Abstract
In this paper, we argue that because of the advent of Artificial Intelligence, the secret ballot is now much less effective at protecting voters from voting related instances of social ostracism and social punishment. If one has access to vast amounts of data about specific electors, then it is possible, at least with respect to a significant subset of electors, to infer with high levels of accuracy how they voted in a past election. Since the accuracy levels of Artificial Intelligence are so high, the practical consequences of someone inferring one’s vote are identical to the practical consequences of having one’s vote revealed directly under an open voting regime. Therefore, if one thinks that the secret ballot is at least partly justified because it protects electors against voting related social ostracism and social punishment, one should be morally troubled by how Artificial Intelligence today can be used to infer individual electors’ past voting behaviour.
Similar content being viewed by others
Availability of data and material
N/A.
Notes
Throughout the paper, we use the term ‘AI’ rather loosely. It denotes several different data analytics techniques, including machine learning and deep neural networks. Nothing in the argument hinges on this, however.
Elklit and Maley, for instance, write: “The secret ballot remains important in protecting voters from forms of pressure, especially within families, that fall into something of a grey area between illegal coercion and legitimate persuasion”(Elklit and Maley 2019: 66). Similarly, Vandamme writes: “As I already suggested, there are prudential reasons for defending the secret ballot. They make us prefer secret ballots for their beneficial effects on vulnerable voters. If, for example, we believe that the voices of vulnerable people are of primary importance in the democratic quest for justice, we might want to make sure that they will be in a position to defend their legitimate interests or express a judgement free from domination” (Vandamme 2018: 392). A voter is certainly not free from domination if she knows that the price for voting in accordance with her political preference is that she will be looked upon with disdain by her co-workers and excluded from social activities at her workplace.
Elklit and Maley write: “Second, the secret ballot can be seen as an instrument that protects voters from the possibility of violence or other coercive action intended to influence their voting decision or to punish them for having voted in a particular way (or indeed for having voted at all)” (Elklit and Maley 2019: 65). Manin claims that open voting in general elections has three undesirable implications, one of which is that it increases the importance of private rewards and punishments in elections (Manin 2015: abstract). J. S. Mill also recognized this justification for the secret ballot. At one point he writes "Thirty years ago it was still true that in the election of members of Parliament the main evil to be guarded against was that which the [secret] ballot would exclude—coercion by landlords, employers, and customers” (Mill 1861: chapter 10). Coercion by, say, a landlord can take either of two forms. First, it can be an ex ante threat to the effect that the tenant will be evicted if she does not vote in accordance with the landlord’s precepts. Second, it can be an ex post punishment to the effect that the tenant will be evicted if she has not voted in accordance with the landlord’s precepts.
For an overview of the content of these lists, and where to purchase access to them, see http://voterlist.electproject.org/home (Accessed May 3, 2022).
Elections where three or more parties/candidates are on the ballot constitute elections where electors face a non-binary voting choice. The same is true of elections where there are two parties/candidates and where electors have the option of returning a blank ballot.
See (Gebru et al. 2017).
A recent study shows that based on Facebook likes alone, it is possible to predict individual electors’ voting intentions in a multi-party system with 60-70 percent accuracy (Kristensen et al. 2017). A recent article asks “Can we predict the voting behavior of Facebook users from their public Facebook profile?”(Idan and Feigenbaum 2019: 816). One model has an accuracy score of 82,5% when it comes to predicting whom an elector would vote for in the 2016 US Presidential election (Idan and Feigenbaum 2019: 823).
The interested reader can try The Economist prediction tool herself at: https://www.economist.com/graphic-detail/2018/11/03/how-to-forecast-an-americans-vote (Accessed May 4, 2022). The Economist prediction tool is not an AI in any way, but it highlights how few data points it requires to reach high levels of predictive accuracy.
Thanks to an anonymous reviewer for bringing this point to our attention.
For a great overview of empirical studies showing how bad people generally are at probabilistic reasoning, see (Reani et al. 2019).
For example, Mary was trivially late with her latest payment of rent or Mary has failed to furnish her balcony according to the exact regulations of St. John's. According to the lease, both offences are grounds for eviction.
AI constitutes a relatively new technology, and we conjecture that such technology will become more accurate, powerful and accessible in the future when it, with increased speed and reduced costs, can analyse an increased number of data points. It is likely that there will be some diminishing marginal accuracy at some point because each extra data point will begin to correlate so highly with already existing data that each new point does little to increase accuracy. However, this does not undermine our main point: namely that a future Mary’s landlord can contract with a data analytics company that has at its disposal an improved AI technology as compared to the technology of the company Mary’s landlord contracts with. See (Parikh, Obermeyer, and Navathe 2019: 810) for an example of the impressive development path of AI-based predictive algorithms within the field of medicine.
See for example this: https://newmillsproperties.com/artificial-intelligence-rental-market/#:~:text=Using%20Artificial%20Intelligence%20Rentberry%2C%20in,to%20recommend%20appropriate%20rental%20prices (Accessed May 3, 2022).
See https://www.nbcnews.com/tech/tech-news/tenant-screening-software-faces-national-reckoning-n1260975 (Accessed May 3, 2022).
See https://www.wsj.com/articles/SB996702441926667410 (Accessed May 3, 2022).
We wish to thank two anonymous reviewers from AI & Society: Knowledge, Culture and Communication for insightful and constructive comments on an earlier version of this paper.
References
Brayne S, Christin A (2021) Technologies of crime prediction: the reception of algorithms in policing and criminal courts. Soc Probl 68(3):608–624. https://doi.org/10.1093/socpro/spaa004
Brennan G, Pettit P (1990) Unveiling the vote. Brit J Polit Sci 20(03):311–333
Christiano T (2021) Algorithms, manipulation, and democracy. Can J Philos 1–16
Dahl RA (1998) On democracy. Yale University Press, New Haven
Elklit J, Maley M (2019) Why ballot secrecy still matters. J Democr 30(3):61–75
Engelen B, Thomas V, Nys R (2013) Against the secret ballot: Toward a new proposal for open voting. Acta Politica 48(4):490–507
Erman E, Furendal M (2022) The global governance of artificial intelligence: some normative concerns. Moral Philos Polit. https://doi.org/10.1515/mopp-2020-0046
Gebru T, Krause J, Wang Y, Chen D, Deng J, Aiden EL, Fei-Fei Li (2017) Using deep learning and Google Street View to estimate the demographic makeup of neighborhoods across the United States. Proc Natl Acad Sci 114(50):13108. https://doi.org/10.1073/pnas.1700035114
Hinds J, Joinson A (2019) Human and computer personality prediction from digital footprints. Curr Dir Psychol Sci 28(2):204–211. https://doi.org/10.1177/0963721419827849
Idan L, Feigenbaum J (2019) Show me your friends, and I will tell you whom you vote for: predicting voting behavior in social networks. In: Proceedings of the 2019 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining
Khemasuwan D, Colt HG (2021) Applications and challenges of AI-based algorithms in the COVID-19 pandemic. BMJ Innov 7(2):387. https://doi.org/10.1136/bmjinnov-2020-000648
Kosinski M (2021) Facial recognition technology can expose political orientation from naturalistic facial images. Sci Rep 11(1):1–7
Kristensen JB, Albrechtsen T, Dahl-Nielsen E, Jensen M, Skovrind M, Bornakke T (2017) Parsimonious data: how a single Facebook like predicts voting behavior in multiparty systems. PLoS ONE 12(9):e0184562. https://doi.org/10.1371/journal.pone.0184562
Lever A (2007) Mill and the secret ballot: beyond coercion and corruption. Utilitas 19(3):354–378
Lever A (2015) Privacy and democracy: what the secret ballot reveals. Law, Cult Hum 11(2):164–183. https://doi.org/10.1177/1743872112458745
Manin B (2015) Why open voting in general elections is undesirable. In: Elster J (ed) Secrecy and publicity in votes and debates. Cambridge University Press, New York
Mares I (2015) From open secrets to secret voting: democratic electoral reforms and voter autonomy. Cambridge University Press, New York
Mill JS (1861) Considerations on representative government. Parker, Son, and Bourn, London
Nickerson DW, Rogers T (2014) Political campaigns and big data. J Econ Perspect 28(2):51–73
Parikh RB, Obermeyer Z, Navathe AS (2019) Regulation of predictive analytics in medicine. Science 363(6429):810. https://doi.org/10.1126/science.aaw0029
Peters U (2022) Algorithmic political bias in artificial intelligence systems. Philos Technol 35(2):1–23
Reani M, Davies A, Peek N, Jay C (2019) Evidencing how experience and problem format affect probabilistic reasoning through interaction analysis. Front Psychol 10:1548–1548. https://doi.org/10.3389/fpsyg.2019.01548
Roberts SL (2019) Big data, algorithmic governmentality and the regulation of pandemic risk. Eur J Risk Regul 10(1):94–115. https://doi.org/10.1017/err.2019.6
Rona-Tas A (2020) Predicting the future: art and algorithms. Soc Econ Rev 18(3):893–911. https://doi.org/10.1093/ser/mwaa040
Skeem J, Lowenkamp C (2020) Using algorithms to address trade-offs inherent in predicting recidivism. Behav Sci Law 38(3):259–278. https://doi.org/10.1002/bsl.2465
Sturgis D (2005) Is voting a private matter? J Soc Philos 36(1):18–30. https://doi.org/10.1111/j.1467-9833.2005.00255.x
Sønderholm J, Mainz J, Uhrenfeldt R (2021) Big data analytics and how to buy an election. Public Aff Q 35(2):119–139. https://doi.org/10.2307/48616070
Thompson DF (2002) Just elections : creating a fair electoral process in the United States. University of Chicago Press, Chicago, London
United Nations (General Assembly) (1948) Universal declaration of human rights. United Nations, New York
Vandamme P-E (2018) Voting secrecy and the right to justification. Constellations 25(3):388
Acknowledgements
The authors would like to thank audiences at Aalborg University, Amsterdam University and Aarhus University for very useful discussions. Also thanks to David Paaske, Beate Rössler, Axel Gosseries, and James Stacey Taylor for useful written comments on an earlier version of this paper.
Funding
JTM’s work is written as part of a research project funded by the Carlsberg Foundation (CF20-0257).
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
N/A.
Ethical approval and consent to participate
N/A
Consent for publication
All authors consent.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Springer Nature or its licensor holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Mainz, J.T., Sønderholm, J. & Uhrenfeldt, R. Artificial intelligence and the secret ballot. AI & Soc 39, 515–522 (2024). https://doi.org/10.1007/s00146-022-01551-7
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00146-022-01551-7