Since posting ‘Solving The Discovery Problem: New Music That Sounds Like Old Music’ on Friday (worth going there first before reading on if you haven’t already), a couple of things drifted through my transom that seemed worth sharing.
On Monday a chap named Brian Hazard followed me on Twitter. Intrigued by the juxtaposition of such a conventional first name with the more thrusting derring-do of his last, I took a peek at his profile. Turns out Brian is an EDM artist who performs under the name Color Theory; he has 1.9m Twitter followers and describes his music as “80’s synthpop = EDM + piano for fans of Depeche Mode”.
Given the particular suburb of the discovery landscape I’ve mentally inhabited the past week or so, the specificity of his description piqued my interest. On the one hand: as a popular artist that directly references both a decade and a heritage act as major influences in their Twitter profile, Color Theory seemed ripe for the same analysis to which I subjected Daniel Romano and others in the original article.
On the other: if you have two million followers on Twitter, perhaps you’ve reached the point where you don’t need to describe the sound of your music to potential new followers. (Although it’s worth pointing out that Hazard follows more than a million people as well. I’m not suggesting foul play; just that popularity on Twitter and Spotify are two very different things.)
All of this got me thinking more about the true nature of ‘popularity’ in the streaming space, and in particular the Spotify popularity index I overlooked in the first piece. To judge from a selection of Color Theory’s biggest songs, there is indeed a resemblance in sound to Depeche Mode, so I pulled both artists’ numbers – just as I had for Romano and Simpson in the original article:
To deal first with their release windows, Color Theory’s first album Sketches In Grey came out in 1994, while his most recent came out this year. His related artists have an average release era of 2002 to 2011. Depeche Mode’s seminal synthpop debut Speak & Spell came out in 1981, their most recent in 2013, while the release era of their related artists averages 1983 to 2012. So Mr. Hazard has been releasing music almost as long as his heroes Depeche Mode. He even released a full-length covers album in 2003: Color Theory presents Depeche Mode.
Things get more interesting when you look at popularity. Color Theory has 916 followers on Spotify and a popularity score of 17 out of 100. His related artists have an average of 270 followers and a mean popularity of 10, so he slightly outranks his peers. Depeche Mode have 612,544 followers and a popularity index of 72, contrasted with 111,045 and 56 respectively for their similar artists. It’s worth noting that, as with the artists in the first article, neither Color Theory nor Depeche Mode appear in each other’s related artist lists.
Spotify describes popularity thusly:
The popularity of a track is a value between 0 and 100, with 100 being the most popular. The popularity is calculated by algorithm and is based, in the most part, on the total number of plays the track has had and how recent those plays are. Generally speaking, songs that are being played a lot now will have a higher popularity than songs that were played a lot in the past. Artist and album popularity is derived mathematically from track popularity.
So Color Theory’s much lower popularity score of 17, compared to 72 for Depeche Mode, is a reflection not just of the fact that his music is played less than theirs, but that it’s being played less now. Probably Spotify’s related artist algorithms are a blend of both recording era and popularity. (If anyone from Spotify would like to confirm, feel free.) By way of comparison, Adele – whose new album 25 isn’t streaming on Spotify – has a popularity ranking of 92.
It makes sense that Spotify would want to factor in currency as a determiner of popularity. But as I pointed out in the original article, it does leave New Music That Sounds Like Old Music out on a limb somewhat, confining similar/related artists more to comparable consumption cycles (e.g. early adopters vs. laggards) than proximity of sound. There’s a downside to this for both the artist and the listener I think.
Of course not every new artist that thinks they sound a bit like an older, bigger one has a God-given right to a chunk of the latter’s audience. Some of them just might not be very good. (Some services disagree on this; a smart radio provider I worked with several years ago offered labels and artists paid-for ‘power plays’ allowing them to define which artists their music appeared alongside in radio streams. The practice made me distinctly uneasy; payola by another name.)
So if neither algorithms nor cold, hard cash can be relied upon to determine similar artist accuracy, I guess that brings us back to humans. How about a system whereby new artists themselves suggest a handful of possible related artists, which genre experts at the services have the final word on? Or maybe use the wisdom of the crowd and get listeners to do (or help with) that? Perhaps this happens already – I’d be similarly happy to hear from you if it does.
The second thing that happened in response to the first article this week is that I heard from Daniel Romano’s PR in the UK, Del Day of Ark PR. With his permission I’m publishing the main points of his email unedited below:
“The PR plot for the new album was a huge success at press and radio with a lot of online site support too. I think this is down to a few factors – the fact the record is undeniably brilliant, the fact that Daniel is an ever-evolving musician that isn’t always easy to label which helps pitch him across a few different genres, and good old word of mouth. (Obviously I worked my socks off as well which helps..) 🙂
I’m intrigued by your piece and, although I can see your points regards Daniel and Spotify and Pandora, I also think internet radio, especially the shows based here in the UK have done loads to help you find his music. We had massive support right across the board on the album from all the key roots/country shows.
As a publicist working almost entirely with country/roots acts, we got as much coverage we could for Daniel. I honestly don’t think we could have got any more. Reviews in five broadsheets, all the key music magazines, a six page feature in CMP arguably our only dedicated country magazine, numerous plays on Bob Harris Country, and reviews/plays on all the relevant country press/radio. I think a hell of a lot of people still ‘find’ artists this way.”
We may have been talking slightly as cross-purposes about ‘internet radio’; Del seems to be talking about webcast linear radio – i.e. with shows – whereas I was using the phrase in the sense of non-linear, ‘smart’ services like Pandora.
But his comments are interesting nonetheless, and not just because they reveal that I somehow managed to miss what sounds like a stellar PR campaign. It’s linear radio, press and online still doing all the heavy lifting for new artists like Daniel. Radio still outranks all other media, including YouTube, for music discovery; streaming will have to do better if it really wants to increase its share of ear.