When government agencies redact material from released records, there are inevitably questions. That was certainly the case on 19 September, when Margaret Munro, a reporter for Postmedia News in Canada, revealed details of several seemingly egregious cases of misconduct that she had obtained under Access to Information laws.

The heavily censored documents, released by the Natural Sciences and Engineering Research Council of Canada (NSERC), the country's science-funding agency, described a case in which a researcher had faked data, then moved to another institution that may have known nothing of the episode. A second, even more remarkable, case involved a scientist who padded his CV with papers that did not exist anywhere in the published literature.

Taxpayers have a right to know about instances in which their money has been misused.

But the names of the wrong-doers were blacked out, as were the citations of retracted papers and the details of the research fields affected. In some cases, information on the misconduct findings and the nature of the wrong-doing was also redacted. One document retained sentences of a letter in which someone — presumably a misconduct perpetrator — offered abject apologies and argued that cutting off his research funding would harm his students, but blacked out the name and numbers of people involved. In another, a black rectangle obscured a list of six problematic journal articles, some of which are apparently fictitious.

Canada's practices take privacy concerns too far. There is an argument for confidentiality while an investigation is going on, or even after it has been resolved, if allegations turn out to be frivolous or malicious. In the United States, many agencies withhold the names of those accused of misconduct. Some officials argue that cases of honest error should also remain confidential, to protect the careers of scientists exonerated of misconduct. But these arguments cease to be relevant once misconduct has been found, as the US government's Office of Research Integrity acknowledges when it makes public the names of those found guilty of misconduct.

Taxpayers have a right to know about instances in which their money has been misused. Research misconduct can affect the reliability of the scientific literature, and other scientists and journal editors have a clear interest in knowing about it. And the government may be failing to protect its own interests if the names of parties disciplined by one branch are not available to others.

Making it clear that names will always be kept confidential — as NSERC does — also increases the chance of leaks from understandably frustrated whistleblowers, and of the publishing of names through unofficial routes. Last week, for example, the blog of Retraction Watch, a US watchdog of the scientific literature, guessed at the identity of one of the researchers referred to in the documents that Munro obtained. Such disclosures, although often better than no information at all, can leave the public and other researchers struggling to determine what is true.

NSERC has also redacted the names of institutions, including one that Munro says had to be 'reminded' twice to investigate allegations after a researcher had left. Yet there is a clear public interest in knowing whether universities — which typically employ thousands of researchers and educate tens of thousands of students — have received misconduct allegations, and how they have responded.

Redacting the conclusions of an investigation or the actions taken in response to wrong-doing is even worse, arguably serving to protect not only the institutions, but the funding agency itself. For example, one document provided to Munro says that the funding agency has decreed “that [black smudge] be declared ineligible indefinitely to hold or apply for a [black smudge]”. It seems absurd to redact information about how misconduct has been punished. NSERC says that it takes misconduct seriously, that cases in Canada are rare and that its actions send a strong message about the consequences of misconduct. But without any details, the message lacks force.