Lost Horizons
Streaming products in 2026 are boring. Product designers occasionally chip away with iterative changes, but most services have remained largely unchanged for roughly 20 years now. The financial models for these services work, so we optimize the existing paradigm, and those optimizations focus less on innovation and more on making the products increasingly lucrative for those who own them and control the rights.
I have been along for this ride for nearly the entire time, and have worked for or with almost all the major streaming services – Amazon Music, Apple Music, Spotify, SoundCloud, Rhapsody– over the last 20 years. Last month, I wrote a piece that described the early days of streaming music. It reminded me how revolutionary the idea was at that time, the promises and opportunities it offered, and how many of those were discarded.
Since writing that piece, I’ve spent a lot of time reflecting on ways to re-engage with those values. With the advent of AI, it feels that now is the time to start talking about what comes next. Most of us think about AI as a way to enhance existing products – or, to put it less charitably, increase efficiency by laying off broad swaths of the workforce – but the technology presents the opportunity to create wholly new and previously unimaginable experiences. If AI is half as significant as the hype says it is – and, despite many reservations, I think it could be – we’re due a new paradigm shift.
I’d like to explore a few ideas for how that might happen over the course of a few, upcoming essays. I’ll do this by looking at ways to rethink context, community, and discovery in digital music. The first essay below tackles context (and also includes a little table-setting for future installations).
Stay tuned, and let me know what you think.
An Arrested Development
There’s something paradoxical about how music works now. The silos our platforms place us in make the experience intensely personalized, but the more music is oriented around our preferences, the more we limit our understanding of the culture, the art form, and ourselves.
The relationship between culture and identity isn’t new – it extends back to well before the digital era – but traditionally it’s been a two-way street. We orient our tastes in culture around our own identity, but culture also helps shape that identity, and we do so collectively and individually through social negotiation. As the Welsh critic and academic Raymond Williams said, “the making of meaning is a social process, and meanings are offered, taken up, resisted, and revised.” Learning and growth emerge from this circulation, and from our relationships are formed, contested, and change over time.
Personalization silos cut off this circulation and detach us from a broader cultural dialogue. When streaming platforms engage with larger cultural trends — through playlists or radio stations — those ideas are borrowed from social media, which also contain other silos. In the end, it’s algorithms interacting with other algorithms.
These algorithms pursue very specific business goals. For streaming platforms, the belief is that the cleanest path to engagement is to connect users with something familiar as quickly as possible. And, by doing this, they obviate the need for context and the friction that creates. The recommendations they provide are generally tautological (you like this because you like this). The better these systems get at convincing us to press play, the less we understand ourselves or the culture. There’s no near-term, obvious ROI on developing a knowledgeable user base, but there’s certainly one in getting users to push the play button.
We become trapped in our own skewed likenesses, unable to see what we’re actually being shown. Culture no longer shapes us—it fills mental space on commutes and insulates us from one another. In 2026, for most people, music does more to separate us than bring us together.
Rebuilding Music’s Circulatory System
This is a well-understood and generally accepted view, and we want to change this and reorient the user’s role in these apps. I’ll get into ideas around new music discovery and community in subsequent posts, but for now, I’d like to focus on how we provide users with more context and a clearer picture of themselves, which is key to developing a curious, knowledgeable, and more deeply engaged user base.
To flip and modernize the constructivist view of learning: what we need isn’t passive consumption of content, but an active engagement with culture. Every surface (playlist, recommendation, narrative) should reveal contextual logic. This contextual level should reflect the user’s aesthetic preferences ( e.g., “You’re drawn to artists who fuse ambient textures with analog imperfections”) as well as broader socio-cultural contexts (“This was the first song by an African American to reach the top of the country charts”), and should provide them pathways to understanding the basic aesthetic paradigms, underlying histories, and connective tissue of music.
But, more importantly, the observations should tie together different threads of the users’ journey ( “This song echoes the harmonies you loved in Cameron Winter”). Under this rubric, the product (and its underlying AI technology) isn’t just a recommender, but a hermeneutic agent that guides and interprets.
And the products should contain and/or link to external cultural outposts and make that part of the journey, whether that means traditional cultural media or the generation of critical influencers who have sprouted up TikTok, IG, and Substack. There’s been a lot of focus on “superfans,” especially as DSPs puzzle through how to raise their subscription rates and create more graduated tiers, but there’s been little discussion about how to support the parts of the music ecosystem that nurture and create superfans and have struggled with their monetization strategies.
For the user, this requires a different mindset and set of intentions. It would require times when this product would be their primary, sole focus. And it certainly introduces friction, albeit meaningful friction. As the cultural theorist Bernard Stiegler argued, “attention is not something that can be consumed; it is something that must be cultivated.”
This isn’t how everyone engages with music, and it’s certainly not how anyone engages with music all the time, or probably even most of the time. As such, this would not replace the lean-back products that are on the market – those would still exist – but it would be supplementary products for a certain type of fan.
From the rights holders’ perspective, this would require a more liberal licensing strategy from the labels that not only accommodated major DSPs, but that also allowed for products focused on wellness, education, and community.
From the product side, this would require building an experience that is more dynamic and malleable, and more transparent about how it understands the user. That transparency shouldn’t live only in isolated moments, but be woven into the core experience of the product. In addition to contextual cues embedded in playlists and recommendations, users should have access to a dedicated space—something like a dashboard or library view—that reflects back how the system interprets their taste. This doesn’t need to be dry or overly analytical. It can be playful and engaging, borrowing cues from moments like Spotify Wrapped, but offering a more continuous and participatory version of that reflection.
This isn’t just a new interface, but a different way of reflecting the user back to themselves. Learning, in this context, becomes a kind of world-building exercise. As users engage with music, make choices, and encounter new contexts, they begin to map their own terrain of taste and meaning. Over time, the experience takes on a cartographical quality: patterns emerge, paths intersect, and growth becomes visible. Each interaction becomes a small creative act, contributing to a world that grows richer and more valuable as the user progresses through it.
I get that this sounds ambitious, but what was previously cost-prohibitive — a dynamic, contextual music experience at scale — will be possible. Advances in generative AI make it feasible to build more malleable products, where context adapts in medium, tone, and vocabulary to meet users on their level.
The creators of this context (formerly the writers and curators responsible for the contextual backbone of these services) wouldn’t be focused on narrow execution of a single artifact (say, making a playlist), but instead establish the structural parameters and define the modular elements and narrative compositions of the experience. Think less in terms of creating a short-form video about Baile funk, and more along the lines of designing a series of containers, narratives and vocabularies that serve as archetypes for telling many different types of stories.
Reframing
I know I’m painting in very broad strokes, and I also get that building a platform like this, even with the advances in AI, would be difficult, and the economics of it might be rough. But my core point is this: things are going to change, there will be a new realm of possibilities, and now is the time to think about the types of paradigms we want to embrace, and how we value culture and ourselves within them. We shouldn’t be thinking about how to create a better recommendation engine, but a different position for us, the listener: not as a consumer of cultural outputs, but as an interpreter and active participant in their self-formation and the shaping of culture. In the long run, this will be essential to both the cultures and the economics of music.