* #BrazilWasStolen, the private audit of the official public data, was presented on Friday, November 4, from Argentina, to avoid the censorship imposed throughout Brazil. Immediately the TSE (Tribunal Superior Electoral: High Electoral Court) downloaded those public data -which had been online since the night of the elections- and modified them.

* The strictest censorship rules in Brazil: no one contesting the electoral result lasts more than a few minutes on social networks, by order of the President of the TSE, Alexandre de Moraes, in agreement with the US government – which has the effective capacity to impose censorship policies on social networks


On November 4, the Argentine consultant Fernando Cerimedo was the one who provided the information, prepared in Brazil by a private group, Brazil Was Stolen, which accessed and analyzed the public electoral data issued by the Superior Electoral Court of Brazil. 

Those original data as they emerged from the election are in this file, which is the official file that was on the TSE’s website until the time the audit was disclosed. Once the audit was disclosed through a press conference and a live broadcast on YouTube (see below), the official file was downloaded by the TSE headed by Alexandre de Moraes. The file with the official information remained offline for 8 hours, and was uploaded again two hours after the audit went viral, modified (the image shows that it was modified at 19:01 on November 4). The type and extent of the modifications is being studied right now by the Brazil Was Stolen group itself, comparing the new file with the old one.

What is the complaint about?

In a nutshell, Brazil Was Stolen’s fundamental observation of the voting process has to do with the existence of two types of electronic ballot boxes (two types of machines) in which people placed their votes, and the markedly different way in which, according to the original official data, those electronic ballot boxes behaved. 

The election was conducted with two different models of machines. Model 2020 machines, which are capable of being audited, and earlier model machines (2009, 2010, 2011, 2013 and 2015), which are not auditable. 

In any election, the type of machine used cannot be a differentiating factor. That is, a citizen arrives at the polling place, is directed to her electronic ballot box, which can be either a 2020 model or an earlier model, and casts her vote. At the end of the count, if machines of different models are distributed throughout the whole country -in this case, even machines of different models were sometimes mixed at the same polling place- there can be no significant statistical difference between the votes cast on a machine of one model and a machine of another model.

However, this is exactly what happened. While the votes cast in auditable machines (2020 model) have a normal behavior, that is, they do not present significant differences or anomalies or biases, the votes cast in non-auditable machines (pre-2020 models) have a strong bias in favor of Lula. People who cast their votes on pre-2020 machines were between 5 and 80 times more likely to vote for Lula than for Bolsonaro. This represents a grossly unjustifiable statistical difference, equivalent to a probability of one in a quintillion (1 in 1,000,000,000,000,000,000,000,000,000,000). 

When comparing data from one machine to another for relatively homogeneous populations – that is, expected to behave relatively homogeneously electorally – the result is eloquent. 

Asking the right questions

The report presented by the private auditing group gives abundant details of this anomaly. How does one determine if there was interference in the voting? By comparing, in the same election, within a population that is similar demographically and geographically. To do this, the audit proceeded according to the following method, that is designed to compare the comparable: 

Brazil → Each state → No capitals → By population range.

→ → → → Demographic and socioeconomic homogenization of the population for comparison purposes → (which serves to compare equal populations).

After proceeding with this reduction of possible “noise” in the comparison, take the ELECTRONIC BALLOT BOX MODEL (2020 or earlier) as a single target difference parameter to detect possible “extraneous interference”.

The main result is the following: 

Given that machines of different model were mixed and distributed in many different locations, this result means that the machines are an interfering factor that biased the results in favor of Lula (candidate 13 according to the codes used) and against Bolsonaro (candidate 22). 

This can only be verified with certainty, as said, if the comparison between ballot boxes is established strictly within homogeneous populations. Otherwise, the argument can be made that the differences occur because the different results between machines of one type and another can be explained by the different types of voters who voted in one or the other.

Let us look at some examples, which show the disparity within homogeneous populations.

In Ceará (Northeast), where in any case Lula won by a wide margin, the graph shows a comparison of the voting results within a homogeneous population, in cities with less than 50,000 votes (that is why the capital city is not taken into account, where heterogeneity is greater). Within that population, on the left are the results obtained in the 2020 auditable machines. On the right, the results obtained in the non-auditable machines of previous years.

In the 2020 machines Lula obtained 66.53%, Bolsonaro 26.58%. In the non-auditable machines, on the other hand, Lula obtained 75.00%, and Bolsonaro 18.01%. This difference cannot be explained by any factor other than the type of machine used to vote. 

The unaudited machine gives a 47% higher probability of casting a vote for Lula. This with machines that, as said, were often located side by side in the same polling place. 

Other examples:

The original report attached at the end shows a wealth of details from all over the country and its regions, confirming what has been explained above. 

From the summary of all the information examined, a series of conclusions emerge. 

The main one is: the electoral machines (or electronic ballot boxes) prior to 2020, not audited, biased (through the use of an algorithm introduced in them) the voting, so that never the votes for Bolsonaro could exceed a certain proportion in relation to the votes for Lula. Doing this, and comparing for the whole country the results according to type of machine, a scandalous phenomenon is obtained: 

The 2020 models show a natural behavior: there are ballot boxes with 100, 200 and 300 votes. There are ballot boxes with 50 votes for Bolsonaro and 50 votes for Lula, and ballot boxes with 50 votes for Bolsonaro and 200 votes for Lula, for example.

That is why the graph obtained is not so “geometric”.

In contrast, the non-2020 electronic ballot box models have a fixed ‘maximum’ angle, from which Bolsonaro’s (or Lula’s) votes ‘cannot pass’. In this ‘limit range’, the sum of Lula+Bolsonaro votes is fixed: 300+0, 200+100 or 100+200, for example.

This can never be statistically expected behavior. Moreover, the votes are much less dispersed, apparently an artificiality in the ‘attempted union’.

The following infographic shows how to understand the distribution charts attached to the entire report.

So the pattern of vote distribution on the 2020 (audited) machines has a normal dispersion. On the other hand, the pattern of vote distribution in the remaining machines shows a line, a regular slope (impossible to obtain except by an algorithm).

The study also reveals that there is an abnormal number of electronic ballot boxes that show 0 (zero) vote for Bolsonaro for president – even though, for example, they show votes for his deputies or governors. And these ballot boxes with 0 vote for Bolsonaro correspond, invariably, to the no-2020 machines. 

In the following interactive map prepared by another group (Brazil Freedom 2022) you can see the information about the places where Bolsonaro got zero vote.

Many memes are spreading in Brazil indicating the localities where this strange phenomenon occurred, encouraging that if anyone in those localities voted for Bolsonaro, to go out and denounce it.

TRE= Tribunal Regional Electoral

Any complaint or discussion of the results is automatically censored.

Who has the power in Brazil? 

As during the pandemic, it seems clear that it is not Jair Bolsonaro who controls what happens in Brazil, but a powerful group that operates – as everywhere else – outside the elected government, and that determines what happens in terms of communication and effective power.

Dr. Alexandre de Moraes, Presidente del Tribunal Superior Electoral (TSE)

As soon as the results of the election were known, and in the face of the rejection expressed by a huge mass of the Brazilian population that took to the streets, the official discourse was: “the official result cannot be questioned”. From then on, under the power of Alexandre de Moraes and the chorus of the local and international mainstream media, censorship began to operate. 

The election is over. Those who question the result will be treated as criminals” declared De Moraes, under the applause of the pro-government press – which is against President Bolsonaro, still in office. In the United States and the European Union, a series of tweets and congratulations made it clear that Lula is the candidate of globalism. 

The same censorship that prevented reporting the truth about Covid19, the same censorship that prevents reporting the truth about the causes and the real progress of the conflict in Ukraine, now prevents reporting the apparent fraud in the Brazilian elections.

In the streets for a week there have been millions of Brazilians insisting on demanding that De Moraes and other leaders of this apparent fraud be called to account. This information and the magnitude of these demonstrations have also disappeared from many mass media, which have folded to “not reporting anything contrary to the official truth” regarding the presidential elections in Brazil. 


This is the link to the official website of Brazil Was Stolen, from where you can download all the research, the original data files -now modified by the TSE-, and so on.

The following is the video of Fernando Cerimedo’s press conference (in Spanish and Portuguese), which we also attach since it has been downloaded from the original sites.