Fragmented Realities: How Algorithms Divide and Conquer Cultural Discourse

Our Inability to Understand Each Other and The Almighty Dollar

I was listening to the Ezra Klein Podcast this past week and he did an interview with a journalist Kyle Chayka to discuss his new book “Filterworld: How Algorithms Flattened Culture” which looks at how algorithms have fundamentally changed culture, our ontology, and what has happened to our ability to discern and produce ‘taste.’

A key concept Chayka discussed in his book is the idea of the homogenization of culture. Klein pushes back on this. I agree with this pushback as there is an unintuitive paradox here — a de-homogenizing logic at work, which is profoundly shaping our shared experiences.

The Internet: From Human Curation, Hybridity to Pure Algorithmic Harvesting

As Klein and Chayka explore the evolution of the Internet they rightly point out that during the early days of the Internet era, we are talking the beginning of this century, that what stood out in this era were humans who had taste profiles they expressed through their collection of other content on the Internet.

I was an active user in this time period, and the figures mentioned representing this era such as Atrios, Instapundit, Andrew Sullivan, Drudge, each of these people established themselves as curators — style engines you could say — that provided a digital space for others to come around and discuss, reflect, argue, grow.

It was all very informal, yet, incredibly rewarding.

You would leave a website of one of these bloggers and feel like you gained something of value. You were qualitatively changed in such a manner that you could feel it — it was an embodied experience.

But something happened to this version of the Internet. As Klein states:

“And that all got eaten by algorithms. And I do think there’s something there in — I miss curators. There are a couple newsletters that have a curatorial function now for me. But it’s hard to find a human being exposing you to their taste. And it’s something I really appreciate now when I find it. What I like about the internet is connecting with actual human beings. And the inability to find them now and hold to them, it feels like a real wrong turn.”

What Went Wrong

I want to bring in the idea of hybridity to encompass a point in the evolution of the Internet where human curation was paired with new emerging algorithmic machine processes. This period circa 2006–2014 is when I view these two modes of the internet coexisting and working harmoniously together.

I was part of this internet, not just as a user, but as an active agent building this new hybrid curatorial Internet. I was a Music Analyst for Pandora Radio working on the Music Genome Project.

My role was to listen to a song and identify the musicological traits of the music. There were roughly 400 attributes that can be scored for each song, identifying the key, changes of key, tempo, orchestration, even elements like the timber of a guitar i.e. is the guitarist strumming near the bridge or closer to the soundhole.

This information would then be sent off to the data scientist wizards who incorporated this human curatorial style process with the emergent new field of data science — big data and machine learning algorithms.

Now, this information was used to generated stations (playlists) when a user came to the platform and told Pandora what their favorite song, or artist is. A station would generate that was informed by the human produced Music Genome Project, along with the algorithm.

Another additional layer of human-centric engagement that Pandora incorporated was the concept of the thumbs up and thumbs-down. The user had the ability to influence their station by further refining it by providing thumbs up and thumbs down.

This model: A team of professionally trained musicians analyzing music, using that data to feed the algorithm, was further augmented with realtime user interaction that further influenced the algorithms performance.

This model represents the high point of this hybridity era of the internet in my mind.

Then Rise of Spotify

What Spotify offered that Pandora did not at the time, was the ability for a user to pick the song they wanted to listen to ‘on-demand,’ rather than the ‘lean-back’ experience Pandora provided.

Spotify filled a need. Any entrepreneur building a startup understands this fundamental — identify a need not being met, and build for that. No need to invent the wheel.

Spotify lasered in on that, and Pandora instead doubled-down on the lean-back experience, which proved a fateful business decision in retrospect.

Pandora’s blind spot in their business model, what they viewed as their ‘moat’ was this hybridity model explained earlier. This was at the core of lean-back experience and was just far enough removed from being personally curated, that the switching cost for most users was low enough to make the switch to Spotify. Thus the mass exodus from Pandora that started in 2015 and has never really abated.

But where Spotify excelled, it also lacked, and that was their algorithmic playlist generator. Everyone I spoke with at the time, my own personal experience, was a resounding rebuke of Spotify’s algorithms ability to generate personalized playlists that actually were ‘good.’

I heard countless times — I was a professional touring musician at this time and had conversations with a wide swath of people on this topic — that yes Spotify provided the ability to create your own playlist ‘on-demand’ their algorithm was just terrible at playlist generation when compared to Pandora.

But as Klein points out in the interview:

…I think that’s one reason the internet feels so crappy to a lot of us now. I mean, you’ve written a lot about this. But the “platform decay” process has simply advanced sufficiently in enough of the major platforms that when I do a Google search, it is so clear that that search is for the advertisers, right? They are giving me so much advertising, and so much of it is SEO crap, that it no longer feels like it is working for me.

I have really felt this in the past two years on Spotify, which is a complicated platform in a lot of ways because I’m such an obsessive listener and listen so much and so weirdly. It’s not that it’s perfect, but it takes me in new directions and unusual directions all the time. It has learned enough about me that it is better at recommending music to me than any friend I have really is. And so I really appreciate the Spotify algorithm. It’s done me a lot of good. I felt like — that beautiful music I mentioned earlier, “Warmth,” I got it through Spotify. Nobody handed me that on vinyl.

KYLE CHAYKA: Totally.

EZRA KLEIN: But man, the degree to which it is now endlessly serving me audiobooks that I don’t listen to there and podcasts that I don’t want, and just trying to get me to do things on the platform that I don’t want to do is really degrading my experience. And I wonder how long I’ll end up on Spotify at this point.

And I do think there is this way in which I would love the thing where I can kind of pay for the algorithm I want. And it does feel like we’ve backed into an internet now that is corroding. And I wonder how long that is actually sustainable for. It’s hard and it’s frustrating. And you’re a little bit locked in because you’ve built a million playlists and you’re used to it. But I can’t remember a time when it feels like everybody on the internet wants alternatives as much as they currently do.

What has happened here? Spotify did in fact get their algorithms more finely tuned. That was no longer the issue as it was in 2015–2018. But, now, something else has happened.

The algorithm was now pushing you towards places that no longer seemed tied to your subjectivity, but rather to what the ‘advertisers’ wanted you to move towards. As Klein points out that is what Google, and really all the current top attention mechant websites do now. This is Cory Doctorow’s platform decay theory.

There is a profound shift here in the role of the algorithm and who and what it serves.

The danger here and Ezra speaks too, is that this shift is resulting in a significant degradation of experiences. The bet that Spotify, Google, Facebook et al are making is that this degradation does not outweigh the switching cost.

Remember the arc of the Internet thus far — As Klein pointed out: “What I like about the internet is connecting with actual human beings. And the inability to find them now and hold to them, it feels like a real wrong turn.”

The general arc of the internet can be viewed as: 1) beginning as human-centric focused, 2) then a middle period of a hybridity where human and machine coexisted productively, 3) and today the Internet runs solely on the logic of the machine, whose governing logic is to extract your attention in order to sell that attention to an advertiser.

This all serves the ultimate teleology which is the increase return on capital invested to shareholders. What is truly revolutionary, and dangerous, in this current stage of the Internet is how the medium of music, which has historically served as the media through which cultural transformation manifested, is now the media through which the ends of capital are achieved.

Music is completely denatured. Removed from intimate subjectivity, to generalizable utility for ends unconnected from the self.

This is not a homogenization though. This is where I disagree with those arguments. There is a ‘samenessification’ steering the algorithms, yes, but this samness should be viewed in a narrow, targeted sense, and not in a broad all enveloping sense.

The metaphor of ‘echo chambers’ works quite well to illustrate my point.

A platform like Spotify knows you, and they serve you the music that falls within a certain ‘bandwidth’ — an upper and lower machine generated boundary, which will keep you passively engaged.

This machinic process also feedbacks into the music creation process where music is now written and produced in a manner that falls within these algorithmically machine generated boundaries. For those who may bristle at the idea that this sort of cultural exchange happens, a great book written a few decades ago traces this symbiosis across history: Susan McClary’s ‘Conventional Wisdom: The Content of Musical Form.’

Algorithms and the Erosion of Common Ground

But what is wrong with this?

Look outside.

Well, how about, metaphorically speaking. Look outside where I am in the United States. Do things seem to be confusing?

People, groups of people, seem to speaking dialects of our language that make no sense. A sense that your neighbor and yourself live in alternate realities?

Does it seem to you that there is a growing current of incoherence that you feel, deep inside you, coursing through our culture?

This is where I see the externalities of Internet echo chambers manifesting in the public sphere. The cultural and informational insularity fostered by algorithms results in groups of people developing distinctly different understandings of the world.

This fragmentation (see: de-homogenizing logic) leads to a situation where public discourse is marked by a profound sense of incoherence and misunderstanding. People, engrossed in their tailored realities, find it increasingly challenging to engage meaningfully with those outside their echo chambers.

This phenomenon is reflected in the recent surge of books centered on the ‘end of Liberalism’ (not liberals, Liberalism). Here are just a few I have read: Patrick Deneen’s ‘Why Liberalism Failed,’ and ‘Regime Change: Toward a Postliberal Future,’ Francis Fukuyama ‘Liberalism and its Discontents,’ Samuel Moyn’s ‘Liberalism Against Itself,’ and Charles Taylor, Dilip Parameshwar Gaonkar, Craig Calhoun’s ‘Degenerations of Democracy.’

Why is all of this happening now? I see the divergent realities cultivated by machine-centric-algorithmic curation as pulverizing our ability to find common ground through constructive dialogue on societal issues as we have been filtered into identifiable (see quantifiable) monetizable echo chambers. As Spotify et al stock prices rise, our ability to communicate in the public sphere becomes increasingly tenuous.

In this landscape, music and art, once mediums that bridged these gaps and fostered shared experiences, are now the tools in fortifying these ‘stock price echo chambers.’ This term encapsulates the direct correlation between the market performance of companies like Spotify and their algorithmic strategies.

Navigating Towards a Human-Centric Digital Future

Is there a way out of this?

I believe there is, but it requires a conscious and deliberate shift. First, we must recognize the role these algorithms play in shaping our cultural landscape and public discourse. Awareness is the first step towards change.

Secondly, we need to reclaim the value of diverse and independent curatorial voices in our digital spaces.

Medium, the website you are reading this on, is a great model which integrates the human curatorial in collaboration with the algorithmic recommendation system.

Pandora’s Music Genome Project is still going, and now offers on-demand along with their human- algorithmic curated stations which you can further refine and engage with through the thumbs up and thumbs down system.

Even websites like Pinterest embrace this ethos and meld the curatorial with the algorithmic.

Ultimately as a society, we need to cultivate a digital environment that prioritizes human connection and resonate relationships over profit. This isn’t just a technological issue, but a cultural and economic one. We need to be protective of the role music and and art have historically played. Music and art transcend the limits of dialogic engagements and resonate in registers that are vital to our ability to flourish.

Voicing these realities is a critical first step on the path to reclaiming our agency and undoing the proliferation of echo chambers who serve only the insatiable needs of investors over the health and prosperity of the public.