[NERD STUFF] Number of metal bands and genre statistics over time

[einride]

or

why it is a statistical fact that metal is dying​

I like statistics. I also like theorizing about metal and the state of it. So I wrote a script that harvests stats of how many bands are started each year, and what genre they are, from Metal-Archives, in order to obtain data about which genres have historically been the most popular and how popular metal as a whole is and have been throughout the years.


METHODOLOGY

• I used a little Python script to scrape data from Metal-Archives. I searched for bands categorized as each genre, by year formed, starting from 1967 and ending in 2020.
  
  
• I separately searched for what I call "modifiers" like "atmospheric", "epic" etc. which can apply to different genres but have their own trends throughout the years.
  
  
• The data has been color coded in the tables with a scale going from red to green, so that green represents "good years" and red represents "bad years" for each genre.

GOTCHAS

• The total per year is not the same as the sum of all genres. This is because 1) I don't actually cover all possible genres, and 2) some bands belong to multiple genres. The total per year is obtained by searching for bands started a particular year, without genre tag.
  
  
• Because Metal-Archives requires at least one official release to accept a band, the falloff in the last couple of years is a lot sharper than it probably should be, because bands formed in 2018 or 2020 might not have gotten around to releasing their first demo or album yet.
  
  
• This also means that we do not adjust for how long it takes to release an album. Maybe it is true, for example, that black metal bands are faster than death metal bands to release their first thing? I don't know, the data does not show this.
  
  
• Metal-Archives has only one field for genre -- this is why you see thrash metal appearing already in the 1970's. There are a lot of bands that evolved and switched genres throughout the years, but we don't have data of that granularity, because they don't assign genres to albums, only to bands.

With that said, let's get to the data:



GENRE POPULARITY BY ABSOLUTE NUMBER OF BANDS STARTED PER YEAR

ANALYSIS

• Metal as a whole peaked in volume 2005; except for a slight but noticable local peak in 2012 (if you have a plausible explanation for this increase, please post your theory) we have been on a constant downward slope in terms of absolute number of bands started since then.
  
  
• Once metal as a whole discovers something more extreme, it never looks back again. Death metal overtook thrash as soon as it was discovered, which in turn had quickly overtaken heavy metal.
  
  
• Black metal finally overtook death metal in 2018 and trends towards keeping the lead.
  
  
• After languishing in relative obscurity for most of its existence, doom metal is now the third most popular subgenre.

GENRE POPULARITY AS PERCENTAGE OF METAL AS A WHOLE

ANALYSIS

• Over 50% of the bands started in 2020 are some kind of black metal.
  
  
• The "death metal trend" of the early 90's that Norwegian black metal rebelled against is very apparent in the data -- in 1991 over 52% of bands started were death metal. This is the same year that Darkthrone released "A Blaze in the Northern Sky" and started the rebellion.
  
  
• You can see black metal becoming a trend and peaking briefly in 1996-1997, right as the second wave was coming to an end and most of the Norwegian bands were winding down.
  
  
• Note the peak in doom metal around 1993 -- this is right around when the death/doom movement led by Anathema, Paradise Lost and My Dying Bride was gaining prominence
  
  
• Industrial metal had an early peak around 1992 in the echoes of Godflesh "Streetcleaner" et al, and then was a thing again around 1998-2000, right when The Matrix and The Internet and all kind of techno-fetischism reigned supreme in the cultural consciousness. This was the time of "666 International", The Kovenant, "Rebel Extravaganza" etc.
  
  
• Plain old heavy metal was the most popular style for quite a while, but as soon as there were faster and more extreme styles around, it was never the most popular genre again after being overtaken by thrash metal in 1986.
  
  
• Thrash metal, unsurprisingly, was a huge trend around 1984-1989 but had to yield the crown to death metal in 1990. You can also see the thrash metal resurgence that happened around 2003-2009 quite clearly. It is now very untrendy, however.
  
  
• Sludge metal is the most modern of all the genres in the sense that it barely existed before the 2000's but has become a sizable chunk of the total of new bands in the 2010's.

POPULARITY OF GENRE MODIFIERS

ANALYSIS

• "Depressive", "atmospheric" and "post" have all been very popular the last decade or so. Color me unsurprised.

discuss

 

[NAD]

Holy fuck dude.

Wait, what sludge band was around in 1979?!

 

[einride]

Holy fuck dude.

Wait, what sludge band was around in 1979?!

no that is an artifact of the fact that there is only one genre field for bands that have been active a long time, so they probably did not in fact play bona fide sludge metal in 1979

the answer to that particular question btw is:

https://www.metal-archives.com/bands/Graveyard_Rodeo/605

apparently

 

[NAD]

And they are from Nawlinz, but of course.

 

[swizzlenuts]

Can I have your script Erik? I think your normalization is wrong, but it's just a hunch. Also shouldn't it be albums/eps a year not band creation?

"After languishing in relative obscurity for most of its existence, doom metal is now the third most popular subgenre."

Fucking knew it. That's why it sounds so shit now a days.

 

[Krilons Resa]

What the fuck is going on?

 

[Ellestin]

Why did you post the cover of the new MFJ album itt?

Great job! Books about metal have been written on less relevant basis.

Could the 2012 aberration be explained by the reformation of Black Sabbath? Them counting for 666 bands and all.

 

[einride]

Can I have your script Erik? I think your normalization is wrong, but it's just a hunch.

i will post the script and the spreadsheet later so you can play with them if you wanna but the script barely does anything. it's like ten lines and pulls raw data from metal-archives and spits out csv. then i pasted it into openoffice calc and massaged it until it looked halfway presentable. this is not peer reviewed science, this is cowboy statistics from someone who barely knows enough to be dangerous

Also shouldn't it be albums/eps a year not band creation?

well that depends on what you're trying to measure innit

band creation to me indicates whether the kids are into it

if you're trying to find out the overall health of metal (i.e. sure, the kids don't really start as many bands as they used to, but on the other hand, judas fucking priest is still releasing albums), or even a better indicator of relative genre popularity, then yes it would be better to look at releases probably, but it is problematic to pull that data from metal-archives because it will for instance count every katatonia album as death and doom metal because albums are not tagged by genre, only bands

if you wanna scrape that data, and i would like to look at that too, it would be better to get it from something like rateyourmusic probably




EDIT

#!/usr/bin/python3

import requests, json, time

headers = {
  'User-Agent': "Mozilla/5.0"
}

genres = ['heavy', 'doom', 'thrash', 'black', 'death', 'folk', 'power', 'alternative', 'viking', 'speed', 'industrial', 'gothic', 'groove', 'sludge']

for genre in genres:
   for year in range(1967, 2021):
      response = json.loads(requests.get("https://www.metal-archives.com/search/ajax-advanced/searching/bands/?genre={}+metal&yearCreationFrom={}&yearCreationTo={}".format(genre, year, year), headers=headers).text)
      print("{},{},{}".format(genre, year, response["iTotalRecords"]), flush=True)
      time.sleep(1)

 

[swizzlenuts]

i will post the script and the spreadsheet later so you can play with them if you wanna but the script barely does anything. it's like ten lines and pulls raw data from metal-archives and spits out csv. then i pasted it into openoffice calc and massaged it until it looked halfway presentable. this is not peer reviewed science, this is cowboy statistics from someone who barely knows enough to be dangerous

of course, i love this shit too.

band creation to me indicates whether the kids are into it

fair enough, though a bit naive.

Thanks for the script.

 

[NAD]

Band creation shows enthusiasm and that's really the driving force behind all this, right? At the beginning of 2020 at the NAMM show (remember crowds? wow, man, wow), I watched a group of very young kids (8 to 10 years old) going nutso in the Jackson Guitars booth wearing '80s era Slayer and Metallica shirts and I looked at my fellow oldmanheadbanger pal and said "look dude, hope for the future!"

Consistently speaking, the origins of a Band or Genre typically bring about the best shit. Yes there are bands that peak late in their career, and yes there are genres that are perfected many years down the line, but these are the exceptions and not the rule. The most enthusiastic releases happen early on for both, because the excitement, drive, and hunger are all major parts of this. You haven't heard it all before if a new idea is being presented.

I don't recall who the axiom is properly assigned to, but it goes something like: a band has their entire life to write their first album, but only a year or so to come up with the second. Led Zeppelin II might be a better album on paper but the first Led Zeppelin just sounds so much more LIVELY.

 

[einride]

WHY METAL IS ACTUALLY IN VERY GOOD HEALTH THOUGH

so i took a look at the number of releases per year too and there are A LOT of releases these days, despite the dwindling number of new bands. to listen to all the metal in 2020 you would have to listen to over 60 releases per day for the whole year on average. and i have no idea about the average length of a release but i think this probably means that it is now literally impossible to listen to everything in the same year it was released, even if you spent 24 hours a day doing it. good luck with your year end lists folks.

curiously, until about 2004, right about when the number of bands started to peak, there were fewer releases per year than bands started. now it's the other way around, by a lot. so this means (maybe) a couple of things:

1) metal bands tend to stick around for a long time and keep releasing albums

2) much it's easier to release music nowadays (also metal-archives accepts digital streaming/download releases now so the bar is a lot lower than it used to be)

3) also there are probably a lot of re-releases and collections of older shit kind of skewing the numbers

also, for shits and giggles, i included a graph of releases per number of bands, cumulative since the beginning. i don't think this actually says very much because there's no way to tell how many of these bands are active but graph's tending up so stonks i guess

 

[swizzlenuts]

yeah, that was what I was thinking when I was talking about normalization.

can't wait to have some time to play around with this script. Fuck being a parent.