Independent audit of iReV portal uploads casts doubt on INEC results for Rivers state, others
An independent picture of the March 25th presidential results is slowly beginning to emerge, as thousands around the world engage in a crowd-sourced audit of the results uploaded to the iReV portal since election day.
The iRev portal is publicly available and anybody can log in to see the results of any polling unit in the country.
Before the elections, the Independent National Electoral Commission (INEC) promised the results sheet (Form EC 8A) from each polling unit would be uploaded to the internet and be publicly accessible in real time on election day, using the Bimodal Voter Accreditation System (BVAS). As at publication, nearly two weeks after the presidential election over 93% of the results sheet had been uploaded.
Rivers of inaccuracy
Initial findings based on an audit being carried out on a website created by tech entrepreneur, Mark Essien, and other tech professionals cast doubt on the Rivers state results announced by INEC, which had declared Bola Tinubu’s APC as the winner of the state.
The audit suggests Peter Obi’s Labour party won the state. With more than 80% of the votes for Rivers tallied so far, it would appear INEC greatly under-counted the votes won by Labour, while more than doubling the votes garnered by the APC.
Based on the audit, Labour had 215,560 votes, APC had 115,698 and PDP with 72,419. The results INEC announced were APC: 231,591; LP: 175,071; and PDP: 88,468.
After collating all of Rivers State Polling Unit Results together (even accepting all modifications), LP had 215,560 votes, APC had 115,698 vote and PDP: 72,419. The results INEC announced: APC: 231,591, LP : 175, 071 and PDP : 88,468.
This is based on https://t.co/DombV8DUyE
- me (@markessien) March 7, 2023
Results sheets which haven’t been uploaded or which have but are illegible are not included in the count.
With 523,651 votes cast in the state according to INEC, around 100,000 or so votes are yet to be added to the tally. Whether the APC is able to catch up the Labour as those votes are added is up for debate, according to Stanley Ikechukwu, Head of Operations SBM Intelligence, at Lagos based security and research consultancy.
“It depends on where the remaining polling units left to be uploaded are. Certain LGAs [local government areas] have more population compared with others. The population centres have always been the battleground in any election. The Rivers case will not be any much different.”
Suspicions and gut feelings
Questions about the accuracy of INEC’s figures began to emerge shortly after the election, with many people from Rivers state expressing shock that Labour had lost the state, where Peter Obi has long been seen as a popular candidate.
Having a feeling about the results though, isn’t enough to indicate that INEC was wrong, which is where independent analysis came in.
Civil society organisation, Yiaga Africa, did a random sample of results across the country, and found that their estimates for Rivers and Imo states were at odds with INEC’s official figures.
In a statement on the 1st of March, Yiaga Africa said, “in sharp contrast to the Yiaga Africa WTV [Watching The Vote] estimates for Rivers which are: APC 21.7 percent ±5.0 percent; for LP 50.8 percent ±10.6 percent; and PDP 22.2 percent ±6.5 percent.”
Yiaga Africa further added, “INEC should clarify the inconsistencies in some of the results, especially presidential election results from Rivers and Imo states.”
Why audit the results?
While different results from Rivers and Imo states may not affect the results materially in that the APC would still have the most votes and the 25% threshold spread, it raises questions about INEC’s overall credibility.
The Electoral Act 2022 allows INEC to investigate concerns before announcing a winner, however, it appeared to only allow recourse (going to court) to aggrieved parties after APC was declared winner.
Mark Essien, a self declared Peter Obi supporter, and founder of the hotel booking portal, hotels.ng, felt something was amiss.
“It was very clear in places, like in Rivers state, that the numbers were not matching,” said Mark Essien.
“But how do you prove it, who’s going to take out 6,000 sheets from Rivers state and tally it all up together somehow with a pen?”
Essien decided to leverage his knowledge of technology to look at the results.
Working with other developers, the website took them just two days to build.
How it works
The website uses crowdsourcing to audit the results.
When a user lands on the webpage of the Polling Unit Validator App, it displays the Form EC 8A of a specific polling unit.
The user enters the number of votes received by the four main parties, APC, LP, NNPP, and PDP, while also adding information such as whether any details on the page appear to have been altered, whether the polling agents signed off on behalf of their party, and if it was stamped by the presiding officer of the polling unit.
Other key identifiers are also included — the state, local government area, and specific polling unit, all of which can be found at the top of the Form EC 8A.
Some of the uploaded images are illegible or are the wrong results sheet, for example, for a race in the House of Assembly. These can be marked as unclear or invalid.
Once submitted, the numbers entered are added to a counter for the parties. These can be viewed on a results page.
Validating each results sheet takes less than two minutes, before another one is displayed for the user to repeat the process.
Maintaining accuracy
A quick look at those sharing the website on twitter indicates many are Obidients, supporters of the Labour Party and Peter Obi. However they can’t artificially inflate his numbers, nor can others who want to derail the process.
The accuracy of the audit is ensured by more than one user checking the same sheet — just like when in school you would pass your test paper to another pupil to mark.
Even if someone puts in the wrong figures, deliberately or otherwise, it won’t matter as there are other users also independently checking the same results sheets. The level of accuracy needed is programmed into the system.
In theory, bad faith actors could try to derail the process by trying to trick the system into accepting the wrong results.
But in order to do so, a coordinated group of people would need to target specific polling units and have multiple people separately entering the same wrong figures, thereby fooling the system into thinking the wrong results are the correct ones, since several users have separately agreed on the results.
However, this wouldn’t work in practice as the Form EC 8A that a user gets is assigned randomly.
The probability of several people sitting in a room getting the same sheet from a pool of tens of thousands at the same time is in the order of one in billions — you have a better chance of winning the lottery.
There’s also an added level of verification not open to the public.
Simply put, there’s no point in trying to game the system.
What impact will this have?
As at the time of publication, around 64,400 polling units had been successfully transcribed, with around 19,600 more to go, and more being constantly added.
At any one time, the website has had up to 850 people around the world working on the sheets, and getting through 20,000 a day.
Mark Essien and the other developers intend to upload all 176,000 EC 8As or whatever is available, for verification. He hopes the information can be used as evidence in any legal challenge to the results.
Essien says he just wants the numbers given by INEC to reflect what really happened.
“I’m not part of the Labour party, I’m not part of Peter Obi’s team in any way… If he turns out that he didn’t win, then I’m fine with that. That’s the truth, and we know that that’s the truth.”
Originally published at https://greycontinent.com on March 8, 2023.
