Thirteen years ago, I started a project called FAWM (February Album Writing Month) with a few friends. It was a way to motivate ourselves to make more music. The goal is to write 14 songs in the 28 days of February, much in the same way NaNoWriMo challenges you to write a whole novel in a month. Maybe it’s crappy, maybe it’s not. Don’t overthink it, just do it!
Inspiration is for amateurs – the rest of us just show up and get to work. ~ Chuck Close
If you’re curious, here is a playlist of my favorite FAWM demo recordings from the past (if you’re familiar with my band delicious pastries, you may recognize nascent versions of a few tracks we reworked for later albums):
FAWM Growth Over Time
Last week marked the end of FAWM 2017, and while I was too busy to take on the challenge myself this year, I kept the website running for others. (Big props to fawmers Eric and Jen Distad in particular, as well as Nancy Rost and Helen Robertson for helping out, as always.) It was a record year for sure: 11,239 new songs that didn’t exist in January! In the early days of the project, I was pretty sure FAWM would keep growing quadratically from year to year, but things leveled off pretty quickly after 2010:
My theory as to why we leveled off has to do with “FAWM Escape Velocity.” If there are 28×24×60 = 40,320 minutes in February, and the average song is 3.5 minutes long, then after 11,520 songs we will have created more music than is actually possible to listen to in the month of February… and that’s assuming you listen non-stop! This seems to be a metaphysical barrier we just can’t penetrate. (Credit goes to fawmer Tim Wille for coining the term.)
So, Who Keeps FAWM Going?
Here is a cartogram/choropleth map showing where “fawmers” came from, and how much they’ve donated to help keep the project alive (as of this writing). The size of each country represents the its active “population” in 2017, and the redness scale indicates how much (in US dollars) people in that country donated per capita. (Note: Song output and most other activities were extremely correlated with “population,” so I plotted donations as an interesting orthogonal variable).
The population/size seem to correlate roughly with Internet access crossed with English speakers (although there are substantial Finnish and German speaking FAWM sub-communities). Donation rates don’t seem to track with the Big Mac Index, so who knows what socioeconomic factors are at play there. In general, the more anglophone the country is, the more they seem to give… although Spanish fawmers were the most generous this year, so who knows? Overall, fawmers give $4 per capita (about 20% are donors), which is enough to pay the server bills and occasionally order Tshirts, stickers, guitar picks, and other cool FAWM swag from time to time.
The Social Network
The hairball graph below visualizes the FAWM social network. Each node represents a fawmer, and each edge indicates a fawmer who left feedback on the song(s) of another. The edge thickness denotes how many comments were given, and the size of each node reflects that fawmer’s PageRank (a measure of how “central” or “influential” s/he is in the network).
The graph itself is fairly dense: there is only one connected component. Each fawmer gaves 48.8 comments on average, and is only 2.7 links away from any other person in the network, on average. (In fact, the longest path from one fawmer to another is only 7!)
At first glance, there don’t appear to be too many cliques in the network. But to test that hypothesis, I ran a community detection algorithm which did discover seven subgroups of fawmers, based on how their commenting behaviors cluster them together. These subgroups are color-coded in the figure above. While these groups do appear to capture something about the network (e.g., the green subgroup has more high-PageRank power users, and all but the red and yellow subgroups seem to cluster pretty tightly together), I’ve had trouble characterizing them or trying to understand what may have caused them to form. Geographical location, number of songs written, number of previous FAWMs, amount donated, etc. — none of these variables seem significantly associated with any particular subgroup (the best model I could come up with is a logistic regression that only does 4% better than random guessing at classifying fawmers into these subgroups). So either the detection algorithm is bunk (not likely), or the forces at play can’t be explained using the variables I have immediately on hand in the database. I only threw a couple of hours at it, though, and maybe something like musical genre (e.g., song tags) can explain it. At any rate, it makes a pretty graphic!
The last thing I want to talk about is collaboration, which is a huge part of FAWM culture these days. “Collabs” started in 2008, which was a leap year with one extra day. We jokingly decided to make the challenge “14½ songs in 29 days,” with the extra ½ song being a collaboration. Roughly 8% of all FAWM songs have been collabs ever since. I’ve done some previous research into how these form and what makes them successful.
FAWM hit a new record in 2017: 943 collabs! Since most fawmers provide an (optional) location field, I used the Google Maps API to pinpoint them on a map and plot all the collab-connections on a map like airline flight routes:
Lots of trans-Atlantic co-writing this year — which isn’t surprising, since most fawmers are in North America and Europe — but plenty that stretch out into Africa and Oceania, too. There’s even someone way out in the Kerguelen Islands who got in some sweet collab action. (Credit goes to fawmer Scott Lake for suggesting this visualization.)
It’s been rewarding to help create something like FAWM which brings people together to make music, and I’m happy to see it still going strong thirteen years later. Next year — FAWM’s fourtheenth year — is gonna be a blowout!