Should data analytics stay out of the creative arts?
Disclaimer: my blog post this week will have a more personal tone than I would normally use for an assignment. This post uses music as a proxy for other creative fields – I think the general logic here applies to visual arts, fashion, and even food and beverage, among others.
Data analytics, artificial intelligence, machine learning – we’ve been hearing these buzzwords everywhere recently, and not just in the context of MS&E238. As is the case with any trend, modern society’s apparent obsession with these topics has permeated the lives of almost everyone, not just the technological gurus developing new algorithms or the entrepreneurs capitalizing on the opportunity. Besides concerns related to privacy and security, one could argue that most applications of analytics are beneficial to society, as we’ve been discussing extensively over the past several weeks. However, I would like to make a case as to why big data may not always be helpful for all industries, especially creative ones.
Spotify, YouTube, and Apple Music are some of the most popular music streaming services today. Millennials praise these services as the best thing since Limewire (long-gone are the days of rampant music piracy), and it’s almost become common knowledge that their recommendation engines use numerous algorithmic models to figure out what each listener might like the most. Spotify’s Discover Weekly is the most notable example – it leverages features from your listening habits, others’ listening habits, textual metadata about songs, albums, and artists, and even the audio files themselves [1]. Personally, I’m mostly ambivalent towards automated playlist creation because I enjoy library curation, but I think most people would agree that having software to help us figure out what we might like listening to is generally helpful in these busy times. However, what most listeners might not see is what goes on behind the scenes when it comes to payouts and royalties. Morally speaking, streaming services will always be better than music piracy, but by how much? In 2015, the average pay per stream through Spotify was only slightly more than half a cent [2]; in other words, an artist was paid less than $0.01 each time their song was played. So, if I played my favorite album 50 times in a year (let’s assume there’s 10 songs on that album), the artist would still be making significantly less than $5 from that. All of this is assuming that I’m paying for a subscription of Spotify Premium – the payouts to the artists would be even less for free users.
Of course, we still see headlines about how lucrative streaming services can be for artists. Drake earned $15 million from over 1.8 billion streams in 2015 [2], for example. But what about for smaller artists that don’t have an established fanbase? Artists that are living paycheck-to-paycheck, trying to ‘make it’ in an industry that, to them, seems to only care about the number of streams you can get? Or artists that happen to thrive in genres and communities where they can’t produce their own music in their bedrooms quite as easily? A rising hip-hop producer or rapper might be able to make an EP in her bedroom with a laptop that she already owns, some free software and plugins, and a microphone and monitoring headphone setup for less than $300, but what about the guitarist or drummer whose instruments and amplifiers easily cost thousands? Of course, digital emulation and sampling are becoming much more popular and affordable with companies like Positive Grid (guitar amplifier emulation) [3] and Toontrack (drum sampling plugin) [4], but any instrumentalist, audiophile, or music lover can tell you that there’s no replacement for the real deal – and decent-sounding software will still cost hundreds of dollars.
Even though the field of data analytics has generally had a positive effect on many businesses and industries, it’s worth pausing and thinking about whether or not it’s helpful to apply it to everything. As an instrumentalist and musician that grew up on more unpopular, obscure genres like metal and shoegaze, it concerns me that technology (combined with a meritocratic-capitalist society) may be contributing to the slow death of smaller, niche communities. I’m fortunate that I work in a field that pays well enough that I can still pursue my creative passions (spare time is another issue…), but many artists don’t have that privilege. So next time you walk by that record store and scoff because you have Spotify, or you wonder why anyone would ever pay for music when it can just be streamed for free, take a moment and think about the paradigm shifts that are happening in the creative community. How important is music and art to you? Do you really need that $10? Or perhaps could it make a much bigger difference to a struggling artist who might one day release your favorite album?
[1] https://medium.com/s/story/spotifys-discover-weekly-how-machine-learning-finds-your-new-music-19a41ab76efe
[2] https://www.theverge.com/2015/12/7/9861372/spotify-year-in-review-artist-payment-royalties
[3] https://www.positivegrid.com/
[4] https://www.toontrack.com/product/superior-drummer-3/
6 comments on “Should data analytics stay out of the creative arts?”
Comments are closed.
Very interesting article, thanks! However, as with most things, I suspect that the concerns you have around the industry’s application of data analytics to the arts is not actually a problem with data analytics (or AI or machine learning) but how people are using it. Data analytics is just another tool that people can leverage for multiple purposes – good or bad – and it is the laws and regulations around it’s use that should be interrogated, rather than the technology. For example, a knife is a tool, and it can be used for good (cooking) or bad (murder). Rather than criminalising the existence of a knife, the act of murder is illegal instead. I think you are absolutely correct in saying that the application of data and analytics technologies needs to be investigated, especially when it comes to a lucrative industry like entertainment, where tweaks of some algorithms could probably make certain people a lot of money. I think that since these tools and technologies are relatively new, and the people in power are often unfamiliar with how they work and the implications they could have, there needs to be a big investment in regulation of all existing technological tools, and somehow standardising this across countries as well.
Users who have LIKED this comment:
I absolutely agree. While my post is definitely meant as a critique for the use of data analytics in the music industry (and the arts at large), I’m not suggesting that we criminalize it – that would be an extreme overreaction, in my opinion. As you said, data analytics (and AI, and ML, and big data, and all these other buzzword subjects that seem to be the same thing to the uninformed) is a tool, and tools can be used for ‘good’ or ‘bad’ depending on relative morality. I also agree that stronger regulation is needed. In the specific context of my post, I think what we really need is better regulation in the music industry so that society doesn’t think of it as being low value.
A very interesting thought. How about we also look at it from this perspective – Artificial Intelligence and Analytics could help us preserve and recreate art to the finest amount of detail making it almost immortal. We could perhaps use it in conservative methods but still stay relevant to the true aspect of Art.
Users who have LIKED this comment:
That’s very strongly dependent on your definition of art. To me, art isn’t about technical perfection, it’s about the intention of artist, the story that they/she/he wanted to tell, and how it touches the lives of people – it’s human. One example – the Magenta Project creates visually stunning imagery, but to me, it’s not the same kind of art as the fine art that its algorithms are trained with. To me, the art that it creates is a testament to how far technology has progressed. Equally impressive, if not more impressive, in my opinion, but it fails to tell the story that human artwork tells.
If we want to preserve true detail, then we should be forgoing digital altogether for art and only using analog (although I fear I may be opening up a whole other can of worms with the analog vs. digital subject…). With respect to music, I think it’s more meaningful to save the original analog tapes from recording sessions so that we can always replicate vinyl presses. But to me, art is something that will always exist as long as humans exist. We may not be able to keep all of it in a “perfect” form (whatever that may mean), but for the sake of archival, I think what we’re doing now already is sufficient. Almost all art already exists in some way, shape, or form in a digital archive.
I do agree to a great extent that AI is negatively affecting music. I personally used to enjoy searching for good music and felt like my effort payed off. Whereas today, with Spotify, music is being handed to us and, as a result, so many people have the same songs and playlists. That’s most certainly a personal issue but it could resonate with other music listeners. However, there’s no doubt that after a while, Discover Weekly would feature amateur artists and that could help them more than ever before. Spotify, like any other company, wants to make profits, and profits come with consumer satisfaction, so it makes complete sense to use AI and big data to offer good music and weed out bad music. Honestly, with the digital age, music-sharing outlets like SoundCloud, and music-recommending algorithms provided by Spotify, good music will be heard and bad music will be neglected. If some genre of music or artist isn’t being recommended often, odds are, that genre or artist isn’t appealing to most people – and there’s nothing wrong with that.
Users who have LIKED this comment:
That’s a great way of thinking about it, in my opinion. My concern is that people will lose the ability to determine what is and isn’t good music, but perhaps that has more to do with my pessimism about our ability to self-curate our tastes rather than mindlessly consume the content that we’re fed. Of course, there’s no “right answer” for what’s good and bad music since it’s all subjective – so long as someone can justify why it’s good or bad music. I think people are generally more likely to spend time figuring out what music they enjoy (and why they enjoy it) if they view it as valuable. Sadly, most people seem to equate value with expensiveness, which is exactly why I have concerns about the devaluation of music and other forms of digital media.