Algorithms in the activity of digital platforms

Размещено на http://www.allbest.ru/

Размещено на http://www.allbest.ru/

Algorithms in the activity of digital platforms

Yu.S. Razmetaeva

PhD in Law, Associate Professor at the

Department of Human Rights and Legal

Methodology,

Yaroslav Mudryi National Law University,

Ukraine, Kharkiv;

Visiting Researcher at the Faculty of Law,

Uppsala University, Sweden, Uppsala

Ю.С. РАЗМЄТАЄВА

кандидатка юридичних наук, доцентка, доцентка кафедри прав людини та юридичної методології Національного юридичного університету імені Ярослава Мудрого, Україна, м. Харків; запрошена дослідниця Уппсальського університету, Швеція, м. Уппсала

АЛГОРИТМИ В ДІЯЛЬНОСТІ ЦИФРОВИХ ПЛАТФОРМ

The platforms are quite often called “digital” or “online” in the synonymous sense of these terms. These platforms can be even more broadly described as “social platforms” (Cobbe, 2021, p. 740), which seems to be an accurate term from the point of view of their noticeable impact on social relations. Despite the fact that the influence of platforms on society is growing every day, the term “digital platforms” will be mainly used in this paper for two reasons: (1) platforms have gained serious influence and power precisely in the digital era, not least due to the immanent features it has, (2) a successful platform business model is built on digital technologies and data which are collected and processed thanks to new technologies.

The term “digital” can have different connotations and be applied in various contexts. The platforms can be described as digital regarding their function (Gawer, 2009), their ability to support online communities (Spagnoletti, Resca & Lee, 2015), their ability to provide collaboration and engagement through social media (Falco & Kleinhans, 2018), etc. This is also applicable to the term “online” platforms, which is becoming widespread mainly due to its use in recent legal acts adopted at the level of the European Union (EU). The European Commission (2022) in its approach to online platforms emphasizes their ability to rely on data, use of information and communication technologies to facilitate interactions between users, as well as network effects. Although the definition of online platforms is often closely related to the digital economy and markets, an interesting view at this type of platform includes studying them as corporations, which are different from traditional ones, including in the use of algorithms for decentralized control (Frenken & Fuenfschilling, 2020).

The platforms “now routinely use algorithmic behaviour modification techniques to manipulate users”behaviour” (Greene, 2022), and they make manipulation part of their economic model. They also “have the ability to shape the information published on the platform, and they profit financially form the interaction that users have with information present on their platforms” (Kettemann, 2022, p. 8). Justification by the economic benefits of algorithmic governing and shaping information, including limiting it, is an important element of platforms' activity. Moreover, it seems that there are attempts to present this purely economical justification as a certain race in which there must be winners, and societies can even be publicly declared such winners and beneficiaries. However, the actual beneficiaries are the platforms themselves - their owners to be clear.

To specify what “algorithms” mean here I will refer mainly to the smart algorithms which have become possible due to the introduction and deployment of digital technologies and especially artificial intelligence. Despite the fact that much of the discussion revolves around software algorithms, it is worth noting that the context can be broader. As Tarleton Gillespie (2013) describes this: “Algorithms need not be software: in the broadest sense, they are encoded procedures for transforming input data into a desired output, based on specified calculations” (p. 167).

An important feature of algorithms in connection with the activity of digital platforms is the lack or insufficiency of human supervision over automated and algorithmized processes. That is, algorithms “analyse data collected from the user in order to deliver content they would supposedly be interested in or more likely to engage with. They perform their activities automatically, with no regular human intervention or oversight” (Thiago, 2020, p. 609). Processes that are once given over to algorithms are rarely returned to human oversight, both for the reasons that algorithmization is economically profitable or at least promising in the long run, and for the reasons that algorithms are able to give those who own and manage them great power over others - from influencing opinions people to control their behavior.

Objectives of the paper. The purpose of this article is to identify the impact of algorithms in the activity of digital platforms and to outline the main existing and potential consequences of this for people and societies, with a particular focus on the consequences for such fundamental values as human rights, the rule of law and democracy. Taking into account the fact that these values are the basis of the European legal order, as well as some national legal orders, it is necessary to study the prospects of reducing the negative impact on them from the side of algorithms and those who are the owners and managers of the latter.

Main findings

There are certain implications that are manifested in the activity of digital platforms and which can be formulated as follows: (1) the platforms aime to maximum involvement users in spending time on a certain digital platform in every possible way; (2) the platforms create the impression that they increase the field of the choices that people make while significantly shaping these choices; (3) the platforms promote goods, services or opinions without distinguishing between what to promote and whether it should be promoted; (4) the platforms are shaping reality at all levels from shaping an individual feed in social media to the shaping public policy in certain areas; (5) the platforms declare commitment to fundamental values while interpreting them in a way that is beneficial to platforms activity or interpreting them in the way that platforms' owners think these values should be interpreted; (6) platforms are building up and expanding algorithmic governing, and contribute to imposing an algorithm-based approach in a wider social context. All this together forms the basis of the activity model of modern digital platforms which is justified mainly and purely by economic considerations, or rather, the benefit of platforms' owners and managers.

Considering COVID-19 situation, Marco Briziarelli and Emiliana Armano (2022) claim that: “digital abstract space represents the framework of social relations mediated by the digital in which machinic fix capital can move between contradictory states of abstraction and subjectification” (p. 55). In this sense, digital platforms have already occupied quite a large part of the digital space. Moreover, platforms make a significant contribution to the fact that social relations are increasingly transferred to the digital environment. By itself, such a transfer might not be dangerous if it were to some extent the evolutionary development of humanity. The problem is that against the background of digitization, which could be conventionally called natural, there is a push to shift the social into the digital form. It should be emphasized that this is in many ways an artificial, imposed choice and not a valuable civilizational transition.

Platforms thus make efforts to recklessly accelerate technological development, which differs from a reasonable, and not only purely economically justified approach. In addition, in most cases, digital platforms, as well as their owners and managers, avoid responsibility for how their activity affects human rights, democracy and the rule of law. As Koen Frenken and Lea Fuenfschilling (2020) wrote, “platforms manage their workforce with a capacity similar to traditional corporations and in the interest of its investors, but without the formal obligations that traditional corporations face regarding their employees and other stakeholders” (p. 103). The problem of the responsibility of digital platforms deepens both against the background of their growing power as owners of new technologies tools and against the background of the fact that they often operate in areas where traditional legal safeguards and measures are not sufficiently effective.

Accumulated power on the one hand and law's failure to keep up with the challenges of the digital era on the other give platforms the opportunity to maneuver and reconfigure their activity. As it was rightly pointed out: “If rulings provide any clarity at all, platforms can quickly re-code their software and/or alter their terms and conditions, creating a new artifact with slightly different workings that would necessitate a new court case, and so forth” (Frenken & Fuenfschilling, 2020, p. 107). In this context, the consequences of such phenomena as algorithmic governing and censorship, widespread in the activity of platforms, may turn out to be particularly dangerous. Besides, acting sometimes factually as governments act, digital platforms have the antifragility of business and not the antifragility of public institutions. This is reflected in their vulnerability and nonsustainability when facing to many challenges produced by crises, emergencies and conflicts. Vulnerability and non-sustainability of platforms ultimately affects people and societies, taking into account the interactions characteristic of the digital era and a significant part of the online activities of all actors in social relations.

1. Platforms ' algorithmic governing. The algorithms of digital platforms are geared towards retaining attention and greater involvement, often overlooking ethical business conduct in pursuit of these goals. They use not only an aggressive business model, but also shaping digital space and the visibility of something or someone in the agenda. Whatever is not in today's agenda is almost non-existent in the minds of people. In addition, the way in which algorithms attract and retain attention and how they shape the agenda is opaque, hidden from the public.

As it was rightly noted, “there is an assertive force about digital platforms able to transform the world in ways specific to their logics of operation” (Rossiter & Zehle, 2022, p. 34). This power could have been directed to the promotion of values and their support. For some time, it seemed that it was so. For example, social media platforms seemed like good spaces and tools for democratic discussions, unity of like-minded people, organization of protests in situations that required immediate public response. However, this turned into a powerful manipulation of users' opinions that spread far beyond the borders of digital spaces, polarization and radicalization, as well as the growing dependence of public opinion and public institutions on seemingly private digital platforms.

Lack of transparency in the activity of digital platforms may lead to violations of human rights, as well as the rule of law and democracy requirements, that will not be monitored in time, and the price of which may be exorbitant for individuals and communities. This problem can be further complicated by the fact that the definition of transparency itself is not established but rather refers to broad and vague principles.

Digital platforms are unlikely to reveal how they operate, including using algorithms, justifying this by the need to preserve commercial secrets. In addition, additional opacity can arise from the fact that some types of algorithms are poorly understood due to their nature or are extremely technically complex. As Dias Oliva Thiago (2020) mentioned, “the way platforms are using these technologies remains largely unclear, with references in transparency reports limited to mentioning if removed content was initially identified through automated flagging, without explaining the role that automated flagging had in the removal decisions” (p. 636). Removal of content that is recognized as inadmissible and its automated flagging as inadmissible can be done based on the platform's own rules, but without taking into account the balancing of freedom of expression with other rights and legitimate interests. Moreover, such actions may not be compatible with the requirements of the rule of law if we compare the procedures and rules for contesting the removal of information offered by platforms with the usual judicial procedures and rules available in national legal systems. This is especially important when it comes to information of public interest.

It is worth mentioning that decision-making based on smart algorithms is gaining momentum at all levels and is penetrating the private and public spheres. In this sense, platforms make a significant contribution to forming the habit of such decisions. This applies both to individual decisions that are made on the basis of algorithmic recommendations by people regarding their life choices, and to decisions that are made at the level of a community or an entire society and that partially or completely rely on algorithmic calculations.

Digital platforms are pushing societies towards total algorithmization. One of the mechanisms of such pushing is the production and promotion of such content and, in a broader sense, such forms of expression that are easily recognized and processed by algorithms. According to Tarleton Gillespie (2013): “There is a powerful and understandable impulse for producers of information to make their content, and themselves, recognizable to an algorithm. A whole industry, search engine optimization (SEO), promises to boost websites to the top of search results” (p. 184). For example, the texts we read online today are often designed in such a way that they are better not for humans but for algorithms. This can be achieved with the help of certain text structuring, the use of keywords that help bring the content higher in the search results, translations performed automatically and on the basis of AI tools.

It should be noted that one of the serious dangers of algorithmic governing, which has negative consequences for human rights at least, is the gradual elimination of people from processes, including decision-making processes. In the context of algorithmic disclosure co-regulation for platforms” business users, Fabiana Di Porto and Marialuisa Zuppetta (2021) argued that: “The human presence [...] is essential to monitor if errors occur in the building of the knowledge graph: technicians supervising in the sandbox may intervene to eventually deactivate any error that may occur in the algorithm” (p. 287). There is no way we can adjust the algorithm once and for all, leaving it in the future without human intervention, and get the results of this algorithm's activity that would correspond to fundamental values requirements. Firstly, the challenges that arise at the level of communities and societies are always dynamic, so certain elements of values can be revised or applied differently in different contexts. Secondly, it is necessary to monitor the algorithms to see if there are any biases or errors, since this cannot always be detected before the deployment of a particular algorithm. This also applies to human supervision of embedded technologies in terms of revising social practices. For example, these can be practices that algorithms follow or learn from, but which we currently consider or will consider unacceptable at some point (discriminatory, illegal, etc.). Thirdly, the variability of life circumstances is higher than any today's algorithm can take into account while working effectively. This means that there will be cases that will not be handled correctly by the algorithms because they deviate and that human supervision should at least follow up on such rare cases and solve them manually.

Undoubtedly, there must be a fair balance between innovation and the protection of values. At the same time, algorithmic governing and the application of technologies as such are not always what should be implemented as soon as possible, even if the real or declared goal is to promote, ensure and protect fundamental values and their elements. In particular, to eliminate discrimination in the workplace, hiring algorithms are used instead of in-person interviews. Parsing algorithms then withdraw all who are not giving their CV in proper machine-readable form. However, instead of contributing to the reduction of discrimination, such technological solutions may lead to its growth. They also can contribute to the emergence of new forms of inequality as an algorithmic discrimination, as it was in the well-known case with the Amazon hiring algorithm that learned from past discriminatory practices and created a pattern to hire men for some positions and not to hire women for these positions. With the growth of algorithmic governing, there will be more and more such cases.

2. Platforms' algorithmic censorship. The issue of content censorship by digital platforms is not new, however, it acquires new connotations against the background of their wide implementation of algorithms, especially content moderation algorithms. Jennifer Cobbe (2021) argued that: “the emergence of extensive algorithmic censorship as a primary form of content moderation by social platforms is an unwelcome development that gives rise to new forms of corporate societal authority” (p. 743). She wrote that it not only increases the power of the platforms but also enables them to insert commercial considerations into everyday communication between people.

The platforms regulate the understanding of what is freedom of speech and hate speech, relying primarily on their own rules, neither on human rights conventions, pacts or other legal acts, nor on legal doctrines established in the practice of authoritative international and national judicial institutions. In addition, they do not have mechanisms for balancing rights in conflict situations, which can be found in national and international law, and adjusted in line with judicial practice.

One of the examples for both cases - reliance on own rules and lack of proper balancing - is the examples of platforms' reacting to Ukrainians and Russians comments regarding Russian aggression against Ukraine. There are hundreds of comments, full of hate speech. written by Russians living in Russia or living in other countries but mentally located in Russia, under the posts about Ukraine, Russian aggression or life stories of Ukrainians in social media on Facebook or X (Twitter) platforms, which continue to remain on platforms and widely distributed. At the same time, many opinion leaders in Ukraine who violate community rules by allowing themselves offensive names for Russians or strong expressions, or even just posting their own photos of what really happened, were banned from the same platforms. Certainly, platforms can set rules, but they also have the properties of a public forum, especially in an environment where it is extremely important to convey an opinion and when this remains the only channel of communication. That raises the question of proportionality, balancing freedom of expression with other human rights, and in a broader sense the question of fundamental values.

Discussing the requirements of the rule of law, Nicolas Suzor (2018) writes about limits societies “should impose on the autonomy of platforms”. Despite the fact that autonomy itself is one of the legal and moral values it is not limitless. Digital platforms should not be uncontrolled, acting as seemingly private subjects in society and the legal order, whose activities should take place within the limits of legal norms and principles, but in fact are only partially covered by them. This is happening because platforms - like eels - are slipping out of the regulatory framework, operating in the growing decentralized digital space, having serious power and testing more and more new tools, including algorithmic ones. They should not avoid responsibility for certain consequences of their activity just because they are protected by autonomy and because the laws of the market supposedly dictate them to put business interests above fundamental values.

3. Platforms ' activity in emergency situations and conflicts. Digital platforms today rely on business models that in turn rely on the data and algorithms economy. One problem here is that these models are not well compatible with the requirements of legal norms and principles, especially at the level of fundamental values. Another problem is that the platforms rely on a business model that is not suitable for most emergencies.

In particular, due to the fact that digital platforms use algorithms that track peoples' interests and quite aggressively populate individuals' device screens with targeted and profiled results, it's platform algorithms that “decide” exactly what people will see and what not. Undoubtedly, this does not mean that the algorithms decide it in the sense in which we understand decision-making as humans. Rather, it is about the fact that algorithmic recommendations, issuing information and assigning it a place in the feed are grounded on calculations that the algorithm performs on the basis of given programs; or on the basis of formed patterns if it is a self-learning algorithm. It does not happen based on value, instinctive or experiential choice. Therefore, if we are talking about an emergency situation, the reactions of the algorithm will be different from human reactions. Moreover, in the vast majority of cases, there will be no reactions that would meet the challenges of extraordinary circumstances, until people reprogram or retrain the algorithm, as was the case during the COVID-19 pandemic.

In the emergency situations of wars and conflicts, natural disasters and other disasters, this can be an extremely serious problem. For example, it is a significant issue in the situation with Russian aggression against Ukraine, and in particular after the full-scale invasion. If social media feed for several months after 24 February 2022 was literally crammed with advertisements of everything with the words “Ukraine” and “war” (mostly requests for donations to various international organizations), posts from ordinary people, from those who offered help to others, or were looking for how to cooperate to temporarily leave particular places, or wrote that they saw Russian armored personnel carriers on a nearby street, - all these posts were taken down in the feed, to a bottom, and often went unnoticed.

It cannot be said that digital platforms do nothing in situations of emergency and conflict, or at least that they do not declare their commitment to proper actions. However, what they are doing or trying to do is not enough, at least when compared to the amount of power and impact they have in the digital age.

Recently, attempts to focus on the construction of so-called ecosystems of artificial intelligence or algorithms, or ecosystems of information technologies in a broader sense, are gaining popularity. On the one hand, this reflects attempts to act ecologically in all senses of the word, and on the other hand, it works for a positive image of algorithmization, primarily because of the associations that the term “ecosystem” has. According to Ned Rossiter and Soenke Zehle (2022) even “a focus on “green IT” itself can't critically address the advance of automated decision-making or the algorithmic bias in the data used to train intelligent systems” (p. 42). It must be borne in mind that the deployment of algorithms can be accompanied by negative consequences for fundamental values that are poorly monitored or that are significantly delayed in time, or have an accumulative effect.

Discussing legal technologies Ryan Whalen (2022) wrote that “the more directly a technology engages with legal rules, the more likely it is to raise normative legal issues, such as concerns about justice, equality and democracy” (p. 67). Algorithmic technologies definitely belong to such technologies. It would be very unwise to treat them without caution, no matter the huge economic benefits, impressive achievements for people and corporations, and the realization of the dreams of mankind, the introduction of such technologies would not promise.

Conclusions

digital platform human rights

Digital platforms use an aggressive business model and promote the ill-advised acceleration of algorithmization, as well as shape the digital space and social relations beyond the digital. They govern an online environment with the help of algorithms, while at the same time occupying a niche that is not regulated by legal instruments quite successfully. The algorithms of platforms are not sensitive to situations of conflict or war, they do not rely on special legal regimes or principles that could be something like the Geneva Conventions online. All this can have fatal consequences for the fundamental values that are the basis of legal orders and the common existence of people in democratic societies.

References

1. Briziarelli, M., & Armano, E. (2022). Domus Capitalismi: Abstract spaces and domesticated subjectivities in times of Covid-19. In E. Armano, M. Briziarelli, & E. Risi (Eds.), Digital platforms and algorithmic subjectivities (pp. 47-61). University of Westminster Press. https://doi.org/10.16997/book54.d

2. Cobbe, J. (2021). Algorithmic censorship by social platforms: Power and resistance.

Philosophy & Technology, 34, 739-766. https://doi.org/10.1007/s13347-020-00429-0

3. Di Porto, F., & Zuppetta, M. (2021). Co-regulating algorithmic disclosure for digital platforms. Policy and Society, 40(2), 272-293. https://doi.org/10.1080/14494035.202

0. 1809052

4. Falco, E., & Kleinhans, R. (2018). Beyond technology: Identifying local government challenges for using digital platforms for citizen engagement. International Journal of Information Management, 40, 17-20. https://doi.org/10.1016/j.ijinfomgt.2018.01.007

5. Frenken, K., & Fuenfschilling, L. (2020). The rise of online platforms and the triumph of the corporation. Sociologica, 14(3), 101-113. https://doi.org/10.6092/issn.1971- 8853/11715

6. Gawer, A. (2009). An introduction. In A. Gawer (Ed.), Platforms, markets and innovation (pp. 1-18). Edward Elgar Publishing.

7. Gillespie, T. (2013). The relevance of algorithms. In T. Gillespie, P. J. Boczkowski, & K. A. Foot (Eds.), Media technologies: Essays on communication, materiality, and society (pp. 167-193). MIT Press.

8. Greene, T., Martens, D., & Shmueli, G. (2022). Barriers to academic data science research in the new realm of algorithmic behaviour modification by digital platforms. Nature Machine Intelligence, 4, 323-330. https://doi.org/10.1038/s42256-022-00475-7

9. Kettemann, M. C. (Ed.). (2022). How platforms respond to human rights conflicts online: Best practices in weighing rights and obligations in hybrid online orders. Verlag Hans-Bredow-Institut. https://doi.org/10.21241/ssoar.8187

10. Rossiter, N., & Zehle, S. (2022). Platform politics and a world beyond catastrophe. In E. Armano, M. Briziarelli, & E. Risi (Eds.), Digital platforms and algorithmic subjectivities (pp. 33-46). University of Westminster Press. https://doi.org/10.16997/ book54.c

11. Spagnoletti, P., Resca, A., & Lee, G. (2015). A design theory for digital platforms supporting online communities: A multiple case study. Journal of Information Technology, 30, 364-380. https://doi.org/10.1057/jit.2014.37

12. Suzor, N. (2018). Digital constitutionalism: Using the rule of law to evaluate the legitimacy of governance by platforms. Social Media + Society. https:// doi.org/10.1177/2056305118787812

13. Shaping Europe's digital future. Online platforms. (2022, June 7). European Commission. https://digital-strategy.ec.europa.eu/en/policies/online-platforms

14. Thiago, D. O. (2020). Content moderation technologies: Applying human rights standards to protect freedom of expression. Human Rights Law Review, 20(4), 607-640. https://doi.org/10.1093/hrlr/ngaa032

15. Whalen, R. (2022). Defining legal technology and its implications. International Journal of Law and Information Technology, 30, 47-67. https://doi.org/10.1093/ijlit/ eaac005

Размещено на Allbest.ru