Have Streaming Platforms Finally Started Fixing Discovery?
Have the major streaming platforms finally run out of patience with their own algorithms — or is the slow pivot toward editorial curation just another round of rebranding? The past eighteen months have seen Netflix, Amazon, and a cluster of mid-tier services make conspicuous moves toward human-led programming, editorial verticals, and social-layer integrations that look, at first glance, like a genuine rethink. For anyone who has spent years watching films like Redirected disappear into catalogue obscurity despite genuine audience appeal, the question is whether any of this actually changes the discovery problem — or just rearranges the furniture.
The context matters here. Streaming's discovery problem isn't new. Video streaming services have been criticised since roughly 2015 for flattening catalogue depth in favour of surface-level trending rows. The standard argument — that recommendation engines optimise for engagement rather than quality — has been rehearsed extensively in the trade press and is by now accepted even inside the platforms themselves. What's changed in 2025 and 2026 is that the criticism has moved from academic complaint to commercial problem. Subscriber growth in mature markets has flatlined, and differentiation through sheer catalogue size is no longer a competitive advantage when every service has more content than any viewer can watch in a lifetime.
What the platforms are actually doing differently
The clearest signal of a genuine shift is Netflix's publicly announced expansion of its "Editorial" team — a dedicated curatorial unit that frames viewing recommendations through human-written context rather than purely behavioural data. The unit, which produces themed collections, retrospectives, and "hidden gem" features, has been running experimentally since 2023 and moved into full deployment across multiple territories in late 2025. It is a small operation relative to Netflix's engineering headcount, but its existence is symbolically significant: the largest streaming platform in the world has formally acknowledged that algorithmic recommendation alone is insufficient.
Amazon Prime Video has taken a parallel route via its integration with Letterboxd — the social film logging platform that has grown to over 15 million users and whose list-based discovery culture has long operated as a shadow recommendation system for the films that algorithms miss. The Amazon-Letterboxd connection, announced in early 2025, allows users to import their watchlists and receive cross-referenced availability data. More importantly from a discovery standpoint, Letterboxd's editorial team — which publishes curated lists, year-end rankings, and genre retrospectives — now surfaces on Amazon as a recommendations source alongside the platform's own engine. The practical effect is that a film sitting on a well-followed Letterboxd list can now generate algorithmic momentum on Amazon in a way that wasn't possible before.
MUBI, the curated streaming service that has always positioned itself explicitly against recommendation-engine cinema, has seen significant subscriber growth over the same period. According to reporting from The Guardian's film desk, MUBI passed four million subscribers globally in 2025, up from approximately two million in 2022. That trajectory matters because MUBI's model — a rotating programme of around thirty films at any given time, each selected by an editorial team — directly inverts the everything-everywhere-all-at-once approach of the major platforms. The growth suggests a meaningful segment of the streaming audience is prepared to pay specifically for curation rather than scale.
The Letterboxd factor and social discovery
The most underappreciated development in streaming discovery over the past two years isn't anything the platforms themselves have built — it's the consolidation of Letterboxd as the de facto social layer for serious film audiences. Research from the University of Southern California's Annenberg School for Communication and Journalism has documented the role of social film communities in driving streaming platform decisions, with findings suggesting that Letterboxd list placements now correlate measurably with streaming request rates for catalogue titles.
This matters enormously for films in the genre-hybrid, cross-national category that the streaming algorithm systematically underserves. A film like Redirected — British-Lithuanian, crime-comedy-thriller, built on an ensemble that rewards patience — is exactly the kind of title that lands well in curated Letterboxd lists around themes of "overlooked British comedies," "cult crime films," or "Eastern European setting, Western stars." Each list placement is effectively a discovery pathway that operates outside the platform's own recommendation logic. The list's author has done the curatorial work; viewers trust the human judgment over the algorithmic suggestion.
The implication for films from the early 2010s that missed their initial window is genuinely optimistic. For catalogue titles with genuine merit but limited marketing infrastructure, the combination of social discovery (Letterboxd lists, film Twitter, dedicated subreddits) and the new editorial verticals on the major platforms creates more viable pathways to audience than existed at any previous point in the streaming era.
What remains broken
Optimism about the direction of travel shouldn't obscure what's still not working. The major platforms' editorial efforts remain small relative to their catalogues — a few hundred editorially-featured titles against libraries of tens of thousands. The curation is still heavily weighted toward prestige cinema, major directors, and films with existing brand recognition. For every "hidden gem" feature that surfaces a genuinely obscure 2014 crime comedy, there are hundreds of titles that remain invisible regardless of their quality.
Metadata quality — the tagging infrastructure that determines what surfaces in genre and mood searches — remains inconsistent, particularly for co-productions that cross national and genre lines. The British Film Institute's industry data on catalogue accessibility consistently shows that films tagged as British-Lithuanian or classified under multiple genre categories have significantly lower visibility metrics than equivalent single-territory, single-genre titles. The technical infrastructure for surfacing genre hybrids hasn't kept pace with the editorial good intentions.
There's also a question about the commercial incentives behind the curation pivot. Editorial features and "hidden gems" verticals generate goodwill and press coverage, but the platforms' core algorithmic infrastructure remains optimised for the engagement metrics that drive subscriber retention. The editorial layer sits on top of the algorithm; it doesn't replace it. A film that underperforms on early engagement data still gets suppressed by the underlying system, regardless of how well it scores in an editorial feature. The curation is a second channel, not a structural fix.
The net verdict for cult cinema in 2026
The streaming landscape in mid-2026 is meaningfully better for cult and genre-hybrid cinema than it was in 2020 — but the improvement is fragile, partial, and still heavily dependent on human networks rather than platform infrastructure. If you want to discover films like Redirected, international crime comedies, or any of the British-Lithuanian productions from the early streaming era, your most reliable routes remain Letterboxd lists, dedicated film journalism, and word-of-mouth from people whose taste you trust. The platforms are starting to catch up. They haven't arrived yet.
What's worth watching over the next two years: whether the major platforms' editorial investments scale or stall, whether the Letterboxd-Amazon integration model gets replicated elsewhere, and whether the metadata quality problem gets serious engineering attention or continues to be treated as a second-order concern. The answer to those questions will determine whether the curation shift is a genuine structural change or just this cycle's version of the "discover something new" button that nobody actually clicks.