Civic
A new analysis from Harvard's Ash Center challenges the field's foundational assumption that making government data public is enough to hold power accountable.
NewsOnScale Staff
November 1, 2026
There is a version of the civic technology story that goes like this: governments hide information, reformers build platforms to expose it, citizens get angry, and accountability follows. It is a clean narrative, and it powered a global wave of open-data initiatives, budget transparency portals, and procurement databases throughout the 2010s. The problem, according to a rigorous new examination from Harvard Kennedy School's Ash Center for Democratic Governance and Innovation, is that the evidence for this chain of causation is far weaker than the field ever acknowledged.
The analysis — drawing on lessons accumulated across civic-tech deployments in multiple countries and contexts — arrives at an uncomfortable conclusion: transparency is necessary but structurally insufficient as an anticorruption mechanism. Publishing the data is not the intervention. What happens after publication is.
## The Assumption That Went Unexamined
The open-government movement inherited a philosophical premise from information economics and investigative journalism: asymmetric information protects bad actors, so equalizing access to information removes their cover. This logic is not wrong, exactly. It simply omits most of the system.
Corruption persists not only because information is hidden, but because the actors positioned to act on disclosed information — prosecutors, auditors, journalists, opposition politicians, civil society organizations — often lack the resources, independence, or legal authority to convert data into consequences. A procurement database that reveals inflated contracts is valuable. It is not, by itself, an enforcement mechanism.
The Ash Center's assessment identifies several structural gaps that transparency tools alone cannot bridge: weak investigative capacity in state audit institutions, legal frameworks that criminalize exposure rather than the underlying misconduct, political economies in which the actors most motivated to use transparency data are also the actors least likely to have access to power, and — critically — the attention problem. Citizens, even outraged ones, face collective action barriers that data portals do not dissolve.
## What the Evidence Actually Shows
The report's most useful contribution may be its granular reading of cases where transparency initiatives produced measurable anticorruption outcomes versus cases where they did not. The pattern that emerges is consistent: impact correlates not with the quality of the data platform, but with the presence of an organized user — a journalist with editorial backing, a civil society group with legal standing, an auditing body with genuine independence.
In other words, the tool functions when someone picks it up. And who gets to pick it up, and what they can legally do with it once they have, is a political question that no database schema resolves.
This is not a counsel of despair. Several programs reviewed in the analysis show real results precisely because they paired data disclosure with user capacity — training investigative reporters to interrogate procurement records, embedding data analysts inside civil society organizations, designing platforms explicitly around the workflows of auditors rather than the aspirations of technologists.
## Implications for the AI-Augmented Accountability Stack
The timing of this reassessment matters. Civic technology is on the verge of a significant capability expansion as AI tools lower the barrier to pattern detection across large government datasets. Anomaly detection in public contracts, automated red-flagging of budget line items, natural language querying of regulatory filings — these are real and arriving fast.
If the transparency-only thesis failed in the era of static spreadsheets, there is no reason to assume AI-powered transparency will fare better by default. The Ash Center's framework suggests the productive question for this next generation of tools is not what they can surface, but what institutional infrastructure exists to receive what they surface and act on it.
Building smarter detection systems into civic platforms without simultaneously investing in the legal, organizational, and political capacity to use what is detected risks replicating the same dynamic at higher resolution: more data, better visualizations, equivalent impunity.
The field built remarkable infrastructure over the past fifteen years. The evidence is now asking it to take seriously what that infrastructure was always for.