While I applaud African Governments’ increased use of technology and e-government resources in tacking the many challenges the economy faces, we often see technology being leveraged to optimise only a subset of a process, leaving out other activities that remain vulnerable to abuse/mishaps. One such instance that still has lots of room for improvement is the electoral process.
For example, from news articles and coverage regarding the Kenyan elections, the focus was more on how the final result counting/tallying process was “hacked”, with little or no mention of how the preceding voting activities were conducted; thus, it can be assumed that these activities were not technology-assisted.
I need not emphasize that when it comes to voting, the outcome (results from counting of ballot papers) is dependent on the credibility of the preceding activities (registration, accreditation and casting of votes). As such, the concept of Garbage-In-Garbage-Out concept is invoked wherein “manipulated” votes, even when counted correctly (assisted by technology), lead to “manipulated” results.
From Kenya, Rwanda, Liberia, Angola to Nigeria, elections in Africa have had a great deal of challenges ranging from security to corruption especially. Time and time again, we hear of the (re)deployment of armed troops, snatching of ballot papers and boxes, under-aged voting, etc.
Apart from the damage these impart on the democracy of the Nation, they also incorporate many direct and indirect costs in the form of relocation allowances, re-printing of ballot papers etc; gulping funds that could be used for many other projects. In its 2015 presidential elections alone, Nigeria was reported to have spent $547 million (excluding costs incurred by political parties and their candidates).
With increased global attention and the call for more credible elections in Africa, we have started seeing advances in optimising other activities of the electoral process with initiatives such as the Nigerian Federal Ministry of Science and Technology announcing the launch of an electronic voting (e-voting) system that was to be evaluated by the Electoral Commission.
This article thus proposes leveraging technology developments in authorising voters and ensuring the integrity/consistency of their votes throughout the process of casting, counting and transmittal of voting results.
Note that this article is not intended to demine the use of physical ballots; however due to the frequent abuse of that system in Africa, a comprehensive digital approach is suggested to tackle the peculiar challenges facing the region’s electoral ecosystem.
To exemplify the concepts proposed in this article, I focus on elections in Nigeria but these can be applied in other countries as well.
Following this introduction, this article first summarises the current electoral process in Nigeria. Proposed improvements are then suggested, in the sequence of the electoral process. Afterwards, the article is concluded.
Note: INEC stands for Independent National Electoral Commission; the body empowered to conduct and organise democratic elections in Nigeria.
STATUS QUO
Just to recap and point out some challenges (marked with *), the current voting process in Nigeria entails:
STEP 1. The candidate makes their way to an INEC station for voter registration which includes capturing biometrics.
*The candidate may meet a large queue and might not be registered on that day;
STEP 2. On registration completion, the voter’s data is stored. The voter is issued a paper-based temporary voters’ card (TVC); which is subsequently replaced with a plastic-based permanent Voters’ card (PVC) after printing. INEC is expected to send the voter an SMS when their PVC is ready for collection.
*Registration database is decentralised so a candidate can register at multiple locations;
* Voter might not be notified of PVC readiness (e.g. due to technical issues) and can only be sure by visiting the registration station regularly;
Step 3. On the Election Day, the voter shows up at the polling station (where they registered) and their PVC is first verified as genuine using a card reader. If genuine, the card reader presents the bio-data of the voter including their picture and finger prints.
STEP 4. The Voter is authenticated by placing their finger on the fingerprint scanner on the card reader; this matches their fingerprint with the one embedded in the voter card (captured during Voter registration). Based on the discretion of the electoral officer, an unauthenticated voter (e.g. due to technical issues with the card reader) may still be allowed to go through to the next step; however, this should only occur after cross-checking the voter’s name on the list of voters registered at that station and the occurrence should be noted in an incident form.
*There is the risk that the electoral officer can “authenticate” whoever they favour;
*What if the voter has no finger?
Step 5: All voters accredited to vote are recorded in a register (accreditation register) and assigned a number based on when they were accredited. The electoral officer also places indelible ink on the finger of the accredited voter.
STEP 6. Voters are queued in an orderly manner beginning from the first person to arrive the voting area.
STEP 7. At their turn, the voter mentions their assigned number which is cross-checked on the accreditation register, to ensure that it is the same person that was accredited that has turned up to vote.
*Crosschecked details are limited to what has been captured on the accreditation register; hence, accredited voters can easily be replaced with unaccredited individuals (with ink on their fingers) who only need to memorise the details captured in the accreditation register (such as first or last name);
*The accreditation register itself could be altered to include unauthorised voters;
STEP 8. Once confirmed, a ballot paper is issued to the accredited voter to exercise their franchise; once completed the ballot paper is cast into the respective ballot box (e.g. Presidential, Gubernatorial, etc.).
*Voter is exposed to outside interference/side talks/threats to influence their choice;
STEP 9. After the last queued voter would have cast their ballot(s), the electoral officer sorts out the cast ballot papers to ensure that they have been cast to the appropriate box.
STEP 10. The votes are counted and the results are announced; announcements usually include number of voters accredited, number of qualified votes, number of disqualified votes (e.g. when a candidate voted for two candidates, instead of one), number of votes per party, winner of election at polling station.
*If not monitored, the electoral officer can substitute cast ballot papers or boxes with pre-casted ones selecting their “preferred” candidate;
*The voter cannot verify if their vote was counted or altered;
In addition to the activity-specific challenges, the following overarching challenges exist through the process:
• Generally a slow process, generating long queues and disruption to daily activities
• Environmental impact of printing ballot papers
• Possible disenfranchisement of individuals without fingers
• Harassment and disruption by armed bandits (e.g. snatching of ballot boxes)
• Risk of compromised (bribed) electoral officers manipulating the process or result
In tackling these challenges, I propose the following activities which leverage technology (almost) throughout the electoral process.
VOTER REGISTRATION
Candidates can register their details on a website and schedule an appointment for data verification and biometrics capture. Appointments can be scheduled to take about 20 minutes per candidate, using an appointment management system.
Registration details and biometrics are stored a central database such that if the candidate had previously registered at another location, they could be penalised. Candidates without fingerprints for biometric capture can be provided with an alternative authentication mechanism (e.g. pin, pattern, etc.)
After successful verification and capture, the candidate is confirmed as a voter and is provided with a smart voter’s card (SVC) in the form of a personalised computer security smart card. The purpose of the card is to ensure the candidate had gone through verification and is an eligible voter. It also acts as first level authentication for the e-voting system to be adopted.
The registration process should also consider and restrict the voter to the polls that they can participate in (e.g. Abuja resident might not be eligible to vote in a Lagos Governorship poll).
THE VENUE & POLLING “STATIONS” (KIOSKS)
Mobile portable computers (having similar form factors as self-service kiosks) with light specifications are proposed to be customised for the actual voting. These computers can comprise mainly a touchscreen, smartcard reader, fingerprint scanner and a portable CPU/board such as the Raspberry Pi. They can even be built from scratch in the country, providing employment through the production process.
Building such kiosks is estimated to be less expensive than the combined costs of providing dedicated laptops, smart card readers and ballot papers/boxes. An even less expensive alternative computer system can dismiss the CPU, touchscreen and fingerprint component by providing an inexpensive Android Tablet with in-build fingerprint scanning and smartcard functionalities.
The polling machine should be situated in “portable” or mobile cubicles; e.g. shipping containers (as there are so many empty containers lying around the country). The cubicle should consist of a window or air conditioning unit for ventilation, the polling kiosk and one CCTV camera facing the voter from a corner. The purpose of having the camera is two-fold – to ensure the candidate is not sabotaging the system; and to record the turnout of events if necessary.
The cubicle should have one-way access control in the sense that the voter can lock themselves in the cubicle with the electoral officer alone having an override access key in case of emergencies or voter delays. The cubicle can also be manned by one security official.
The electoral officer should be provided with a stall and demonstration polling kiosk which is disconnected from the live voting system, for educating voters that are not familiar with the voting process. They should also be provided with a queue management system wherein voters are provided with a queue number when they arrive to vote and if they are absent when their number is called, they have to get another number.
The process thus goes:
• Voter arrives at venue and take a number
• Voter enters and locks cubicle when it is their turn
• Voter votes and exits the cubicle
• Repeat until last vote
VOTER AUTHENTICATION
The polling kiosk’s default state is such that a lock screen is displayed requesting for the voter’s smartcard. All polling kiosks should have decentralised copies of voter authentication details (i.e. finalised central voter registration database) for faster authentication service.
Once the voter inserts their card, the system queries the central database of votes if the voter had previously voted. If so, the voter cannot make another vote. Voting in remote locations (without access to internet) is covered in a subsequent section.
If the voter has not previously voted, the system requests for their fingerprint to finish authentication; that way, a stolen card cannot be used without the presence of the owner. The voter is also authenticated to see if they are eligible to vote for the particular election/region being contested.
Voters without fingers or fingerprints are provided with the alternative of providing a pin, or other form of authentication.
CASTING THE BALLOT
On authentication, the voter should be presented with a screen of candidates and/or their political parties. These can be displayed with a radio button as the selection element since it allows for only one “option” to be selected.
If there are more than one election on that day (e.g. senate and president, etc.), the voter is presented with a screen of possible candidates for one position contested, before proceeding to another. On final candidate selection, the voter is asked to confirm their selections before final submission.
Depending on the authentication of the voter, polling screens can be restricted; i.e. a Lagos Governorship election voter, even if situated in Kaduna, can have the Kaduna polling screen restricted when their card is inserted, but has the Lagos one open.
MAINTAIN VOTE INTEGRITY DURING RESULT TRANSMISSION AND COLLATION
With this system, there is no physical movement of ballot boxes as all votes are made electronically.
The voting system would be supported by two Blockchain-based databases – one central and remote, the other local and decentralised.
On vote submission, both databases are updated in real-time and a hash value of the vote (transaction) is generated which is provided to the voter as confirmation and a means to verify that their vote was not altered.
The hash value of any vote is linked to the hash value of the preceding vote in the database such that any alteration would affect the hash of the subsequent vote. Note that these hash values are proposed to be store on both the central and local databases. The local ledger is used as a backup of original votes, in case of suspicious tampering of the remote ledger. Thus, any bad actor would have to update the entire database across the affected nodes on the network.
In the least, the remote/central database entries should contain a reference to the local database entries and these entries should be provided in a ledger. Voters should be provided with a portal where they can verify their votes which would emanate from the central database.
REMOTE LOCATIONS
Considering the state of power supply and the digital divide in Nigeria where a lot of communities are with little or no electricity and internet access, the following can be adopted:
• Tablet computers for e-voting:
• Since a lot of them can go for at least 12 hours without charging
• Electricity from solar panels:
• Solely for charging tablet computers
• Local Blockchain database alone:
• To be appended to the central database after the elections. In case voters might have voted at more than one decentralised location, this would flag such instances. In such instances however, any entry already on the central database should take precedence.
BENEFITS
Amongst other benefits, the proposed approach should bring about:
• Seamless voter registration;
• Voting by authorised electorate only;
• Reduced environmental footprint from reduced paper use;
• Cost savings associated with reduced paper use, amongst other resources;
• Flexible voting condition wherein voters are not confined to location of registration and movement does not have to be restricted;
• High availability of e-voting systems enabled via the decentralised failover structure
• Elimination of physical counting of ballots and real-time compilation of voting results (excluding remote locations)
• Verifiable integrity of votes;
• Privacy during actual polling;
• Less risk of bribing with an electoral official since they have minimal influence on the actual voting
THE OTHER STUFF
In this section, I detail some other things to consider in realising the proposed system:
• Transmitted data should be encrypted (e.g. via https)
• Database passwords should comprise multiple parts; i.e. the password to the databases can be a combination/concatenation of passwords from multiple parties. For example, if development of the e-voting system is outsourced, the developer can set a password like “ABC” and the INEC representative can set their password as “123”. Thus, the actual password to access the databases should be “ABC123”.
• One hashing algorithm that can be adopted is MD5 and the each new hash generated should be linked/linkable to previous hash (e.g. as a function of the voting timestamp and voter ID).
Database queries
CONCLUSION
This article is not intended to demine the use of physical ballot papers in the electoral process; but proposes e-voting as a preferred solution in Africa due to the peculiar challenges the region faces.
By adopting the proposals in this article, general elections just might be held in countries like Nigeria without the need to restrict movement or halt business activities as voters could easily “branch” their polling station, vote and leave with minutes.
As leading change management practices suggest, stakeholder education and enlightenment can aid a successful deployment of any new system such as this. A phased approach is also recommended; starting with pilots on less critical elections such as those of political parties’ primaries or Local Council Officials.
You might notice that I made no mention of a “mining” process from the proposed adaptation of the Blockchain/distributed database. Mining in this context, refers to the validation of the voters and votes by a representative number of other nodes. This is because the time it might take to validate one voter or vote might be significantly high considering the volume of requests from nodes at different locations, ultimately causing the voting process to span longer than expected; not to mention that the e-voting systems proposed to be adopted should be light-weight and might become overwhelmed with such voluminous requests. Furthermore, nodes in the remote locations with limited internet resources might be entirely cut out of mining process. Nevertheless, Blockchain adoption on voting is not a new topic and we just might see a broader approach being adopted in African elections.
Do let me know if you have suggestions, contrary opinions or further proposed improvements to the electoral process; either technology-assisted or not.
Jude Okosun, an IT Consultant writes from Lagos
and then