Metallica 129,158,165 1,614,913
2 System of a Down 110,234,856 1,859,323
3 Nine Inch Nails 98,243,492 1,269,462
4 Nirvana 91,601,656 2,246,164
5 In Flames 77,571,473 648,985
6 Rammstein 74,874,781 1,080,014
7 Iron Maiden 71,102,754 1,118,305
8 Nightwish 66,186,512 830,576
9 Tool 63,722,501 1,021,247
10 Pearl Jam 59,373,659 1,405,955
11 Korn 59,012,020 1,316,256
12 Slipknot 56,894,546 1,112,724
13 AC/DC 56,358,687 1,414,917
14 Marilyn Manson 54,675,138 1,223,915
15 Queens of the Stone Age 52,970,037 1,273,775
16 Guns N Roses 51,825,756 1,673,338
17 Disturbed 48,310,602 1,116,727
18 Rage Against the Machine 45,688,191 1,556,924
19 Megadeth 40,307,528 729,446
20 Children of Bodom 38,347,666 562,446

Well, that’s a bit of a difference! Only 4 of the critics’ choice bands appear in the top 20 most listened. This is far from scientific, but it would seem that (Metallica excepted) metallers have a somewhat different perspective on what they think is of ‘quality’ and what they actually listen to. Clear critical favourites and genre instigators Black Sabbath are right down the rankings with 32,298,137 listens, lower even than perennial noughties favourites Opeth with 35,744,109 listens. And what’s with Korn in the top 15, I really didn’t think anyone seriously listened to them any more. And Nightwish? Really? The European influence I guess.

There’s quite a few caveats around this list, for example last.fm attract a certain demographic that’s perhaps not universally representative, plus song length will play a part (you can listen to 4 AC/DC songs in the time it takes to listen to 1 of Opeth’s), and this is last.fm’s lifetime list, so more established bands are more likely to appear, but it’s still indicative.

Anyway, I couldn’t make last.fm give me any data on overall lifetime listening stats for the metal genre, so calculating the long tail was impossible, and by now I was being distracted by the list I had compiled, and its various curiosities. For example, why the hell are In Flames so high? I realise they’re a pretty influential band, but they’re above Iron Maiden for gawd’s sake! The answer to this conundrum is in the listener figures. If you divide the total listens per band by the listeners you get a rough indication of listens per user. Order list by this value and you get a somewhat different picture.

Band Listens Listeners Plays per Listener
In Flames 77,571,473 648,985 120
Metallica 129,158,165 1,614,913 80
Nightwish 66,186,512 830,576 80
Nine Inch Nails 98,243,492 1,269,462 77
Opeth 35,744,109 462,064 77
Rammstein 74,874,781 1,080,014 69
Children of Bodom 38,347,666 562,446 68
Blind Guardian 26,270,993 407,215 65
Iron Maiden 71,102,754 1,118,305 64
Tool 63,722,501 1,021,247 62

As you can see, certain bands have more modest numbers of ‘fans’ with a propensity to listen to their music a hell of a lot. In Flames fans seem particularly enthusiastic about the band’s music. You’ll notice also that the list is considerably less ‘commercial’ with the likes of AC/DC, Slipknot and System of a Down disappearing from the top 10. This is explained when you look at the track play volume profiles of In Flames vs AC/DC:

In Flames

In Flames top tracks

AC/DC

AC/DC top tracks

The slope on the AC/DC is much more pronounced. Essentially, many more people are listening to a much smaller selection of AC/DC tracks, probably hitting Back in Black in their collection every so often, amongst Lady GaGa or U2. In Flames fans listen to a much greater selection of tracks, and probably favour albums over single tracks, thus driving the overall play count per listener up.

What’s also curious is that Black Sabbath, who dominate the critics album list, have a somewhat low plays per listener count of 27. Coming into this with no knowledge of the Sabbath you may assume, on viewing the critics’ list, that  they are an albums band, which is true to an extent, however, Sabbath’s profile is insanely skewed towards Paranoid and Iron Man which suggests that they attract a lot of casual listeners:

Black Sabbath play list

Anyway, I could noodle about with figures and stats all day, but where is this getting us? Well, we could surmise that the critics and fans aren’t really in agreement when it comes to voting with their feet. This is also evident in the public arena – Terrorizer Magazine’s albums of 2009 put Converge and Cobalt at the top and Behemoth somewhere in the 30′s, whereas the fan survey put Behemoth at the top of almost every category. On the other hand, Metalsucks famously triggered a minor insurgency when their poll of the top 21 albums of the 21st century – voted for by critics, bands and industry types – was mauled by their more discerning readership.

I’m really keen to understand whether metal is a long tail genre in comparison to, for example, Pop which I assume is by definition a populist genre, but decent stats are hard to come by. It’s a murky picture and one that dominated by dodgy data, countless caveats, spurious statistics, cliques and fraternities, but my quest will continue, to understand what makes the metal world tick.

Tags: , , , , , ,