How many Red Notices, published by Interpol, are later found to be out of compliance with its rules? This is an important question for U.S. policy, and for the policies of all democratic nations. There is rising concern, and press and political interest, in the problem of Interpol abuse — that is, the use of Interpol by autocratic states (including Russia) to harass dissidents, refugees, and political opponents. But we do not have a handle on how big a problem this is.

It should be easy to find out how many Red Notices are later found non-compliant. In fact, it is anything but easy. A logical place to look for this information is in Interpol’s Annual Reports. But they provide no information relevant to this question.

Another possible source is the Commission for the Control of Interpol Files, an independent appellate body. CCF reports do provide useful information. Unfortunately, it is presented in statistics that do not line up with each other, that do not add up, and which change their basis from year to year. Moreover, these changes are not explained, making it all too easy to miss major differences in the CCF’s reporting methodology. CCF reports have no consistent format, do not reliably present the same types of information, and on occasion even contain typos that make it difficult to know what the report is saying.

The following is a good faith effort to determine how many Red Notices were found to be non-compliant in 2016, but because of the number of assumptions required to make sense of the CCF’s statistics, it is no more than an estimate. It is not — for reasons that will be explained below — a safe estimate of the number of Red Notices from any given year that are later found to be non-compliant. Nor can it be used to generate an average of the number of Red Notices from a given set of years that are later found to be non-compliant. It is only about the number of Red Notices that were deleted in 2016.

One approach to calculating the number of Red Notices later found to be non-compliant would be to start with the number of requests the CCF received and processed in 2016 and work back from this number to calculate the number of deletions. Unfortunately, this doesn’t work, because the CCF receives many requests that it cannot consider — because they do not meet its fairly simple criteria for admissibility — and even more requests from individuals who are not in Interpol’s files. It then (sometimes) runs these requests into its count of processed requests — though sometimes it excludes them — which makes it impossible to figure out what proportion of the processed requests required a substantive decision.

So instead of starting with outcomes, we must start with categories. The CCF receives (or at least classifies) four types of requests: complaints, requests for access, preventative requests, and requests for review. It does not state how many preventative requests and requests for review involve a Red Notice. But fortunately, the CCF received only 70 preventative requests and 24 requests for review, and of these only 29 (7 preventative, 22 review) were from individuals who were in Interpol’s files. Ignoring these 29 cases will bias the final figures, but not by much.

The CCF’s two main areas of activities are dealing with complaints and requests for access. In 2016, it received 357 complaints from individuals in Interpol’s files, 67% of which were from individuals named in Red Notices. It therefore received (approximately) 239 complaints from named individuals. In the same year, it received 115 requests for access from individuals in Interpol’s files, 65% of which were from individuals named in Red Notices. It therefore received (approximately) 75 such requests, for a total of 314 complaints and requests for access in 2016 from individuals named in Red Notices.

The CCF also states that, of all the requests it examined, 54 percent involved cases in which it recommended the deletion of non-compliant information. If we apply that 54 percent deletion rate to the figure of 314 complaints and requests, we get a total of 170 cases involving the deletion of non-compliant information and Red Notices in 2016.

It would be convenient if all non-compliant deletions involving Red Notices were deletions of Red Notices. Unfortunately, they probably are not. The 2016 CCF Report says nothing useful about this, but if we go back to the 2012 CCF Report, we see that the CCF did not simply delete or not delete information: it also updated information, added an addendum to a published Notice, or blocked access to a Notice. The 2016 Report says nothing about the CCF taking any of these actions, but it seems unlikely that the CCF abandoned them entirely. Thus, what the CCF reports as deletions in 2016 might well include actions that are not actually deletions. Moreover, it is possible that one individual could be the subject of more than one deletion. So the figure of 170 is unlike to be much higher, and could be a bit lower.

There are other reasons for recognizing this figure is inaccurate (and likely too high). It ignores any deletions made as a result of preventative requests and requests for review. It assumes that the 54 percent deletion rate applies evenly across all types of requests, an assumption which undoubtedly introduces a margin of error. But since the deletion rate can only apply to admissible cases in which there was information in Interpol’s files, and complaints and requests for access comprise the overwhelming share of such cases, the estimate of 170 Red Notices deleted in 2016 is unlikely to be wrong by very much.

The 2016 CCF Report gives us another reason to believe that the estimate of 170 deleted Red Notices is at least reasonable. It notes that 10 countries (led by Russia) comprise the “main” source of new applicants, and that there were 271 applicants from these 10 countries. As other countries added a smaller but presumably not negligible number of applicants, it seems reasonable to assume that the total number of such applications was certainly more than 300: if the top 10 countries yield 271 applicants, it does not seem unreasonable that the bottom 180 would yield at least 29 more.

As the CCF implies that these applicants were the subject of its decisions to delete or retain data, if we apply 54 percent deletion rate to our low-ball estimate of 300 applicants, we get a figure of 162 deletions — which is close to our previous estimate, derived in part independently, of 170 deletions. Thus, I believe that in 2016 the CCF recommended the deletion of at least 160 Red Notices, with 170 being a more likely figure.

This does not, however, account for the deletion of diffusions, which are another kind of Interpol communication that is akin to, but different from, Red Notices. The CCF Report notes that it received approximately 32 complaints from wanted individuals who were not subject to a Red Notice: these individuals were likely named in diffusions. Similarly, it received approximately 21 requests for access from such individuals, giving us a total of 53 complaints and requests. If we apply the 54 percent deletion rate to this figure, we arrive at a total of 29 deleted diffusions in 2016.

But these deleted diffusions are accounted for — somehow — in the overall figure of 271 applicants from the “main” countries. The logical conclusion is that the total number of new applicants from all countries was not 300, but about 360, of which about 50 were related to diffusions and 310 to Red Notices. Applying the 54 percent deletion rate, 360 applicants gives us a figure of about 30 deleted diffusions and about 170 deleted Red Notices. Until the CCF decides to provide the world with clearer information, these figures are as close as we are likely to get to the truth.

Even then, we must enter a further caveat. The CCF is not the only Interpol entity that can delete Red Notices and diffusions: the Interpol Secretariat can also review Interpol communications. We do not know how often this happens, but we do know it happens on occasion. It is likely, therefore, that the number of diffusions (and probably to a lesser extent, Red Notices) deleted by Interpol (both the CCF and the Secretariat) is higher than the total of 200, but we cannot say how much higher.

And there is another source of imprecision. This figure of 200 deletions does not relate to Red Notices published and diffusions transmitted in 2016: it relates to CCF decisions taken in 2016, and many of those decisions were about Red Notices and diffusions from 2015 or previous years. This is why, for example, we cannot assert that 170 Red Notices published in 2016 were later found to be non-compliant. It would be pleasant to be able to go back through previous CCF reports and — by the same process as above — work out how many deletions were made in every CCF year. In part 2 of this series, I’ll examine that question, and wrap up this series by looking at what the deletion rate means for Interpol as a whole.