The small rooms: how DOD's podcast archive reads from 2026
The DOD podcast archive runs from 2017 to 2023. DOD is a discussion forum — these were working conversations among practitioners thinking through how democracy functions and fails, not policy proposals or predictions. Revisiting them from 2026 is an exercise in calibration: did the observations hold up? Where did the analysis prove accurate, where is it still unresolved, and what did it miss entirely?
The summary and analysis in this post were drafted by Claude Code from the episode transcripts and DOD's published blog posts. Quotes are reconstructed from auto-generated transcripts — verify against the recordings before citing. A human editor reviewed it for general accuracy.
The question isn't whether DOD influenced anything — it's whether the diagnosis was right. That's worth asking honestly, including about the parts where the picture is more complicated than the analysis suggested.
Where the observations held up
Citizens' assemblies as a serious policy option. When Nicholas Gruen argued in 2017 that randomly selected citizen juries were a structural corrective to elections — that elections select for charisma and emotional mobilisation rather than considered judgement — it was a reformist position with limited institutional uptake. By December 2025, the Victorian parliament had completed an inquiry into its Upper House electoral system and recommended a citizens' assembly as the appropriate mechanism for deciding any reform. The 2025 Australian Election Study found that 48% of Australians support the idea, against only 20% opposed — and that only 32% trust government at all. The Victorian inquiry is documented on this site. The 2017 Gruen argument is in Citizens' Democracy; the 2020 version in Isegoria.
Gruen's framing was precise: it's not that politicians are bad people, it's that the system rewards the wrong qualities. "We are getting exactly what we're asking for." The trust figures suggest that the public has arrived, intuitively, at a similar diagnosis.
Trust as a measurable substrate. The 2019 DOD Trust discussion treated trust not as a warm feeling but as the structural substrate without which collective intelligence can't emerge at scale. The 32% government trust figure from the 2025 AES is a concrete measurement of that substrate deteriorating. The podcast was examining the mechanism; the survey is measuring the outcome.
Wealth concentration as a governance problem. At the December 2022 Basil's Table panel, Jose Ramos framed the central contemporary governance problem as unprecedented private wealth concentration distorting political input. Within three years, Australia passed the Electoral Legislation Amendment (Electoral Reform) Act 2025, introducing donation caps and tighter disclosure requirements as a direct policy response to exactly that concern. Whether the legislation is sufficient is a separate question — Gruen's structural argument would suggest it addresses the symptom (who funds the game) without changing the game itself.
Where the picture is more complicated
The Flux Party's issue-based direct democracy model and MiVote's consensus platform were both presented at DOD with genuine conviction. Both have the same core insight: that with appropriate mechanism design, you can get considered, cross-partisan agreement on specific policy questions at scale. The Flux discussion from 2020 and the MiVote catch-up from 2021 are detailed and technically serious.
Whether those specific mechanisms are the right vehicle remains an open question — neither has achieved significant institutional adoption. But the underlying observation (that people can reach considered cross-partisan agreement when given the right conditions) has, if anything, been strengthened by the citizens' assembly evidence. The harder question the 2019 Trust discussion named is still live: why can't organisations working on democratic reform coordinate around a shared structural agenda? The reform movement coordination problem — groups with compatible diagnoses but different mechanisms, unable to unite — is itself a collective intelligence failure. It's the accountability sink applied inward.
Write In Stone's research-trail transparency model — showing your sources as a form of credibility rather than vulnerability — has found more uptake in adjacent domains (fact-checking, AI-assisted journalism) than in the political mainstream it was aimed at. The principle is now embedded in how DOD documents its own philosophy development. But as a broad reform of political discourse, it remains ahead of its moment.
The structural problem none of them resolved
Every discussion in the archive, across eight years, circles the same underlying problem from different angles.
The 2019 Trust discussion named it most precisely: "the people who are influenced are not the people who are influencing." The feedback loop between political outcomes and political decision-making is broken. Dan Davies' accountability sink concept — developed in The Unaccountability Machine (2024) — gives it a more recent theoretical frame: a structural feature of large organisations in which the chain between decision and consequence is broken so that no individual can be held responsible.
The citizens' assembly proposals, the donation caps, the deliberative democracy mechanisms — all of these are input reforms. They address who decides and how they're selected and whose money flows into the process. None of them directly addresses the output problem: how do the consequences of decisions reach back to the people who made them? That's the feedback loop Gruen and Hofkirchner were both circling in 2017, from opposite directions (deliberation vs. prediction markets), and it remains the least-addressed item in democratic reform discussions.
Hubertus Hofkirchner's argument for prediction markets as collective intelligence infrastructure was, at its core, a feedback mechanism: prices aggregate dispersed knowledge about consequences, making the link between decision and outcome visible in real time. Gruen's counter — that this only works for factual, resolvable questions and breaks down on value questions — identifies exactly why no single mechanism can close the loop alone.
The CI-vs-AI thread
Alexar Pendashteh's framing from Basil's Table 2022 — that the goal is collective intelligence, not artificial intelligence, and that the criterion for any tool is whether it raises or lowers the group's CI — was ahead of its moment when she said it. In 2026, it's one of the central questions in technology governance across every sector. The DOD philosophy review using AI systems from different national contexts to stress-test the framework's own claims is a direct application of that principle: using AI as a tool to raise collective institutional intelligence, not to replace the human judgement doing the work.
What the archive is, and isn't
These discussions weren't predictions. They were working sessions, attended by practitioners trying to solve specific problems, often disagreeing with each other in productive ways. The Gruen/Hofkirchner debate about what deliberation can and can't do — what shall be versus what will be — is more useful as a standing tension than as a settled conclusion. Keeping it in the archive as a tension, rather than resolving it into a DOD position, is probably the right call.
The gap the archive doesn't fill: almost all of it is Australian, and almost all of it is in the liberal-democratic reform tradition. The accountability and feedback-loop problems it identifies are not unique to Australia or to liberal democracy. But the mechanisms it proposes are largely designed for and within those contexts. That parochialism is worth naming as the archive's main limitation — and a direction for the next eight years.
DOD podcast posts referenced in this article: Citizens' Democracy (2017) · Trust: A Concept Analysis (2019) · Isegoria (2020) · Flux/IBDD (2020) · MiVote (2021) · Write In Stone (2021) · Basil's Table (2022)