ICT Innovation Cycles and their Relation to the Electoral Cycle
ICT systems evolve very quickly compared to the duration of an electoral cycle. Most ICT equipment, both hardware and software, has a useful life of about 3–5 years. After this period, equipment needs to be either replaced or significantly refurbished and upgraded. For electoral ICT applications, this can pose a significant challenge. With many electoral cycles lasting about four years, an ICT system that was successfully deployed for one election can be expected to require major upgrades or even a complete replacement for the next election.
Where the time between elections and use approaches the end of the useful life of ICT equipment, the sustainability and efficiency of purchasing, owning, storing and maintaining equipment needs to be considered. To avoid the high costs of storing unused equipment, EMBs can lease selected equipment, share it with other countries, or plan to reuse, sell or donate some of it after an election for non-electoral use.
Figure 12 provides a visual comparison of how operating systems, data transfer technology, mobile computing and fingerprint standards have evolved between elections, assuming a four-year electoral cycle starting in 1992.
The ‘Bathtub Curve’ and Electoral Cycles
Another challenge related to long electoral cycles and the comparatively short useful life of ICT systems becomes evident when considering the bathtub curve, a concept that is widely used in engineering to describe the occurrence of failures in technical systems.
The curve starts with a ‘burn-in’ phase in the early stages of using a system. This phase is characterized by the high failure rate of a still-immature system. As problems are detected and addressed, the failure rate decreases and the system enters the period of its useful life with a very low failure rate. Finally, as the useful life of a system comes to an end, wear and tear lead to an increased failure rate and require the eventual upgrade or replacement of the system.
While most non-election systems will be used primarily during the second phase of this curve, the sporadic nature of elections implies that electoral technology will often be used only shortly after its development and may therefore still be in the failure-prone burn-in phase. The period of useful life of election technology therefore coincides mostly with the time between elections when the system is not used. And when the next election occurs, after several years, the system may already be close to or in its wear-out phase with increasing failure rates.
The bathtub curve demonstrates that long preparation times are required for the introduction of electoral technology to make sure the burn-in phase takes place before election day, and that the time between elections must be used to identify and implement all upgrades required for the next election period.
Total Cost of Ownership of ICT Systems
One of the main claims offered in support of elections technology is cost effectiveness. However, experience suggests that the initial capital investment required to purchase an ICT system can be as low as 25 per cent of its total cost of ownership (TCO) over its expected lifetime. The TCO is the sum of all direct and indirect costs involved in purchasing, operating and eventually disposing of the equipment. It can be easy to underestimate the TCO when only looking at the initial purchase price of ICT equipment.
The TCO of electoral ICT systems includes the costs of:
- public dialogue about the introduction of the new system;
- purchasing or developing system hardware and software licences;
- system certifications and auditing;
- securing the system physically and technically;
- preparing the infrastructure required to operate the system;
- deploying, configuring and testing the system for an election;
- training staff and operators to use the system;
- operating the equipment during the election;
- retrieving the equipment after the election;
- storing and protecting the systems between elections;
- disaster recovery: preparing for back-up planning;
- replacements or upgrades, possibly for every election; and
- disposal of outdated equipment.
Many of the costs above are difficult to quantify, which makes it difficult to establish the precise costs of using an ICT solution, to compare ICT costs between different solutions, or even to estimate any cost savings incurred by automating parts of the electoral process. While a realistic estimation of the TCO and securing the required resources is necessary to effectively control and manage the use of electoral technology, claims about savings through introducing new technologies or between different types of technologies should always be carefully examined.
ICT Security and Costs
ICT security works like preventive medicine: one does not know if it is working or how well it is working until it fails. Moreover, there is no perfect security. Securing electronic systems is an arms race between those trying to protect them and those prepared to invest in breaking into them. Electoral stakeholders often demand the highest possible security standards for election technology, since a country’s democracy is at stake when election technology seriously fails or is manipulated. But the highest security levels are expensive. Each EMB must determine, in consultation with all electoral stakeholders, how at risk its ICT systems are; how likely failure or various types of attacks are; what the consequences of such events are; and how much security investment is needed, possible and justifiable.
Internet-based systems raise additional security concerns since they are connected to a public network and thus exposed to an unknown threat environment with similarly unknown threat capabilities. This environment includes not only national actors but also, by virtue of the online environment, foreign governments and hackers.