Pell-Mell in Data Published by Electoral Commission Raises Serious Questions about Vote Counting

Atanas Tchobanov

The Bulgarian Central Electoral Commission published online the European Parliament election data in an open, machine-readable format in accordance with European Parliament Directive 2003/98/EC. The data were published according to CEC Decision No. 445-EP/May 29, 2019. However, their analysis shows discrepancies that would not have existed if the CEC methods and controls had been respected.

The Institute for Public Environment Development already published an analysis based on data from 12,032 voting polls, revealing that 1,139 voting polls had not taken into account the preferences from the machine vote. All parties and coalitions involved in the elections are affected.

The published open data currently includes 12,223 voting polls and 2,977 voting polls with machine vote. Their exploitation is problematic because, for example, there are 20 columns in the description file and 28 in the file. We combined this data with the online version of the protocols available on the CEC site and created a searchable database where suspicious results can be compared with the published PDF versions of the election protocols. The search engine can be found at

Has the machine vote been reported at all?

The first observation is that the information about unused preferences in the counting of the machine vote is confirmed. However, even worse discrepancies are detected, such as the presence of machine-cast votes for a political party that are more than the total number of votes counted for this same party. Here are some examples for different parties:

Voting poll 042200013 in the town of Pavlikeni: Two machine votes for Democratic Bulgaria, compared to the reported zero valid.

Voting poll 010300003 in the city of Blagoevgrad: Four machine votes for Vazrazhdane (Revival), compared to the reported zero valid.

Voting poll 030601062 in the city of Varna: Five machine votes for Coalition for Bulgaria and three valid reported.

Voting poll 192700119 in the city of Ruse: 29 machine voices for VMRO and 26 valid reported…

and so on, in total, there are 91 such cases in different voting polls and for different parties.

By the publication of the article, the CEC did not respond to the request for information to what this phenomenon with more machine votes than the valid ones can be attributed.

If the valid votes are more than the machine ones, the check is not that easy. In any case, the question is how many machine votes had not been counted as valid ones, and whether or not there are deprived and privileged parties because of these inconsistencies.

Malfunctioning oversight

The number of ballots in a voting poll must be equal to the number of ballots outside the ballot box plus the ballots found in the ballot box. However, this simple math has not worked in a large number of voting polls in different regions of the country.

For example, in voting poll 020400033 in the city of Burgas, 400 ballots had been delivered, 151 had been found in the ballot box. This means that there should be 249 ballots outside the ballot box. Yes, but there are 226 ballots outside the ballot box i.e. 23 are missing.

In voting poll 244601038 (24th electoral district in the capital Sofia), the opposite phenomenon is observed: 700 ballots had been delivered with 330 in the box 330 and 374 outside the box or four more than they should be.

In another voting poll in Sofia – 234608034 – there is a huge number of ballots for which it is not known how they had been delivered. According to the protocol, 358 had been delivered, however, the total number of ballots in and outside the box is 898, or 540 ballots more!

We will not here discuss the obvious handwritten additions to protocols, such as that of the headline image, in which few votes have been added to the votes for certain parties. The situation with the machine voting and the CEC’s non-working oversight is serious enough and the CEC owes to the public clear and meaningful responses.

P.S. The open data file is provided in a ZIP format that at the time of the publication of this article had the following control sums: MD5 8259a2bd2d2881c82d2661d0dfcaf302 and SHA1 b7dfe2a1e4c19c64a3e44b414ca4b6abf3994b2a. After unzipping the file, one finds the following files: votes.txt, votes_mv.txt, sections.txt, protocols.txt, preferences_mv.txt, preferences.txt, cik_parties.txt, cik_candidates.txt and the explanatory file readme.txt. In this case, we are interested in whether the votes of votes_mv.txt are included in votes.txt


If you find this article useful, support our work with a small donation.

Pay a Bivol Tax!

We will highly appreciate if you decide to support us with monthly donations keeping the option Make this donation monthly.

Select Payment Method
Personal Info

Credit Card Info
This is a secure SSL encrypted payment.

Donation Total: 5€

Извършвайки плащане Вие се съгласявате с Общите условия, които предварително сте прочели тук.

Биволъ не записва и не съхранява номера на Вашата банкова карта. Плащанията се обработват през системата Stripe. Даренията за Биволъ с банкови карти се управляват от френската неправителствена организация Data for Reporters Journalists and Investigations - DRJI.

This post is also available in: Bulgarian

Вижте също / Read Also