Business News Digital Digital Royalties Timeline Labels & Publishers Legal

Dissecting The Streaming Inquiry #10: Curation and algorithms

By | Published on Friday 5 February 2021

Houses Of Parliament

We are currently reviewing and dissecting submissions made to the UK Parliament’s ongoing inquiry into the economics of streaming.

Based on the five years of research CMU Insights has undertaken with the Music Managers Forum as part of the ‘Dissecting The Digital Dollar’ project, we explain the background to the key debates, helping you navigate and understand each issue and the proposed solutions.

The streaming services in general – and Spotify in particular – get a great deal of critism online, especially from music-makers, over the streaming music business model and the low per-play royalties that are paid. So, it’s interesting that the services themselves have so far avoided too much harsh criticism during this Parliamentary inquiry.

Of course, the services haven’t yet appeared before MPs – that starts next week – and the criticism of the digital companies might increase at those sessions. But it’s pretty common that, while concerns about the streaming music business often start with Spotify dissing, once people start digging attention shifts more to how the monies paid over by the streaming services are shared out around the music community.

However, there is one element to this debate that is getting increased attention which is entirely based on the decisions and practices of the streaming services themselves. And that’s curation and algorithms.

With so much music available on the streaming services, and so much new music being added every single week, most users rely heavily on the streaming platforms to help them navigate the tens of million of tracks they can choose from.

That begins with simple user-experience factors like how music is presented within the platforms and how search tools work. But it then quickly gets into more sophisticated curation and discovery functionality.

Quite how this works varies from platform to platform. Indeed, with pretty much every service having the exact same catalogue of music, the platforms have often used their navigation and discovery tools to try to gain competitive advantage, so “our discovery tools are better than their discovery tools”.

The approach each platform takes has also changed over time. The services often respond to what the data tells them about consumer behaviour, or just change their minds over how they want curation and discovery to work.

In the early days Spotify encouraged users and third-party influencers – including artists and labels – to be the curators. Then it decided that it wanted to lead the curation process, and now very much pushes its own playlists and curation tools to the fore. In more recent years, Spotify has also put ever more effort into and focus on its algorithm, both to inform and refine its human curated playlists, and with its entirely data-driven recommendation tools.

The streaming services are always very focused on consumer experience and understand how people interact with their platforms and catalogues better than anyone. Therefore many of the developments in this space are good developments.

However, there are increasing concerns about the decisions the services make in this regard and the impact that might have on what kinds of artists, tracks and genres get pushed – whether by human curators or the algorithm – and whether that, deliberately or inadvertently, puts certain artists, tracks or genres at a disadvantage.

Others question what impact those decisions have of the economics of the music industry. Through their playlists, the streaming services are good at delivering sudden spikes in listening for tracks, but do they encourage the kind of sustained repeat listening labels need to make a return on their investment?

And do they encourage users to engage with an artist and their wider output, to help grow the fanbase an artist needs to build a sustainable business around their music?

There are also transparency concerns. Although Spotify in particular has gone out of its way to open up its playlist pitching processes to the wider music community – and other services have followed that lead to one extent or another – there isn’t always total transparency about how things work, and all the more so once it’s an algorithm doing the curation.

As with social media algorithms, this increases the amount of trial and error artist and labels must undertake to figure out what kinds of marketing strategies work – and once you’ve done that, the algorithm usually changes.

And on top of all that, some also fear that – if commercial considerations don’t already currently impact on curation decisions – they might do down the line.

A number of the submissions made to the select committee discuss these issues.

In its submission, Beggars writes: “The algorithm is now in charge, it has largely taken the place of charts, chart shows and even reviews. Spotify in particular is very focused on utilising algorithms to deliver what they think the user will listen to”.

“Spotify resists attempts for rights owners to promote their own recordings via third-party owned playlists on the platform”, it goes on. “The whole ecosystem is very much Spotify’s USP and they resist any non-Spotify offering”.

Not only that, but “there is a clear policy to overlook albums and concentrate on individual tracks. For example, on the new release page on Spotify there is no distinction between EPs, singles and albums”.

Noting that an increasingly significant percentage of listening on Spotify is algorithm driven, Beggars concedes that, on one level, “this can be good for music discovery”, but, it adds, “it can also lean towards the homogenous – as with all algorithms”.

In its submission, music distribution firm state51 also makes some interesting observations about the impact of the streaming services’ curation systems on the kinds of music people consume, and even the kinds of music artists make.

“Platforms more recently have moved away from the notion of ‘consumers as curators’, preferring to regain effective control of the music discovery process through their own algorithmic and curated playlists”, it writes.

“Through their artist tools, platforms are also now intervening to influence the style of music being made. It’s not clear where this might take the industry; one model would suggest that music could become much more ‘average’, with less room in the market for radical innovation or subcultures”.

In its submission, contemporary classical label NMC expresses concern that curation decisions made by the services might disadvantage more niche genres, and the artists and labels that occupy those scenes.

“While algorithmic interventions can help users to easily find tracks in the same style as the ones they’ve been listening to”, it writes, “algorithms mitigate against promoting the music of ‘outliers’, and growing audiences for more experimental or different music”.

“In other words”, it goes on, “they do not promote diversity of expression or creativity. Ultimately, this must have a deadening effect on the industry. Critics have written about the tendency for ‘personalisation to the point of banality’ and the ‘one-size fits all approach'”.

Some of the submissions also make reference the potential of commercial considerations becoming part of the curation process. The major labels are adamant they do not secure preferential treatment in this domain in their licensing deals, insisting that they are pitching tracks to curators and learning how to play the algorithm just like everyone else.

However, under the current model, some tracks are cheaper for a streaming service to deliver than others. And while that may not be influencing curation choices today, it could in the future. And, of course, Spotify recently began piloting a new service whereby artists and labels can inform its algorithm about priority or newsworthy tracks in return for agreeing to a discount when those tracks are then streamed.

Opinion is divided on that proposal. Some see it as a modern version of payola – ie paying radio stations to play your music – and a sign that all the concerns were justified over commercial considerations becoming part of the curation process.

Others see it as a desirable development, allowing the industry to ensure that the Spotify algorithm is considering the priorities of artists and labels, and without any upfront costs that would price out independents. Several submissions make reference to this development.

Beggars writes: “There is a risk that certain cheaper content is prioritised over more expensive content – and, in fact, [recently] Spotify announced it is going to offer rights owners the chance to get additional plays in return for accepting a discount to the amounts payable. The result of this will, we fear, be that the service increasingly chooses to push music according to how much it costs them”.

In its submission BMG notes: “Spotify’s announcement that it is to offer labels paid-for personalised recommendations which influence algorithmic playlists has been widely criticised by artists as a form of digital age ‘payola’. While it is too early to say whether such language is justified, any mechanism which is seen to rig the market in favour of the biggest and best-funded players will inevitably raise concerns about market manipulation”.

Meanwhile, in its submission, the Ivors Academy writes of Spotify’s pilot: “This development is a trial and many details are unknown, such as the level of reduction, duration of the reduction and whether it is intended to affect the writers’ royalties. But on the face of it this is a concerning development, as it reduces the royalty rate for creators when they thought the royalty rates could not get any lower”.

This aspect of the economics of streaming debate hasn’t been as widely discussed over the years – although it was the focus of the first One Step Ahead report from IMPALA and CMU Insights last year, which all IMPALA members can download.

Most in the music community would acknowledge that the streaming services are the experts when it comes to user experience. However, there is a case for more transparency being needed as services evolve their curation systems and policies. And possibly a more wide-ranging debate involving the platforms and the music industry to discuss the various concerns being raised. We will see if any of that happens.

You can follow all our coverage of the Parliamentary inquiry into the economics of streaming via this CMU timeline here.



READ MORE ABOUT: | | | | | | |