GSAC Analytics

Leaked lists?

Leaked lists?

The published study on voter lists was originally prepared for a new project that we are doing for the 2020 elections. We are creating a web portal that will put a lot of archival data and research that I did on the table in 2014-2016 and at the headquarters of the opposition in 2018. Also, there will be an analysis of the current situation. By the way, in connection with the expectation of gerrymandering, I finally got my hands on mastering GIS analysis, which I will also post on the site. Materials will be in English. We are launching closer to the summer. I would not force events and publish now the material that I have under the serial number 0, if not for yesterday’s leak of lists, about which there is quite a lot of talk now. On this score, I have my own theory, which I want to tell a narrow circle of my readers. There will be little raw graphs, for the site I prepare interactive and clickable ones. When I worked at the headquarters for the presidential elections, as befits an electoral analyst, I had access to the voter lists. I must say that all parties and some NGOs receive lists in their hands and this is spelled out in the rules of the game. Legally, these lists have a conditionally open status. That is, they cannot be published, but they can be researched. You can only publish statistical information, which I will now do. The study refers to the list drawn up for the second round of the presidential elections in 2018 (this is the most recent one distributed to the parties at the moment). The lists contain the names, addresses and dates of birth of voters. There are no phones, as in the leaked lists. First, let’s take a look at the demographic profile:

To unravel the sources of the leakage of the list, this graph does not carry information, but firstly demographic histograms are always interesting to consider, and secondly, if I have enough hands and brains, I plan to code a SEIR epidemic model for Georgia, in which age distribution plays an important role (invite to work together), but the following graph leads directly to the answer to the question of leaked sources. I’ll tell you more about it. In total, there are 3,528,658 entries in the voter list that I research. I grouped 15 of the most common surnames from these records and built a histogram. Then, I went to the Public Relation Service (PRS) website where I found a form to request the number of carriers of a particular surname (from time to time, this data is published by some news agencies). On the presented histogram, I also noted the difference in the number of people in the CESKO and PRS lists (red bars). It turned out that, on average, the PRS list contains 35% more entries than the CESKO list. This proportion is preserved even with a larger sample (in fact, I did this for the first 70 surnames (16% of the list) and for many others at random). Thus, we come to the conclusion that there are approximately 3.5 + 35% million entries in the list of the House of Justice, i.e. somewhere 4.8 million (lower limit), which surprisingly coincides with the number of records in the leaked data. The number of people with the 15 most common surnames in Georgia. For our analytical portal I prepared this information to show how many real voters live in Georgia. The counting is pretty straightforward. As we understood, there are about 4.8 million people in the register, of which 3.5 million are of voting age and 1.3 million are adolescents (this is consistent with the birth rate statistics for the last 18 years). I would like to note that voter lists are obviously formed according to the lists of the register. According to the census, which I am inclined to believe in terms of methodology and honesty, there are 3.8 million people in Georgia. Thus, about 1 million emigrated. I am inclined to believe that this million includes not only people who actually left the country, but also those who moved to a large city in Georgia and did not participate in the census (and in elections, due to the lack of practice in Georgia of detachment from the site). From my drawing (I hope), taking a rough calculation that people leave their places of registration in a uniform age proportion (which is of course rough), about 2.6 million people are participating in the elections in Georgia. More accurate allocation requires more local data and a link to urban / rural districts.

And now, finally, I’ll turn to the reason why I started posting research materials in advance. I have a hunch that the leaked lists are real and from the House of Justice registry. I will go a little further in my guesses and assume that they “leaked” from the register much earlier than yesterday. I seem to have already gone so far that I began to find answers to the questions that bothered me in 2018. For example, where did the authorities get their phones from to call citizens, then, unlike the opposition, they did not conduct street campaigning and did not recruit a database supporters? Here I, perhaps, will stop in my guesses. Apart from this story, I will say that data engineers who prepare pipelines for working with data sometimes earn extra money by selling them on the darknet. And I praised the dream back office so much … All this once again suggests that Georgian Dream is not going anywhere and is now paying maximum attention to preparing for the elections. I must say that the curfew is clearly part of the election campaign. I heard many enthusiastic responses to her.

Тамаз Хунджуа, (Tamaz Khunjua) Georgian Strategic Analysis Center, GSAC