On this page

Platform audience

Mastodon

Kind
federated, INT
Readable by this system
read
Audience leaning
left-right (US): left; moderate; medium confidence
Leaning by community (2)
  • chaos.social left; moderate; medium confidence
  • mathstodon.xyz left; moderate; low confidence

How audiences are measured: perception and platform audiences.

Federated microblogging network of independently run servers on the open-source Mastodon software (ActivityPub); no central operator of the audience. Users are spread over many countries: a self-selected 2026 survey by the Mastodon project had 33% of respondents in the US, 19% in Germany and 14% in France (block 4).

Audience leaning#

Left-right (US): left; moderate; medium confidence.

Scale: no single country holds a majority of users (block 4; 85 self-selected respondents, so only indicative), hence country INT. The US is the largest single group, so the US scale is kept as the common reference; on European scales the step may differ.

External evidence:

  • Mosleh, Allen and Rand (PNAS 2025, block 1): every post with a link to a rated news domain in January 2024 on 345 public Mastodon servers with 1,000+ users, compared with six other platforms. Left-leaning news is relatively more popular on Mastodon (with Bluesky and LinkedIn) and right-leaning news on GETTR, Gab and TruthSocial. The Cornell summary of the study groups Mastodon with the platforms “considered more left-leaning or neutral” (block 2). This measures the news shared, not a survey of users; it is non-representative of users who post no news links.
  • The Guardian (2022, block 3): “most users agree that the network has a left-leaning bent”, and the project’s vetted server list includes queer-themed and climate-justice servers. Journalism, not a measurement.
  • No representative survey of Mastodon users’ politics was found (searched 2026-11-15): party identification and ideology of users not measured.

Reading: one large behavioral study and reporting agree on a left lean; the PNAS study places Mastodon with LinkedIn as well as Bluesky, so the lean is less one-sided than Bluesky’s on the evidence available. Step left, strength moderate, medium confidence (a non-representative study, no user survey).

Internal evidence: no trust-gradient fit yet (0 posts).

Communities#

chaos.social#

Leaning: left; moderate; medium confidence (US scale for comparability; the server is German).

Evidence: started for “the Chaos community around CCC” (the Chaos Computer Club; block 6), run by the association chaos.social e.V. (block 7), about 14,400 accounts and 6,200 monthly active users on 2026-11-15 (block 7). Its rules ban “right or alt-right” content, Nazi content and conspiracy narratives as well as content advocating oppression of marginalized groups (block 5). An operator rule that excludes the right is a strong documented position, which supports medium confidence for a left lean; the users’ own politics are not measured, so the strength stays moderate. Posts recorded so far: 0.

mathstodon.xyz#

Leaning: left; moderate; low confidence (platform figure).

Evidence: a server “for maths people” with LaTeX rendering, run voluntarily by @christianp and @ColinTheMathmo, about 27,500 accounts and 3,400 monthly active users (block 8). Its rules are topic-neutral (UK and EU law, inclusion, no harassment or exclusionary jokes) and take no political position (block 8). No study or survey of its users was found: its own leaning is not measured, so the platform figure is used, with confidence lowered to low because nothing was measured for the server itself. Posts recorded so far: 0.

Readability#

Public server APIs (instance, nodeinfo, extended description) and the chaos.social info site were read without login on 2026-11-15. Post-level reading was not tested in this external assessment.

Fits#

No fit yet: internal trust-gradient fits start at 30 posts.

Perception posts so far: Xinhua News Agency (2). Other platform profiles: Hacker News, Bluesky.

Sources behind this assessment#

  1. Mosleh, Allen and Rand, Divergent patterns of engagement with partisan and low-quality news across seven social media platforms (PNAS; Europe PMC full text) (PNAS 2025-10-30)
  2. Cornell Chronicle, Right or left, low-quality news links popular on social media (2025-10-31)
  3. The Guardian, Mastodon gained 70,000 users after Musk’s Twitter takeover. I joined them (2022-11-01)
  4. Mastodon, Discovery Week 2026 Report and Appendix (2026-08)
  5. chaos.social, Rules (date not shown)
  6. chaos.social info, One year chaos.social (2018-04-15)
  7. chaos.social instance API and extended description (retrieved 2026-11-15)
  8. mathstodon.xyz instance API and about page (retrieved 2026-11-15)