In response to an earlier blog posting on possible fraud in the voter register, the company contracted for the issuing of registration cards has written a rebuttal. I will post some comments on this letter later. A French version of this letter can also be found here.
In response to the blog “Document may suggest fraud in voter register” by Jason Stearns We would like to address here to some of the preoccupation addressed in this blog which was subsequently mentioned in the local press in Kinshasa (mostly in “Le Potentiel” 5343 and “Le Phare” 4176 on the 30 of September), for our company’s name, Zetes, has been referred to in reference to the some of the work performed by our AFIS (Automated Fingerprint Identification System) on the electoral enrollment data. First of all, the title of the blog, even if in the conditional (“may suggest fraud”), has been transformed in the press in Kinshasa where fraud is now supposedly a fact, this without enquiry nor respect for the work being done. As a partner of the CENI, Zetes participates in a joint effort with the objective of providing the best information possible to the CENI so that all in DRC can participate in the coming election. In this perspective, we, as well as other partners, provide information so that the CENI can analyze, compile, compare and clean data gathered on the field as to obtain the most reliable collection of records with a minimized margin of error, errors that are inevitably encountered in such complex operations. There is no contradictory report to the official report of the CENI, sole depositary of the official and final information. We take this opportunity to remind everyone that this kind of large-scale and complex operation always brings its share of problems we are here to help solving. We put our reputation at stake as our credibility and transparency in such operations we undertake and do not wish to see this work being used for political purposes by feeding rumors that do not have reason to exist at this stage. Any given number can be made to speak different kinds of truth. We strongly regret having been exploited and cited in the local press without having had the opportunity to comment but we hope this message can bring some reassurance to readers and will prevent the incorrect use of preliminary and outdated information in the future. Having a long experience on the subject in hand, here is a statement, as bold as it may seem, that all involved in electoral processes should very well be aware of: fraud is impossible to organize at the level of centralization systems / databases. Here are different reasons why:
- It is impossible the create/delete persons in a database without anyone being aware of it, and there are many actors of different origin involved, when following centralization processes that lasts months.
- All transactions in the databases from the field and in the central system are logged and linked to the operator who performed the actions so we are capable of tracing everything that has been done from beginning to end.
- In terms of logistics, it is impossible to organize a fraud from an electronic file in proportion that cannot be detected or are of nature to have a significant impact. Try creating 100,000 fake voter within a database of more than 30 million by duplication. You would then need more than 2,000 buses (45 seater) or more than 400 planes (230-seater) to move them to other polling stations or manage to instruct 100,000 people without anybody else noticing. Having done all that, you would not even impact 0.3% of the electoral population.
- We always find in the end the same amount of “real doubles or duplicates” here and in many different countries; call it law of numbers. A large number of real duplicates for example is 0.5% based on the total electoral population of which in 90% of the cases are people getting another card because their name was misspelled or didn’t like their picture or just wanted to have two card in case they loose one.
- The nature of a duplicate is complex. The proven frauds (that can only be determined after human verification) are a small subset of the real duplicates, themselves a subset of the raw duplicates. Even in the eventuality that all of these where fraud attempts we are talking about way less than half a percent of the number of persons concerned (a double being a pair of at least 2 persons).
Computerized solution have the advantage of bringing credibility at this stage of the process by being able to cut down the error margin to bellow the percent which is seldom the case even in “mature” democracies but remains necessary in the absence of a computerized national registry to prevent fraud. Errors on electoral lists in G20 countries are often greater than that even with electronic voting. With the experience of many projects, people should now be better aware, where fraud is possible or not, and it is not at the level of centralization systems that they can occur simply because it doesn’t make any sense. As a result, we have transmitted information several time in the course of the process to the CENI and have continued working in order for the CENI to be able corroborate this with other information available from other sources and partners. The very presence of many players guarantees a degree of transparency through the exchange and ongoing consultation, which reveals that no alteration is possible. These information help improve the overall viability and confirmed that even though there were technical glitches, as you mention in you blog, we have early on stated that there where no major problems to foresee as we have sufficient means to address them. The numbers indicated in the blog are of different nature. Raw data is compared here with cleaned data, already deprived of technical errors. The comparison is therefore not possible. In the hope that this, if not addressing the use of incomplete statistics will be reassurance enough to not concede to panic and invite those preoccupied to ask questions before relaying concerns expressed in this blog as facts. We invite all who have used the information in this blog to read it again carefully, it expresses concerns that are not facts and it is irresponsible to use sparse elements in the press without further inquiry to obtain tangible information that would have inevitably lifted any doubts. Kinshasa, October 3, 2011 Fabien MARIE Program Manager Zetes PASS www.zetes.com