I'm a fan of Bill Maher and regularly listen to the podcast of his HBO show "Real Time with Bill Maher" on my Saturday morning runs. I appreciate how he brings on political rivals, discusses uncomfortable topics, and calls out over-reaction to political correctness. However, I sometimes cringe when I hear him reduce complex ideas based on his lay understanding of a topic.    

A good example of this reductionist argument happened on this week's show (Sunday, March 19, 2021) during his New Rules segment. He talked about the concept of "Equality of Outcomes" in which everyone is given equal treatment independent of merit.  He connected it to the idea that people who believe in the equality of outcomes should not watch the Grammys since it rewards some artists and not others. 

Mr. Maher laments that "Trophy Syndrome", where everyone gets a trophy no matter how good or bad they perform,  has resulted in generations of people who have an over-inflated sense of their own abilities and talents. And now when an individual is not as successful or famous as others, they think this is simply not fair.  

He may have a point but then he turns from a recent article in Rolling Stones entitled "Adding Up Music's 1 Percent: Data Shows 90 Percent of Streams Go to the Top 1 Percent of Artists" by Emily Blake from September 2020. He jokes "And yet Rolling Stone complains 'today's streaming landscape looks a lot like the music industry use to; a small class of artists see not just the majority of activity but damn near all of it.' ... These [artists] are called The Good Ones." 

The problem here is that there are many talented musicians that live and perform all around us. Yes, there is a lot of crap too but in my research on local music discovery, I have listened to amazing artists from many different cities and towns that practice their craft and put on amazing live shows. Not all of them play music that everyone likes, but they often have a dedicated group of fans. 

Mr. Maher goes on to quote the article again "streaming hasn't just upheld the gap between music's haves and have-nots; it's widened it."  He says this proves his point that since there are no more gatekeepers (i.e., record labels, radio DJs) it all comes down to talent. He says, "some musicians are have-nots because they have a voice but we [music consumers] have ears." 

The long tail of music consumption. The top 1 percent of artists receive 90% of music steams. This ratio is more equitable for physical album sales, but less equitable for radio plays.

This is a funny quip but what Mr. Maher is missing is that steaming services like Spotify, YouTube Music, and Apple Music are, in fact, still gatekeepers. Artists rise and fall based on whether or not they get placed on important curated playlists like Rap Caviar or algorithmically-generated personalized playlists. This is not unlike how businesses succeed or fail based on their page rank on the Google search engine. 

Furthermore, unlike AM/FM radio play, these streaming music services are not prevented by law from taking money (i.e, payola) from well-funded record labels or 3rd party companies (like SubmitHub) to feature specific songs by specific artists.  And what we know from the "mere exposure effect" is that people tend to prefer music that is familiar. This creates a recipe for a rich-get-richer marketplace that limits consumer awareness and prevents a larger group of artists from being able to support themselves without a day job on the side. 

But we can do better. We can develop music recommendation systems like Localify.org which have been explicitly designed to lift up local music communities. Or we can put pressure on commercial streaming services to do a better job of recommending music by female artists (see Liz Pelly's article "Discovery Weakly: Sexism on Spotify".) We can demand interactive features that give us more fine-grained control over what is recommended based on personal values and priorities. 

Mr. Maher is correct that talent is a pre-requisite for being successful as a musician. However, we can make music discovery more equitable and more humane if we better understand how coercive business models manipulate recommendation technologies to affect our (unconscious) biases and preferences as music consumers.

He says, "some musicians are have-nots because they have a voice but we [music consumers] have ears." This is a funny quip but what Mr. Maher is missing is that steaming services like Spotify, YouTube Music, and Apple Music are, in fact, still gatekeepers.

What Bill Maher got Wrong about “Equality of Outcomes” for Musicians

I just finished watching The Social Dilemma and it rocked me to my core. It is a documentary about societal problems and how they are exacerbated by social media: childhood depression, suicide, political polarization, radicalization, etc.

The film is told from the perspective of Silicon Valley tech insiders who had helped to design and develop services like Facebook, Instagram, Twitter, and YouTube. It also uses dramatic re-enactments to personify how a recommender system algorithm is designed to capture a user’s attention with sophisticated machine learning techniques not unlike the ones that I teach in some of my courses. 

These recommender systems have been trained on enormous amounts of personal data that is being collected while users engage with these addictive services. What The Social Dilemma points out is that these “free” services produce terrible externalities and face almost no regulations. They are being used both by advertisers and political actors to influence their users in often harmful ways without the users even knowing it.

When it comes to media recommendations (i.e., movies, TV shows, music), there is a less detrimental, but still significant, problem caused by commercial bias. The problem is the record labels pay streaming music services (e.g., Spotify, Apple Music) to surface the songs by their artists on high-traffic playlists and through personalized recommendations.  

As a result, users become unfairly exposed to specific songs, and due to the mere exposure effect, they will tend to prefer these more familiar “sponsored songs” over time.  This, in turn, prevents other talented artists from being discovered as they might in a more fair marketplace.

While this is not illegal, I would argue that it is unethical because

users are often unaware they are being fed “sponsored songs” instead of high-quality recommendations, and

it creates a “pay to play” barrier for many artists.

Together, these two problems combine to potentially create a homogenous music landscape in which a few wealthy record labels can dictate music tastes to large groups of users through the control of a small number of music streaming platforms.

A failed promise to “Democratize the Music Industry”

When I was starting out my academic career in music technology as an undergrad in the late 90s, there was a grand vision for a “Celestial Jukebox”. This service and/or device would give listeners instant access to any song anywhere anytime. The path started with peer-to-peer sharing on Napster then moved on to the Apple iTunes store, and finally reached the end game with music streaming apps like Spotify that are now installed on millions of smartphones.

The realization of this Celestial Jukebox along with the availability of inexpensive music recording software has been long thought to lead to the democratization of the music industry (Huffpost 2008.) However, having music be accessible, and having music be surfaced are two very different things. Music streaming services are the modern-day equivalent to radio DJs but unlike terrestrial radio, there are no laws that prevent streaming services from creating a pay-to-play marketplace.    

I have many friends, colleagues, and former students working for the big music streaming services. They are good people who love music and want to support artists. But as I watched The Social Dilemma, I was struck by the refrain from many designers and software developers: They all start out with beautiful ideas and the best of intentions but lament their creations once a greedy business model and the need to satisfy investors takes over. They feel a sense of guilt once their runaway social media companies lose their way and inflict harm on their users. Sadly, I believe that this may also be the case for many of the music streaming services as well.

Tristan Harris was one of the central voices featured in the Social Dilemma. He spent years working on various Google products before becoming disillusioned and starting the Center for Humane Technology. The center’s main goal is to educate people about the potential evils of social media. As a parent, technologist, and concerned citizen, I think this is a noble cause that we need now more than ever.

For the past few years, my students and I have been developing a locally-focused music recommendation system called Localify.org. The idea is to connect a listener’s favorite artists on Spotify with talented local artists through automatic playlist generation and live event recommendation. The idea that exposure to a local artist on Spotify will lead to interest in attending a recommended show, and ultimately supporting the local artist over a longer period of time.  

My hope is that Localify.org can serve as an example of humane technology. Our goal is to use music recommendation to lift up local music communities by surfacing music from hard-working artists. These artists might live in the community or come to town to share their musical talents through live shows.

Developing local music communities, in turn, helps to financially sustain both artists and small businesses (e.g., venues, recording studios.) It gets people out into the parks and bars and clubs to socialize and to dance. And speaking from personal experience, it motivates kids to learn to play music and to find friends to play music with.

As discussed in the documentary, we vote with our clicks so by choosing to stream local music is a way to show support for your local community.

The Music Streaming Dilemma

Ithaca has a superb local music scene. I grew up here and spent a good part of my youth going to late-night shows, outdoor music festivals, and free shows in the public parks. More often then not, the music that was spinning on my trusty Discman was a CD that I had picked up at one of these live music events.

But when I went off to college and then moved to various cities, I found it hard to get into the various local music scenes. It took much time and effort to find events, learn about local bands, and acquire a taste for the local sound.

At the same time, I was researching computational music analysis and reading a bit about music cognition as part of my Ph.D. studies. I began to envision a music recommendation system that would take a user’s mainstream music interests and use them to provide locally-focused artist recommendations.

Flash forward over a decade later through an unsuccessful start-up (Music Search, Inc.) and an 8-year academic project (MegsRadio.fm), I am excited to announce the latest attempt to get people to listen to local music: Localify.org

MegsRadio.fm used the standard personalized radio model (Pandora, Slacker) but the playlist algorithm mixed in songs by local artists with songs by mainstream artists. This might have been a good idea when we started in 2010 but it took us too long to develop the cross-platform app with a (frequently graduating) team of undergraduate students. In the meantime, on-demand celestial jukeboxes like Spotify and Apple Music became a reality and quickly have replaced the personalized radio player model as the dominant way to stream music.

Given the limitations of academic music licenses and the prohibitive costs of developing our own academic celestial jukebox, we decided to explore ways we could leverage existing commercial music streaming services to see if we could co-opt them to help promote local music. Lucky for us, Spotify provides extensive ways to integrate their platform using their expansive public API. So rather than beat them, we decide to join them.

You create a Localify account by logging in using your Spotify username and password.

We recommend upcoming music events in your city based on your “heavy rotation” artists on Spotify

We create personalized playlists that mix together your artists with similar local artists. This playlist is automatically updated every week.

We also send you a personalized email digest every week with shows that you have selected on our web app, as well as event recommendations for your city.

The idea is that there are great local artists all around us. They just need to be heard.By exposing them to you on a playlist along with your favorite artists, the local artists will become more familiar and preferred over time.Then, when you see an upcoming event featuring one of these local artists, you will be more likely to go check it out.

Supporting local farmers and small business owners is good for a number of economic, social and environmental reasons. So is supporting local artists. For example, if you go to 12 shows per year, on average you would spend $270 on local businesses and directly contribute $135 to salaries for people in your community. More importantly, supporting the local artistic community increases the quality of life for everyone! Just imagine life without music. 

Free-to-Use, Ad-Free, Non-for-profit Academic Project

Localify has been primarily designed and developed by students at Ithaca College and Cornell University. It is free-to-use and will never be supported by ads due to the generous support of the National Science Foundation.  In addition, Localify is a platform for research related to machine learning, recommender systems, software engineering, and human-computer interaction.

The idea is that there are great local artists all around us. They just need to be heard.

Introducing Localify

Back in March when COVID-19 effectively canceled all live music events for the foreseeable future, we were forced to take down Localify because much of our initial design we focused on event recommendation. 

We went back to the drawing board and decided to re-imagine Localify as a tool to help promote music scenes. Our big new improvements are:

Recommendation of artists who originate from or are based out of specific music scenes

Including colleges and universities as scenes

Detailed artist, venue, and scene pages for deeper exploration

Crowdscoring features to help us learn about which artists are part of which scenes. 

Give it a try and let us know what you think!

Localify is back with a renewed focus on support local artists and local music scenes.

Doug Turnbull