Is Modern Music Boring?

by on July 28, 2012 at 8:08 pm in Music, Science | Permalink

Here, via Kevin Drum, is statistical evidence that modern pop music is boring or at least more homogeneous than in the past (yes, Tyler already linked to Kevin’s post but I wanted to link to the underlying dataset (see below)).

We find three important trends in the evolution of musical discourse: the restriction of pitch sequences (with metrics showing less variety in pitch progressions), the homogenization of the timbral palette (with frequent timbres becoming more frequent), and growing average loudness levels.

The picture at right shows the timbral variety:

Smaller values of β indicate less timbral variety: frequent codewords become more frequent, and infrequent ones become even less frequent. This evidences a growing homogenization of the global timbral palette. It also points towards a progressive tendency to follow more fashionable, mainstream sonorities.

The underlying data is from the Million Song Dataset which looks pretty cool and is open.

DK July 28, 2012 at 10:56 pm

File under “stereotypes are always right”. Although, to be fair, the graph above simply illustrates wider variety of genres being popular enough in the 1960s.

Rahul July 29, 2012 at 5:49 am

Could it be that a lot of this is an averaging methodology artifact? Let’s assume good songs have timbral variety and mediocre songs don’t. Over the decades a lot more songs get produced (mostly mediocre) therefore β falls.

Averaging over songs is irrelevant since probably 1% of songs get heard 99% of the time and this effect magnifies as your song base grows in size.

Tim August 1, 2012 at 4:58 pm

Wider variety of genres? I would argue that there were far fewer genres in the mid 1960s than there are now. Now when you go to describe a band, you practically have to describe down to sub-genres, they are so specialized. However, I think this specialization of genres allows bands to put out more homogenized music: nowadays a band only has to stand out from other members of the same, highly specific, genre.

Ray Lopez July 28, 2012 at 10:57 pm

Peak Music around 1969!

Rahul July 29, 2012 at 5:54 am

……and we still haven’t run out of it!

mulp July 28, 2012 at 11:45 pm

Just shows the impact of conservatives on even rock music. Diversity is not rewarded. Radical changes are quashed. Money and profit dictates what gets through the gate.

dan1111 July 29, 2012 at 4:13 am

When your theory leads you to the conclusion that conservatives are controlling rock music (!!!!), you need a new theory.

Millian July 29, 2012 at 6:17 am

Relative to other forms of music, rock is for conservatives. R&B? Get those kids off my lawn. Classical? That’s for those people with fancy coastal graduate degrees. Disco or rap? Don’t even go there.

TGGP July 29, 2012 at 1:58 pm

As both Will Wilkinson and South Park have pointed out, country is the conservative genre relative to rock.

But seriously, I think all the genres mentioned except classical (which is high culture rather than mass culture) were counted in the study.

ziel July 29, 2012 at 2:32 pm

According to the paper “the dataset includes the year annotations and audio descriptions of 464,411 distinct music recordings (from 1955 to 2010)…[that] span a variety of popular genres, including rock, pop, hip hop, metal, or electronic.” I can’t find anything more specific than that, though.

Millian July 30, 2012 at 12:02 pm

Country is like supporting Newt Gingrich. It means you’re very, very likely to be a Republican. It does not mean that everybody else is very, very unlikely to be a Republican.

dan1111 July 30, 2012 at 2:50 am

Conservatives may listen to rock, but the politics of the industry sure isn’t conservative. Just look at what happens when a Republican politician tries to use a rock song at a campaign rally.

Millian July 30, 2012 at 12:06 pm

Assume you’re a Republican who wants to do that. In the general case? I don’t think you’d have any problem. In the specific case where you are dumb enough to try to use a song by Bruce freakin’ Springsteen? Yeah, yeah you are going to have a problem.

Frank Lantz July 29, 2012 at 12:19 am

Maybe next they can do a statistical analysis of the kids who refuse to get off my lawn.

Eric July 29, 2012 at 4:35 am

Damn kids!

Andrew' July 29, 2012 at 12:21 pm
Barnley July 29, 2012 at 1:00 am

Maybe more genres were included in popular music in the past, but does this take into consideration ‘alternative’ music genres today which have audiences far larger than what was considered ‘pop music’ in the past? In other words were they listening to dubstep?

Bill S. July 29, 2012 at 1:21 am

As with all such things, I question their metrics for ‘boring.’ I have a hard time distinguishing between various ‘British Invasion’ groups, but you don’t see me whining about it. Also, go back further: Big Band is awful boring. So is Baroque. Stupid studies are stupid.

dan1111 July 29, 2012 at 3:34 am

It’s worth pointing out that the original study never makes such a claim. It simply observes that there is a trend toward less variation in pitch, timbre, and volume. I agree that using these results as a metric for “boring” is flawed. Though I do think modern pop music is boring.

Willitts July 29, 2012 at 5:19 pm

I agree. If anything, the music of the 60s was remarkably similar in its differences from established genres. It appears someone selected a particular measure to achieve a particular conclusion.

I grew up on 60s music, and I can’t escape the fact that its distinction was artificially induced by anarchy, rebellion, shouts for attention, and drugs. It certainly wasn’t boring, but it really made no sense.

Love Me Do is only slightly more sophisticated than a Justin Bieber song. Lucy in the Sky with Diamonds is nonsense.

Country songs, old rock, Blues, Gospel, all had coherent messages. Patsy Cline had much more to say than Brittany Spears. Michael Jackson and Elton John are probably exceptions.

Alex' July 29, 2012 at 8:39 pm

” Lucy in the Sky with Diamonds is nonsense.”

Only if you’ve never done acid.

justin July 29, 2012 at 1:51 am

These studies are meaningless to me. Of course pop music today is probably simpler than ever but you can find plenty of complex music if you look for it. Given how easy it is to access music online these days, I can’t imagine someone from 1950 had an easier time finding a complex album released that year than someone today would.

Ray Lopez July 29, 2012 at 1:59 am

I don’t think the paper was meant to be telling you what to do, just reporting the facts. What you do with the facts is up to you. As you say the standard deviation from the average may be smaller today but the overall size of music out there is bigger.

dan1111 July 29, 2012 at 3:37 am

I would speculate that the wider availability of music today is actually driving this trend. Today, music is available through so many alternative sources that those who are looking for variety are no longer consumers of pop music at all. As a result, pop becomes even more homogeneous and conventional than it was in the past.

ziel July 29, 2012 at 11:48 am

That’s an interesting point. Back in 1967, FM radio was in it’s infancy and home high fidelity stereo systems were still rare. So AM Radio and records (45’s) were still pretty much the only game in town. So you had lots of different interests competing in the “Pop” market – so you had both The Doors and Frank Sinatra in the Billboard top 10 for that year.

Willitts July 29, 2012 at 5:24 pm

Yes, barriers to entry were stronger in the past, and average distribution costs were higher, so scarce resources had to be conserved for the best assets.

Today, you can put a song on YouTube or go on a TV show and become a star. Every high school in America has aspiring rappers of every color.

The improvement of editing technology has turned poor to mediocre singers into studio sensations. They are selling an image, not a song.

Rahul July 29, 2012 at 6:19 pm

Is that an argument for the social utility of barriers to entry in certain contexts. I thought barriers to entry were always bad for society.

ziel July 30, 2012 at 12:16 am

I thought barriers to entry were always bad for society.

Ha! If nothing else, this is one market failure that surely can’t be pinned on government interference.

Doc Merlin July 30, 2012 at 7:55 am

Incorrect. The radio music market and various other music markets markets have federally mandated price fixing and contract fixing.

Anonymous42 July 30, 2012 at 10:48 am

Let’s grant monopolies to artists for 80 years after they die. No way will that harm culture, by giving the established the power to sue the new. The truth will win in court regardless of who can afford the most lawyers…

Right?

Matt July 29, 2012 at 2:00 am

It’s definitely not clear that a reduction timbral variety entails more boring-ness. More boring music would be simply music which makes people feel more bored than some other sort of music. You could study that, by playing people music and trying to figure out how bored they are. This is not that study.

ziel July 29, 2012 at 11:50 am

Again, the study did not make that claim – that was Alex’s editorial comment. Boring though can be used in different senses – the one you seem to be using is “lack of excitement”. Another is “quality of not being very interesting” which is the sense I think Alex had in mind.

ziel July 29, 2012 at 11:54 am

Sorry – it was actually Kevin Drum’s editorial comment, but Alex affirmed it.

Matt July 30, 2012 at 10:08 am

Like the property of ‘being boring, the property ‘not being very interesting’ (I’d probably have said these two are co-extensive) is still dependent on some actual or hypothetical observer. The move from ‘lacks timbral variety’ (or some other objective fact about the music) to ‘boring’ or ‘not interesting’ requires a premise about how boring or interesting listeners find certain intrinsic properties of music.

Nyongesa July 29, 2012 at 2:48 am

I cringed at the heading, as i suspect my age sorely biases my take on this, But, from what I can tell of POP music, which is what the vast majority of people are hearing, the club hits or top 10 is composed of, have allot of converged sounds, techniques and rhythmic qualities. The advent of hit producing machines, beginning with the famous Motown system, has evolved considerably since. A new sound or musical sense, regeatone for example, or, trance, that emerges within the club scene (where there’s real creativity), is rapidly re-packaged for global consumption by the music industry. Cross pollination has accelerated to a fever pitch, with sound engineering, and assembly of dance, star, and marketing machines cycling much faster. The result is much more homogeneity at the top. Pop stars in Bulgaria, South Africa, and Thailand, are using the same timberland riff, and synth sounds of the latest “pop” sound. That said, the base globally is getting much wider and deeper. Many countries around the world have growing homegrown music scenes, beyond traditional music, with good production values, which used to be the pervue of the U.S. Music Industry. I was quite impressed/depressed by the local music video channels during a recent trip to East Africa, Allot of new music and new sound, but also allot of the same lame booty girls as well.

Hadur July 29, 2012 at 3:26 am

Or science has discovered the ideal set of stimuli to please the human ear, and thanks to AutoTune can now hit it over and over again?

Rahul July 29, 2012 at 5:57 am

“growing average loudness levels”

Does recorded music come with an encoded absolute loudness level or only relative levels? I thought absolute levels were a function of your amplifier / speakers?

ziel July 29, 2012 at 11:17 am

The “loudness” is a reflection of dynamic range – the higher the loudness, the less dynamic range. From the MSD documentation:

Overall song loudness is a formula combining segments: local maximum loudness, dynamic range, overall top loudness, and segment rate. The greater the dynamic range, the more influential it is on lowering the overall loudness. As a result, highly compressed music sounds louder than non compressed music, even if their maximum loudnesses are similar.

gwern July 29, 2012 at 12:49 pm

My understanding of the loudness wars is that it is the latter, and the ‘average’ here is the average relative level – more time spent at 90%-100% part of the range (peak loudness), say, and less time spent wandering around 0-50%.

Doc Merlin July 30, 2012 at 7:56 am

It has both.

Millian July 29, 2012 at 6:08 am

I think this is much, much closer to the truth than the interpretation provided by nostalgic journalists.

Piros July 29, 2012 at 6:47 am

oh yes, now autotune is just waiting for a genius of the caliber of Björn Ulvaeus and then there will be no variation in music at all.

freethinker July 29, 2012 at 4:03 am

Ray Lopez says “I don’t think the paper was meant to be telling you what to do, just reporting the facts”. So branding something as “boring” is just reporting a fact?

ziel July 29, 2012 at 11:20 am

I guess being a ‘freethinker’ means never having to click on a link to see what a paper actually says.

Larry Siegel July 29, 2012 at 4:43 am

So the peak of musical complexity was around 1965? Thanks for confirming that the Beatles, Bob Dylan, and the Jefferson Airplane are more interesting than Katy Perry. I kind of thought so.

Barry Oblivion July 29, 2012 at 6:03 am

The author clearly needs to work on the definition of what makes “Boring”.

Boring != variety in pitch | Loudness.

I always have considered the 1960s the dullest decade. Verse chorus verse.

Bill Benzon July 29, 2012 at 6:45 am

As far as I can tell the data set doesn’t contain much information about rhythm, though there is an estimate of time signature, which is, at best, the merest beginning of rhythmic analysis. Without rhythmic analysis, assertions about boring-or-not aren’t worth much.

Millian July 29, 2012 at 6:57 am

The method is privileging some kinds of diversity, but not others.

One perceptive person on Kevin Drum’s thread notes that rap should do poorly on the measures used by this paper. Yet rap is much more diverse (and intelligent) than the “love baby love baby love” songs of the 1960s, on average. It’s just that the types of diversity within hip hop aren’t all captured by this methodology.

ziel July 29, 2012 at 11:36 am

I agree adding some rhythmic analytics would add to the discussion. But do you seriously believe today’s popular music would show more rhythmic diversity than the 1960’s? For example, the top 10 songs of 1967 were “To Sir, with Love”, “The Letter”, “Ode to Billie Joe”, “Windy”, “I’m a Believer”, “Light My Fire”, “Somethin’ Stupid”, “Happy Together”, “Groovin”, “Can’t Take My Eyes off You”. That’s quite a varied collection of rhythmic styles.

Jim K July 29, 2012 at 3:50 pm

And, in large part, bubblegum pop.

ziel July 29, 2012 at 4:06 pm

No they weren’t, with the possible exception of “I’m a Believer” as the Monkees were themselves a ‘manufactured’ band – but applied to that particular song that’s a stretch. You may not like them, but the pejorative “bubblegum pop” does not apply.

Doogie July 29, 2012 at 10:49 am

Could the prevalence of MP3s have an impact here? Anyone producing music with a wide variety of pitches would see much of that lost in the compression process. Producers are creating music that will get through MP3 compression without substantial change, thus songs without high and low ends.

Boring is a different question though. It can’t be easily measured and thus proxies, such as this one, are used. This long-time listener says that hit music is without question more boring than it was in the 60s/70s, but there is plenty of under-the-radar music that is still quite enjoyable.

Sbard July 29, 2012 at 10:22 pm

It’s not compression to mp3 format per se, it’s that pop music is mastered largely under the assumption that it will mostly be listened to on cheap iPod earbuds. Classical and jazz recordings, on the other hand, tend to be mastered for an older demographic that will likely be listening on higher end equipment.

Ed July 29, 2012 at 11:09 am

“Peak music” may be correct though. It could be that there are just so many combinations of notes that are pleasing to people.

Willitts July 29, 2012 at 5:28 pm

Centuries of classical music would seem to disagree.

On the other hand, maybe it is we who have changed. I would describe most modern music as ‘mind destroying’. Why are people so inclined? Maybe the rest of the world has become so uninteresting or perhaps real life is now considered too hard. Music has become a drug.

Millian July 30, 2012 at 12:11 pm

Why would you describe it as mind destroying? It’s not obvious that the modern era has lower intelligence levels than the 1800s.

ziel July 29, 2012 at 11:41 am

Could the prevalence of MP3s have an impact here? Anyone producing music with a wide variety of pitches would see much of that lost in the compression process. Producers are creating music that will get through MP3 compression without substantial change, thus songs without high and low ends.

No. MP3’s have far greater dynamic range than 33 1/3 lp’s – never mind 45’s and AM radio, which were the distribution medium for songs in the 1960’s.

Trent July 29, 2012 at 11:56 am

I’d like to get hold of the data behind Pandora/the Music Genome Project (http://www.nytimes.com/2009/10/18/magazine/18Pandora-t.html?_r=1&pagewanted=allhttp://en.wikipedia.org/wiki/List_of_Music_Genome_Project_attributes) and compute some diversity/entropy statistics to really look at music homogeneity/innovation. Something could be done similar to Frenken K, Saviotti PP, Trommetter M (1999), where they looked at ‘variety and niche creation in aircraft helicopters, motorcycles and microcomputers’ (http://www.sciencedirect.com/science/article/pii/S0048733399000086). Unfortunately this data is proprietary as it backs up Pandora’s internet radio site and they aren’t keen on sharing (I asked, but they wouldn’t budge).

FWIW I think that while mainstream/pop music might have become more homogenized since the 70s (as per above), the overall diversity of musical output (using such a ‘characteristics approach’) must have surely increased. I think Pandora/MGP + Frenken et al (1999) approach would be a good way of trying to answer this one.

ziel July 29, 2012 at 12:25 pm

the overall diversity of musical output (using such a ‘characteristics approach’)

What does that mean?

Trent July 29, 2012 at 1:24 pm

I don’t know, I would have thought many artists are playing and releasing similar music (qualitatively, i.e., same genres, styles) to earlier periods. On top of that, there’s all manner of crazy electronica and avante pop that is being produced with completely different qualities and characteristics. (Not to mention fusions and micro-genres.) If you were to look at everything that’s being created and released I think you would see more variety (a cumulative effect). But then I suppose diversity is also a function of population/niche size.

Perhaps you could say: more variety, less diversity. So while the mainstream (in terms of pop charts) might be homogenized, overall there is be more variety produced. I have to agree that modern pop music is boring though. But the fringes are vibrant, more so then they have ever been.

ziel July 29, 2012 at 2:28 pm

Yeah it’s hard to tell what exactly was sampled – was it just “the most popular” music of each era or just whatever’s in this “million song databas”. I would agree that we’d expect current available music to be much more diverse than in the past, but since the topic is “Popular Music” you’d expect that popularity was taken into account somehow. The paper suggests that that’s what they did, but not real clear how.

dirk July 29, 2012 at 2:38 pm

Surely it’s a statistical artifact that music made for the generation that was obsessed with music is less boring than music made for the generation obsessed with video games.

Millian July 30, 2012 at 12:12 pm

The video games were a lot worse back then, though.

Keith Williams July 29, 2012 at 5:06 pm

Offhand I’d guess that this has absolutely nothing to do with actual variety in music. It merely shows the transition from analogue to digital.

“[The period from the late 1950s into the early 1960s] is most widely regarded as “The Golden Age of Hi-Fi”, when tube equipment manufacturers of the time produced many models considered endearing by modern audiophiles, and just before solid state equipment was introduced to the market, subsequently replacing tube equipment as mainstream.”

http://en.wikipedia.org/wiki/High_fidelity

DK July 29, 2012 at 8:24 pm

I had a conversation like this with a friend of mine who is a write. It’s the difference between a good book, and good writing. A good book is one that sells, because that’s the point of doing the work. Good writing might get you a lecturing gig, but often isn’t commercially exciting. Early on in the rock and roll era, there was not only much more experimenting, but many of those artists were trained in the jazz of the 40’s and 50’s and a more sonically diverse palette was part of their experience. But since the ’70’s, corporate music has been more dominant than ever and they don’t care about what’s good, they care about what sells. Just look at the music trade press – almost all ads are for “Do you sound like XXX? Send us your demo! In fact, before you can submit a song to an agent or label anymore, you have to identify who you sound like. So, I don;t blame art for this one. I blame commerce. We’re just giving people what they want. Music as commodity.

The good news? Cost of producing music has gotten so low, we’re likely to see more diversity in the future. Downside, is that any yahoo can make and publish a record now, so pouring through the chaff is going to be harder. And since you can’t get rich selling music anymore, the best artists will probably go into finance anyway.

Joseph Ward July 30, 2012 at 1:39 pm

I did a recent article on my blog based on some writings by Dr. Joel Waldfogel of the University of Minnesota and NBER. He wrote an article wondering if music quality had gone down since Napster, and basically concluded that it had not, and had actually gone up slightly. When I looked at my own iTunes data, I found similar findings, but at the top level of music quality, there had been marked decline since the late 1960’s. Here’s a link: http://makingofaneconomist.blogspot.com/2012/06/has-music-quality-gone-down-post.html

beefheart August 2, 2012 at 11:30 pm

Trout Mask Replica was ’69, so this makes perfect sense.

Comments on this entry are closed.

Previous post:

Next post: