Digital Distribution Changed Everything Except the Part That Matters

The Problem That Got Solved

For the independent labels that built the cumbia digital canon, the pre-streaming logistics were genuinely brutal. A small Buenos Aires imprint wanting to reach a listener in Berlin had to press physical units, absorb the cost of international shipping, negotiate with a distributor who may or may not have believed in the record, and wait. The minimum order quantities at pressing plants made the math punishing for any release that wasn't already moving — and the structural reasons for that haven't disappeared so much as been routed around. Digital distribution through aggregators like DistroKid, TuneCore, or the more label-oriented Amuse and Fuga collapsed all of that into a flat annual fee and a few days' processing time. A cumbia digital release in Medellín is now available to a listener in Warsaw within the same week it drops. That part of the problem is solved, and it would be churlish not to acknowledge it.

The numbers confirm the shift. Spotify's catalog now runs into the hundreds of millions of tracks; Apple Music's figures are comparable. Latin American artists and labels can upload directly. No gatekeeping by an import buyer, no dependence on a sympathetic editor at a physical distributor who happens to know your city. The geometry of access is genuinely flat in a way it was not before 2010.

The Problem That Wasn't

What digital distribution did not solve is the discovery architecture — the set of mechanisms by which a listener who doesn't already know they're looking for something comes to find it. And this is the part that matters more.

Algorithmic recommendation systems are trained on engagement signals: saves, streams, playlist adds, skip rates. A track that arrives without an existing audience generates no initial engagement, which means the algorithm has nothing to work with, which means it recommends the track to no one, which confirms the absence of an audience. The feedback loop is not hostile to Latin American electronic music specifically; it is indifferent to it, which has much the same effect. Indifference at scale is its own kind of wall.

Editorial playlisting does slightly better, but not much. The major streaming platforms employ genre editors whose remit includes Latin music, but the dominant editorial logic still flows through mainstream reggaeton, Latin pop, and the more internationally legible end of Brazilian funk — genres with existing label infrastructure, PR budgets, and familiarity in the listener's frame of reference. A label releasing experimental cumbia, or the harder, more percussive strains of baile sound-system music, exists at several removes from that infrastructure. The editorial attention it receives reflects that distance.

Context is the underlying issue. A track from a São Paulo baile exists inside a web of meaning — the sound-system culture it came from, the neighborhood it belongs to, the DJ practice it was built for, the way its sub frequencies are designed to behave in a specific acoustic environment. Strip away that context and upload the file, and the listener abroad receives audio data and a genre tag. The genre tag is usually inadequate — "Brazilian bass" covers an enormous range of intentions and tempos, and "Latin" as a classification is essentially useless — and there is no mechanism within the distribution chain itself to supply the rest. The distributor's job ends at delivery.

This is not a new observation, but the streaming era has sharpened it. Physical distribution, whatever its costs, moved objects through networks of people who understood them: record shop staff, importers, DJs who traveled. Those people provided exactly the contextual layer that the algorithm omits. The crate at the back of a shop in Kreuzberg or Hackney, containing twelve records a buyer had sourced from Buenos Aires, was also a curatorial argument — a set of choices made by someone with knowledge and a position. That argument has largely been replaced by a playlist compiled by an editor under commercial pressure or a recommendation engine trained on play counts.

What Actually Moves the Music

What continues to work is what has always worked: the social transmission of enthusiasm among people who know what they're talking about. A DJ in London plays a record from a Santiago label; someone asks about it afterward; that person tells three others; the label's Bandcamp page gains a handful of followers. Bandcamp, notably, still functions as something closer to the physical-distribution model — a page can carry text, liner notes, context — and the way it surfaces music through the feeds of followers and label pages still rewards curation over engagement metrics, at least to a degree. The DJ sets uploaded to Mixcloud and the radio show archives on NTS or Dublab do similar contextual work.

The truth is that the geography problem and the context problem were never the same problem, even when they looked that way. Digital distribution solved one efficiently and left the other entirely intact. The music from the region is available everywhere and heard — in any meaningful sense — in relatively few places. For labels navigating this: the upload is the beginning of the work, not the end of it.