VMDO-commissioned report into algorithms published
Published 6th February, 2026
Today sees the publication of some of the results of Swinburne University and The University of Melbourne’s report into the effects of algorithms on Australian music.
In 2025, a team of academics from Swinburne Institute of Technology and The University of Melbourne reviewed both qualitative and quantitative data to assess the impact of streaming algorithms on Australian artists . Their research used data from Australia’s dominant streaming platform, Spotify, as the basis for quantitive analysis. You can read the Executive Summary via the link, or read some of the topline results below…
Concentration Bias favours Established Artists
The report found evidence indicating that the largest geographic or cultural group within a language group provides the most data, and therefore dominates algorithmic visibility and probability. In this report the US-audience shares 99% similarity with average global genre preferences based on recommended music tracks via Spotify playlists.
Proportional bias also means that “established” artists are favoured within the algorithm. Research showed that 77% of US tracks in the datasets were produced by “established” artists as opposed to 22% of Australian, further exacerbating the imbalance.
There is evidence to support the existence of Filter Bubbles
The report suggests Filter Bubbles exist where the listening habits of the dominant audience leads to a narrowing of selections favoured in algorithmic playlists. Analysis showed algorithmic playlists recommended only from a pool of 47,319 unique tracks (only 2.1% of the tracks in the dataset). In comparison, editorial playlists placed a significantly broader range of 190,034 tracks (8.35%). Editorial playlists in Australia featured 45% domestic content, but this fell to 25% in algorithmic playlists. Encouragingly, Australian listeners did show a preference for local music, but their small number was not enough to influence global algorithms.
Recent reports indicate a decline in Australian music visibility
The Australia Institute’s recent Reversing the Decline of Australian Music report shows that the total share of streams in Australia held by Australian artists has fallen from 12% in 2021 to a mere 8%. Given the popularity of algorithmic playlists, the minimal presence of Australian content on them is a clear factor in this decline.
This research emphasises that … It’s Complicated
There are multiple reasons and issues that influence recommendations, such as:
Promotion drivers (Discovery Mode, paid third party playlists)
Behaviour rejections (skipping tracks)
Career stage and familiarity
Concentration bias, and
Proportional bias
Recommendation is also subject to complex and interrelated structural issues impacting visibility
including :
Label global power
Access to popular editorial playlists
Collaboration between artists
VMDO CALL TO ACTION
Encourage access to education on the impact and costs of promotional options.
Continue to work with platforms to explore new approaches.
Invest in emerging artists to build fanbases locally and globally
Increase Australian’s familiarity with local music by increasing off-platform activity such as broadcast, live opportunities and other usage.