Explanation of difference of processed and biometric data about 137,000 votes

KABUL (Pajhwok): Pajhwok Afghan News has obtained information from the Independent Election Commission (IEC) and other sources about a difference of approximately 137,000 voter records between the number of biometric datasets registered on the IEC computer server and the number of voters processed by the voter verification devices used at the polling stations on Election Day.

BIOMETRIC DATASET DISCREPANCY:

The IEC has confirmed and clarified a discrepancy between two separate sources of the same information, the total number of biometric datasets and the number of processed voters.

Biometric datasets are voter data that contain each voter’s biometric specifications, such as the voter registration ID number, tezkira ID paper, photograph of the face, and fingerprints. This data was recorded by the verification device on an SD memory card as unique files, which were then transmitted to the IEC by the device on or after Election Day.

The number of processed voters is an internal counter of how many voters were verified as eligible for casting a ballot with a quick response (QR) code on the backside. This processed voters number was recorded in the internal memory of the device and also transmitted to the IEC’s server along with the manually keyed in results data and an image of the results form on Election Day or afterwards.

According to the IEC and a series of letters in November from the Dermalog Company, which provided the biometric devices and analyzed both sets of transmitted data, the number of registered and uploaded biometric datasets is 1,929,333, while the number of processed voters counted by the individual devices is 1,791,703. The difference between these two numbers counting the same information is 137,630 voter records, or in other words, the processed voters number is already incorporated into the biometric datasets number and thus has no affect on the total number of voters recorded with biometric datasets.

In addition, the Dermalog company also identified 86,226 duplicate voters who were matched against voter registration certificate numbers, photographs of faces, and fingerprints to identify multiple voting. When these 86,226 invalid votes are removed from the total number of transmitted biometric datasets, the actual number of verified voters is 1,843,107. This final number, however, does not include any polling stations where no results or biometric data was transmitted due to the device or its SD card having gone missing for some reason.

Over the past three weeks, this 137,000 discrepancy between voter’s biometric records created confusion among the presidential campaigns, election observers, and other electoral stakeholders. On November 14, the IEC issued Decision No. 108-2019 on this matter, where the IEC decided that “the 137,630 biometric data where there was a discrepancy with the “processed voters” and which was caused by a technical problem by Dermalog, these cases do not have any specific problems”. The IEC decision was signed by all seven commissioners.

According to the IEC, there is no longer any technical issue with the 137,000 discrepancy; however, the presidential campaigns of Abdullah Abdullah, Gulbudin Hekmatyar, and Rahmatullah Nabil have not yet accepted this decision despite repeated explanations and presentations by the IEC to representatives of these campaigns and others. The result is the current boycott of the audit/recount process and the prevention of that process in at least seven provinces in the northern regions. The rest of the provinces have either completed their audit/recount work or are continuing to conduct their audit/recount operations.

IEC DATA ANALYSIS OF BIOMETRIC DATASET DISCREPANCY:

The IEC documents indicate 10 provinces with the largest difference between processed information and biometric datasets of voters. Kabul has the largest difference of 23,921 voter records.

The difference in Nangarhar province is 13,544 voters, in Ghazni 10,904 voters, in Paktia 9,230 voters, in Paktika 8,173 voters, in Khost 8,018 voters, in Badakhshan 6,989 voters, in Kandahar 6,739 voters, in Herat 5,694 voters and in Baghlan 5,332 voters and the rest 39,088 differences in votes related to other 24 provinces.

The report, containing information about differences in voter data, was prepared by technical team of the IEC secretariat and international advisers using Dermalog letters about the 137,630 discrepancy between the same voters.

This information shows that the processed voter data of 4,563 polling stations is not equal to the biometric datasets and the difference in these stations is shown in two forms. In 1,466 polling stations, the number of voters in processed data is higher compared to that of biometric datasets transmitted and the difference is 30,608 votes.

However, the number of voters in processed data of 3,097 polling stations is lower compared to that of the biometric datasets and the difference is 168,238 votes. The total difference between processed data and biometric data of 4,563 polling stations is 137,630 voter records.

INSERT A TABLE FROM THE PRESENTATION (SEE TRANSLATED VERSION)

The IEC’s report explains the differences in data and the source of the discrepancies. This information is divided into 11 categories for accurate understanding.

PART ONE: THE NUMBER OF VOTES IN PROCESSED DATA IS HIGHER COMPARED TO THAT IN BIOMETRIC DATASETS

First five categories: 1) the results of multiple polling stations recorded by a single biometric device, 2) missing biometric data of some polling stations, 3) partial biometric data of some polling stations, 4) polling stations with difference of five or less than five and 5) polling stations with difference more than five votes in processed data and biometric data.

First category:

This category includes polling stations where the results were transferred by a single device and explained as follows:

• One device is used for transfer of result sheets of different polling stations using different SD cards;
• One device transferred equal number of figures it had already stored in its own memory, rather than transferring the data of SD cards of different polling stations;
• 115 SD cards are used in 36 biometric devices for taking and transferring of results in different provinces; and
• The number of voters in processed information is 3,740 while biometric data shows 2,438 voters. The difference between processed information and biometric data is 1,302 votes.

Second category:

Missing biometric data of devices whose results have been transferred:

• In some cases, no information from biometric devices could be extracted while the processed information of the polling stations exists;
• This may have many reasons and most probably due to missing SD cards as the result forms of the same polling stations have been transferred to the center; and
• Digital forms, photo forms and processed figures that contain 100,038 votes of 141 polling stations have been transferred to the central server.

Third category:

Partial biometric data transfer from some devices explained as follow:

• In some cases, all biometric data could not be transferred from computer to the server due to (errors in the file or the server did not identify the file) and that resulted in difference between figures;
• In 159 polling stations, the number of voters in biometric data is lower compared to that in processed data;
• The number of voters in processed data in these stations is 14,572 while the total biometric number of voters transferred to the server is 3,163. The total difference between processed data and biometric data is 11,409 votes; and

Cases in this category show that biometric data of voters has been attained but it has not been extracted due to some problems. This was probably due to unidentified files, which have been skipped after the system could not read them.

Fourth category:

Those polling stations where the difference between biometric datasets and processed data is five or less than five votes:

• Polling stations of which the number of voters in their processed data is five or less compared to their biometric data.
• In some cases, the difference of voters is one to five between processed data and biometric data in a polling station.
• This happened in 859 polling stations and their processed data shows 88,298 voters compared to 86,869 votes based on biometric data, which is varied by 1,429 votes.

Fifth category:

Polling stations, of which the number of voters in their processed data is more than five compared to their biometric data:

• The difference of votes between processed data and biometric data is more than five in some cases and this happened in 271 polling stations, and
• The number of votes based on processed data in these polling stations is 40,951 compared to 34,521 votes based on biometric data which shows a difference of 6,430 votes.

According to the IEC report, the technical logic of the fourth and fifth categories is as follows. The reason of differences in figures and information was probably due to technical problems in biometric registration software, which in some cases increased the number of voters by one.

PART TWO: THE NUMBER OF VOTERS IS LESS IN PROCESSED DATA THAN BIOMETRIC DATA

Six categories of the second part: 6) the number of voters is less in processed data compared to that in biometric data, which includes repeated uploading of records, 7) transfer of results of multiple polling stations by a single device, 8) polling stations that did not process data but their biometric data is transferred, 9) polling stations of which biometric data is transferred but their digital forms and 10) photos are not transferred, more than one device used for a single polling station. The eleventh category contains exceptional cases that do fit the other categories.

Sixth category:

The information below is about records uploaded twice to the system:

• Dermalog Company in a letter on November 4 said that some information related 1,929,333 votes have been entered twice to the system and 105 polling stations come in this category.
• The number of votes based on processed data is 11,114 compared to 13,301 votes according to biometric data.

Seventh category:

The results of multiple polling stations transferred by a single device:

• This category is similar to the first category, as one device is used for SD cards of different polling stations for transferring their photos and result sheets; but with the difference that the number of voters in biometric data is higher compared to the figures in processed data in this case.
• A total of 23 polling stations come in this category as the number of votes based on processed data is 785 compared to 1,440 in biometric data which shows a difference of 695 votes.

Eighth category:

Polling stations which have not processed data but their biometric data is transferred:

• A total of 2,105 polling stations related to 34 provinces of the country have no processed data but their biometric data has been transferred to the system.
• This category is divided into two parts. In the first part, this is because of not filling the digital forms of biometric devices and only the photo of the result sheet has been transferred, so the software did not record processed figures in the digital form.
• Most of these polling stations relate to quarantined biometric devices, of which their information is transferred by computer.
• A total of 2,053 polling stations containing 112,373 biometric votes in all provinces have no processed data.
• In the second part, in some polling stations, one device is used for recording biometric data while a new device is used for transfer of results and the new device showed zero processed data.
• Fifty two polling stations in 34 provinces of the country that contain 2,998 biometric votes also have no processed information.

Ninth category:

Polling stations of which biometric data is transferred but their digital forms and photos not transferred:

• In some cases, only biometric data from a polling station has been transferred while no digital forms, photos and result sheets were transferred.
• This happened due to the expiration of certificate in Dermalog server on October 16 as the server could not process the information being transferred to it.
• When the process started again, biometric data of these polling stations was added to the server but information related to result sheets and processed data did not appear there.
• A total of 94 polling stations in 11 provinces encountered this problem.
• In total, information of 8,320 biometric votes transferred from these devices.

Tenth category:

More than one device used in a single polling station:

• In 758 polling stations, more than one precautionary biometric device are used in a single polling station and it was probably due to faults in the device and draining battery, but only one device is used for transfer of result sheets.
• This issue caused only processed data of biometric device reached the system without biometric data.
• The number of voters in processed data in these polling stations is 44,515 compared to 83,251 based on biometric data which shows a difference of 37,736 votes.
• In explanation of this category, only the device which transferred the data included only its own data while the information recorded by other devices has been ignored.

Eleventh category:

This category includes 49 different polling stations with no differences related to figures mentioned in the previous categories. The number of votes in processed information is 4,142 compared to 7,341 votes based on biometric data.

Independent Election Commission (IEC):

Hawa Alam Nuristani, IEC Chairperson, on November 16, told a press conference that a detailed technical report of 137,630 voter records regarding their processed data and biometric data had been presented by technical teams during multiple discussions with presidential candidates.

About the Decision No. 108-2019 of the commission, she said that considering analytical information from secretariat of the commission and national and international specialists, the difference of 137,630 votes in the processed and biometric data was due to some technical problems created by Dermalog company. There were no special problems in this information.

An IEC official, who wished to go unnamed, said that 137,630 voter records were among the votes which would be audited and recounted.

About the 137,630 quarantined votes, the source said that these votes related to devices which encountered faults after recording many votes on the Election Day and precautionary devices were used instead.

Citing an example, the official said that a device failed after recording 50 votes and another device was used for continuing registration of voter data.

The source said the information of one of the two devices used in a polling station was entered directly from the device to the server while the information of another device was entered through computer to the server and that was one of the reasons that resulted in around 137,000 voter difference.

Javid Faisal, spokesman for presidential candidate Ashraf Ghani, called the IEC decision as important and said, “IEC decisions make sense and we support them and we ask candidates to respect independence of the commission and put the culture of opposition to an end.”

Pajhwok Afghan News could not contact Dr. Abdullah Abdullah’s team despite several attempts for comment in this regard.

However, the team in a resolution letter on November 6 said that the IEC should deal with four types of fraudulent votes including 137,630 unidentified votes in quarantine, 102,012 votes cast out of official time on the Election Day, around 700 missing devices and SD cards (among approximately 2,400 polling stations), and votes with duplicate faces (a reference to the 86,266 duplicate votes).

On November 5, presidential candidate, Rahmatullah Nabil said that the IEC was yet to address their demands and termed the vote recount and audit against the electoral law and procedure.

The former NDS head accused the IEC of trying to enter over 137,000 votes into the system that had been invalidated by Dermalog Company.

He said the country was facing different crises and the IEC should not confront the country with electoral crisis.

Nabil said that the country was passing a democratic test and security forces should not be used for the benefit of a few and against the public Afghans.

This comes as Dermalog Company on October 27 announced verifying 1,791,703 votes. It said that 137,630 other votes have been quarantined due to differences created by precautionary devices used on the Election Day.

Dermalog is a German company which provides biometric systems for the IEC.

On November 9, the IEC announced audit and recount of votes from 8,255 polling stations for separating fraudulent votes from clean ones, but the process faced opposition from a number of presidential candidates including Dr. Abdullah Abdullah, Gulbuddin Hekmatyar and Rahmatullah Nabil.

The IEC issued Decision No. 106-2019 on November 13 that stopped the audit and recount process of votes due to oppositions from the candidates and said the announcement of preliminary results of the election would be delayed.

However, the commission issued Decision NO. 110-2019 on November 17 that resumed the audit and recount process of votes after consulting with candidates.

The presidential election was held on September 28. Earlier the IEC had given October 19 the date for announcement of preliminary result while November 7 for announcement of final result of the presidential election.

But due to technical issues and problems during separation of clean and fraudulent votes, the IEC could not announce the result on its due date.

On October 27, the IEC said that primary result of the presidential election would be announced on November 14, but the commission again delayed for an indefinite period of time.

Azizullah Hamdard

Hits: 6253