Machine learning models are becoming increasingly prevalent in algorithmic trading and investment management. The spread of machine learning in finance challenges existing practices of modelling and model use and creates a demand for practical solutions for how to manage the complexity pertaining to these techniques. Drawing on interviews with quants applying machine learning techniques to financial problems, the article examines how these people manage model complexity in the process of devising machine learning-powered trading algorithms. The analysis shows (...) that machine learning quants use Ockham’s razor – things should not be multiplied without necessity – as a heuristic tool to prevent excess model complexity and secure a certain level of human control and interpretability in the modelling process. I argue that understanding the way quants handle the complexity of learning models is a key to grasping the transformation of the human’s role in contemporary data and model-driven finance. The study contributes to social studies of finance research on the human–model interplay by exploring it in the context of machine learning model use. (shrink)

This article examines what the rise in machine learning systems might mean for social theory. Focusing on financial markets, in which algorithmic securities trading founded on ML-based decision-making is gaining traction, I discuss the extent to which established sociological notions remain relevant or demand a reconsideration when applied to an ML context. I argue that ML systems have some capacity for agency and for engaging in forms of collective machine behaviour, in which ML systems interact with other machines. (...) However, ML-based collective machine behaviour is irreducible to human decision-making and thereby challenges established sociological notions of financial markets. I argue that such behaviour can nonetheless be analysed through an adaptation of sociological theories of interaction and collective behaviour. (shrink)

The effect of COVID-induced public anxiety on stock markets, particularly in European stock market returns, is examined in this research. The search volumes for the notion of COVID-19 gathered by Google Trends and Wikipedia were used as proxies for COVID-induced public anxiety. COVID-induced public anxiety was shown to be linked with negative returns in European stock markets when a panel data method was used to a sample of data from 14 European stock markets from January 2, 2020 to September 17, (...) 2020. Using an automated trading system, we used this finding to suggest investment methods based on COVID-induced anxiety. The findings of back-testing indicate that these techniques have the potential to generate exceptional profits. These results have significant consequences for government officials, the media, and investors. (shrink)

Algorithmic decision-making based on profiling may significantly affect people’s destinies. As a rule, however, explanations for such decisions are lacking. What are the chances for a “right to explanation” to be realized soon? After an exploration of the regulatory efforts that are currently pushing for such a right it is concluded that, at the moment, the GDPR stands out as the main force to be reckoned with. In cases of profiling, data subjects are granted the right to receive meaningful (...) information about the functionality of the system in use; for fully automated profiling decisions even an explanation has to be given. However, the trade secrets and intellectual property rights (IPRs) involved must be respected as well. These conflicting rights must be balanced against each other; what will be the outcome? Looking back to 1995, when a similar kind of balancing had been decreed in Europe concerning the right of access (DPD), Wachter et al. (2017) find that according to judicial opinion only generalities of the algorithm had to be disclosed, not specific details. This hardly augurs well for a future right of access let alone to explanation. Thereupon the landscape of IPRs for machine learning (ML) is analysed. Spurred by new USPTO guidelines that clarify when inventions are eligible to be patented, the number of patent applications in the US related to ML in general, and to “predictive analytics” in particular, has soared since 2010—and Europe has followed. I conjecture that in such a climate of intensified protection of intellectual property, companies may legitimately claim that the more their application combines several ML assets that, in addition, are useful in multiple sectors, the more value is at stake when confronted with a call for explanation by data subjects. Consequently, the right to explanation may be severely crippled. (shrink)

The article discusses the use of algorithmic models in finance (algo or high frequency trading). Algo trading is widespread but also somewhat controversial in modern financial markets. It is a form of automated trading technology, which critics claim can, among other things, lead to market manipulation. Drawing on three cases, this article shows that manipulation also can happen in the reverse way, meaning that human traders attempt to make algorithms ‘make mistakes’ by ‘misleading’ them. These attempts (...) to manipulate are very simple and immediately transparent to humans. Nevertheless, financial regulators increasingly penalize such attempts to manipulate algos. The article explains this as an institutionalization of algo trading, a trading practice which is vulnerable enough to need regulatory protection. (shrink)

In a talk in 2013, Karin Knorr Cetina referred to ‘the interaction order of algorithms’, a phrase that implicitly invokes Erving Goffman's ‘interaction order’. This paper explores the application of the latter notion to the interaction of automated-trading algorithms, viewing algorithms as material entities (programs running on physical machines) and conceiving of the interaction order of algorithms as the ensemble of their effects on each other. The paper identifies the main way in which trading algorithms interact (via electronic (...) ‘order books’, which algorithms both ‘observe’ and populate) and focuses on two particularly Goffmanesque aspects of algorithmic interaction: queuing and ‘spoofing’, or deliberate deception. Following Goffman's injunction not to ignore the influence on interaction of matters external to it, the paper examines some prominent such matters. Empirically, the paper draws on documentary analysis and 338 interviews conducted by the author with high-frequency traders and others involved in automated trading. (shrink)

This article addresses a dilemma about autonomous vehicles: how to respond to trade-off scenarios in which all possible responses involve the loss of life but there is a choice about whose life or lives are lost. I consider four options: kill fewer people, protect passengers, equal concern for survival, and recognize everyone’s interests. I solve this dilemma via what I call the new trolley problem, which seeks a rationale for the intuition that it is unethical to kill a smaller number (...) of people to avoid killing a greater number of people based on numbers alone. I argue that killing a smaller number of people to avoid killing a greater number of people based on numbers alone is unethical because it disrespects the humanity of the individuals in the smaller-numbered group. I defend the recognize-everyone’s-interests algorithm, which will probably kill fewer people but will not do so based on numbers alone. (shrink)

To address the rising concern that algorithmic decision-making may reinforce discriminatory biases, researchers have proposed many notions of fairness and corresponding mathematical formalizations. Each of these notions is often presented as a one-size-fits-all, absolute condition; however, in reality, the practical and ethical trade-offs are unavoidable and more complex. We introduce a new approach that considers fairness—not as a binary, absolute mathematical condition—but rather, as a relational notion in comparison to alternative decisionmaking processes. Using US mortgage lending as an example (...) use case, we discuss the ethical foundations of each definition of fairness and demonstrate that our proposed methodology more closely captures the ethical trade-offs of the decision-maker, as well as forcing a more explicit representation of which values and objectives are prioritised. (shrink)

There is growing concern that decision-making informed by machine learning (ML) algorithms may unfairly discriminate based on personal demographic attributes, such as race and gender. Scholars have responded by introducing numerous mathematical definitions of fairness to test the algorithm, many of which are in conflict with one another. However, these reductionist representations of fairness often bear little resemblance to real-life fairness considerations, which in practice are highly contextual. Moreover, fairness metrics tend to be implemented in narrow and targeted toolkits that (...) are difficult to integrate into an algorithm’s broader ethical assessment. In this paper, we derive lessons from ethical philosophy and welfare economics as they relate to the contextual factors relevant for fairness. In particular we highlight the debate around acceptability of particular inequalities and the inextricable links between fairness, welfare and autonomy. We propose Key Ethics Indicators (KEIs) as a way towards providing a more holistic understanding of whether or not an algorithm is aligned to the decision-maker’s ethical values. (shrink)

Here, we provide an ethical analysis of discrimination in private insurance to guide the application of non-discriminatory algorithms for risk prediction in the insurance context. This addresses the need for ethical guidance of data-science experts, business managers, and regulators, proposing a framework of moral reasoning behind the choice of fairness goals for prediction-based decisions in the insurance domain. The reference to private insurance as a business practice is essential in our approach, because the consequences of discrimination and predictive inaccuracy in (...) underwriting are different from those of using predictive algorithms in other sectors. Here we focus on the trade-off in the extent to which one can pursue indirect non-discrimination versus predictive accuracy. The moral assessment of this trade-off is related to the context of application—to the consequences of inaccurate risk predictions in the insurance domain. (shrink)

Wittgenstein's conception of the general form of a truth function given in thesis 6 can be presented as a sort of a trade-off: the author of the Tractatus is unable to reconcile the simplicity of his original idea of a series of forms with the simplicity of his generalisation of Sheffer's stroke; therefore, he is forced to sacrifice one of them. As we argue in this paper, the choice he makes – to weaken the logical constraints put on the concept (...) of a series of forms, thus effectively metaphorising that concept for the sake of upholding the N-operation's role of generating the series – is unfortunate. An actual expansion of a series of truth functions as defined in 6 would require either making decisions at each step (Anscombe) or outwardly rejecting the concept of a series (Sundholm). However, neither is faithful to Wittgenstein's own fundamental intuitions regarding the nature of logic. For this reason, a different trade-off that prioritises upholding the basic features of a series of forms over the simplicity of the operation that generates that series seems to be much more reasonable. We offer such a trade-off by developing the schema already present in the Tractatus (5.101). The key element of our alternative solution is the construction of an operation that can perform the task of producing all consecutive truth functions of a given collection of atomic propositions as an invariant difference between any base and its successor throughout the series. We define it as a composition of three functions (in an algebraic sense): the first turns a symbol of a given truth function into a binary number, the second increments that number, the third turns the result back into a symbol of another truth function. We also show how the operation can be defined by means of a different notation without invoking binary arithmetic. (shrink)

This article puts into perspective the practice of financial regulation in contemporary financial markets, while a new normative order has emerged. This order, heralded by algorithmic technologies, changes the conditions for the exercise of regulation: to date, it has not yet been fully acknowledged nor understood by regulatory bodies. Computer code, replacing speech and writing, induces a changeover from one normative order to another in contemporary markets: the norm, previously explicated with recourse to interpretation, is now replaced by an (...) order characterized by calculation. Such paradigm change is identified based on a discussion of works by Gilles Deleuze. In so doing, it contributes to the developing research programme in the cultures of high-frequency trading (Lange et al. 2016) and also points at broader issues for conceptualising financial regulation. (shrink)

Liquidity risk arises from the inability to unwind or hedge trading positions at the prevailing market prices. The risk of liquidity is a wide and complex topic as it depends on several factors and causes. While much has been written on the subject, there exists no clear-cut mathematical description of the phenomena and typical market risk modeling methods fail to identify the effect of illiquidity risk. In this paper, we do not propose a definitive one either, but we attempt (...) to derive novel mathematical algorithms for the dynamic modeling of trading volumes during the closeout period from the perspective of multiple-asset portfolio, as well as for financial entities with different subsidiary firms and multiple agents. The robust modeling techniques are based on the application of initial-value-problem differential equations technique for portfolio selection and risk management purposes. This paper provides some crucial parameters for the assessment of the trading volumes of multiple-asset portfolio during the closeout period, where the mathematical proofs for each theorem and corollary are provided. Based on the new developed econophysics theory, this paper presents for the first time a closed-form solution for key parameters for the estimation of trading volumes and liquidity risk, such as the unwinding constant, half-life, and mean lifetime and discusses how these novel parameters can be estimated and incorporated into the proposed techniques. The developed modeling algorithms are appealing in terms of theory and are promising for practical econophysics applications, particularly in developing dynamic and robust portfolio management algorithms in light of the 2007–2009 global financial crunch. In addition, they can be applied to artificial intelligence and machine learning for the policymaking process, reinforcement machine learning techniques for the Internet of Things data analytics, expert systems in finance, FinTech, and within big data ecosystems. (shrink)

The rapid advancement of algorithmic trading has demonstrated the success of AI automation, as well as gaps in our understanding of the implications of this technology proliferation. We explore ethical issues in the context of autonomous trading agents, both to address problems in this domain and as a case study for regulating autonomous agents more generally. We argue that increasingly competent trading agents will be capable of initiative at wider levels, necessitating clarification of ethical and legal (...) boundaries, and corresponding development of norms and enforcement capability. (shrink)

Inspired by recent breakthroughs in predictive modeling, practitioners in both industry and government have turned to machine learning with hopes of operationalizing predictions to drive automated decisions. Unfortunately, many social desiderata concerning consequential decisions, such as justice or fairness, have no natural formulation within a purely predictive framework. In efforts to mitigate these problems, researchers have proposed a variety of metrics for quantifying deviations from various statistical parities that we might expect to observe in a fair world and offered a (...) variety of algorithms in attempts to satisfy subsets of these parities or to trade o the degree to which they are satised against utility. In this paper, we connect this approach to fair machine learning to the literature on ideal and non-ideal methodological approaches in political philosophy. The ideal approach requires positing the principles according to which a just world would operate. In the most straightforward application of ideal theory, one supports a proposed policy by arguing that it closes a discrepancy between the real and the perfectly just world. However, by failing to account for the mechanisms by which our non-ideal world arose, the responsibilities of various decision-makers, and the impacts of proposed policies, naive applications of ideal thinking can lead to misguided interventions. In this paper, we demonstrate a connection between the fair machine learning literature and the ideal approach in political philosophy, and argue that the increasingly apparent shortcomings of proposed fair machine learning algorithms reflect broader troubles faced by the ideal approach. We conclude with a critical discussion of the harms of misguided solutions, a reinterpretation of impossibility results, and directions for future research. (shrink)

The power trading system has the characteristics of nonlinearity, dynamics, and complexity. Part of the business data in the trading system needs to be exposed to numerous external business systems. The traditional centralized power trading model has some problems, such as low data security and trust crisis of regulators. Blockchain technology provides prominent ideas for solving these problems. Firstly, the improved AdaBoost algorithm is used to predict the supply and demand gap of power trading nodes. Secondly, (...) based on the fact that the information on the blockchain is only open to the power supply side, a two-stage game model of asymmetric information between the power supply side and the power user is constructed to capture the price competition behavior between them in order to find the Nash equilibrium price in two stages. Finally, the US PJM market electricity market data are used to carry out an example analysis to verify the effectiveness of the algorithm and model. (shrink)

Clark & Thornton have demonstrated the paradox between the opacity of the transformations that underlie relational mappings and the ease with which people learn such mappings. However, C&T's trading-spaces proposal resolves the paradox only in the broadest outline. The general-purpose algorithm promised by C&T remains to be developed. The strategy of doing so is to analyze and formulate computational mechanisms for known cases of recoding.

Foundations of Algorithms, Fifth Edition offers a well-balanced presentation of algorithm design, complexity analysis of algorithms, and computational complexity. Ideal for any computer science students with a background in college algebra and discrete structures, the text presents mathematical concepts using standard English and simple notation to maximize accessibility and user-friendliness. Concrete examples, appendices reviewing essential mathematical concepts, and a student-focused approach reinforce theoretical explanations and promote learning and retention. C++ and Java pseudocode help students better understand complex algorithms. A chapter (...) on numerical algorithms includes a review of basic number theory, Euclid's Algorithm for finding the greatest common divisor, a review of modular arithmetic, an algorithm for solving modular linear equations, an algorithm for computing modular powers, and the new polynomial-time algorithm for determining whether a number is prime. The revised and updated Fifth Edition features an all-new chapter on genetic algorithms and genetic programming, including approximate solutions to the traveling salesperson problem, an algorithm for an artificial ant that navigates along a trail of food, and an application to financial trading. With fully updated exercises and examples throughout and improved instructor resources including complete solutions, an Instructor's Manual and PowerPoint lecture outlines, Foundations of Algorithms is an essential text for undergraduate and graduate courses in the design and analysis of algorithms. Key features include: • The only text of its kind with a chapter on genetic algorithms • Use of C++ and Java pseudocode to help students better understand complex algorithms • No calculus background required • Numerous clear and student-friendly examples throughout the text • Fully updated exercises and examples throughout • Improved instructor resources, including complete solutions, an Instructor's Manual, and PowerPoint lecture outlines. (shrink)

Governments and citizens of nearly every nation have been compelled to respond to COVID-19. Many measures have been adopted, including contact tracing and risk assessment algorithms, whereby citizen whereabouts are monitored to trace contact with other infectious individuals in order to generate a risk status via algorithmic evaluation. Based on 38 in-depth interviews, we investigate how people make sense of Health Code, the Chinese contact tracing and risk assessment algorithmic sociotechnical assemblage. We probe how people accept or resist (...) Health Code by examining their ongoing, dynamic, and relational interactions with it. Participants display a rich variety of attitudes toward privacy and surveillance, ranging from fatalism to the possibility of privacy to trade-offs for surveillance in exchange for public health, which is mediated by the perceived effectiveness of Health Code and changing views on the intentions of institutions who deploy it. We show how perceived competency varies not just on how well the technology works, but on the social and cultural enforcement of various non-technical aspects like quarantine, citizen data inputs, and cell reception. Furthermore, we illustrate how perceptions of Health Code are nested in people’s broader interpretations of disease control at the national and global level, and unexpectedly strengthen the Chinese authority’s legitimacy. None of the Chinese public, Health Code, or people’s perceptions toward Health Code are predetermined, fixed, or categorically consistent, but are co-constitutive and dynamic over time. We conclude with a theorization of a relational perception and methodological reflections to study algorithmic sociotechnical assemblages beyond COVID-19. (shrink)

According to a recent survey by the HR Research Institute, as the presence of artificial intelligence (AI) becomes increasingly common in the workplace, HR professionals are worried that the use of recruitment algorithms will lead to a “dehumanization” of the hiring process. Our main goals in this paper are threefold: i) to bring attention to this neglected issue, ii) to clarify what exactly this concern about dehumanization might amount to, and iii) to sketch an argument for why dehumanizing the hiring (...) process is ethically suspect. After distinguishing the use of the term “dehumanization” in this context (i.e., removing the human presence) from its more common meaning in the interdisciplinary field of dehumanization studies (i.e., conceiving of other humans as subhuman), we argue that the use of hiring algorithms may negatively impact the employee-employer relationship. We argue that there are good independent reasons to accept a substantive employee-employer relationship, as well as an applicant-employer relationship, both of which are consistent with a stakeholder theory of corporate obligations. We further argue that dehumanizing the hiring process may negatively impact these relationships because of the difference between the values of human recruiters and the values embedded in recruitment algorithms. Drawing on Nguyen’s (2021) critique of how Twitter “gamifies communication”, we argue that replacing human recruiters with algorithms imports artificial values into the hiring process. We close by briefly considering some ways to potentially mitigate the problems posed by recruitment algorithms, along with the possibility that some difficult trade-offs will need to be made. (shrink)

The article examines two digital phenomena linked to money and finance, which are the bitcoin and high-frequency trading, through the lens of Vilém Flusser’s concept of technical image. Flusser’s theory highlights three aspects of technical images: they are engendered by the act of organizing particles, are produced by people who operate devices through keys, and are mediated by code, which is linear and pertains to the era of written text, which Flusser conflates with the notion of history. In this (...) article, the argument focuses on the aspects of high-frequency trading and the blockchain that relate to the idea of images. The sections investigate how money has historically operated in the logic of Flusser’s description of symbol manipulation in relation to consciousness and temporality: from the circular time of magical consciousness to the linear time of historical consciousness, and more recently, to the non-dimensional time of post-historical consciousness. The article argues that there are two levels, or layers, in the images involved in cryptocurrencies or algorithmic trading, but only one of them is visible to the human operator or viewer, in the form of a price. The other, which indeed represents the operations in the system, is only accessible to the machines themselves. The questions raised concern the role of these recent digital financial and monetary forms in the development of a reality dominated by apparatuses and technical images. (shrink)

Certain strands of contemporary media theory are concerned with the ways in which computational environments exploit the ‘missing half-second’ of human perception and thereby influence, control or exploit humans at an affective level. The ‘technological unconscious’ of our times is often understood to work at this affective level, and high-frequency trading is regularly provided as a primary illustrative example of the contagious dynamics it produces. We challenge and complicate this account of the relation between consciousness, affect and media technologies (...) by drawing on the recent work of N. Katherine Hayles and by focusing in detail on the ways in which the ‘costs of consciousness’ are accounted for and negotiated in high-frequency trading. We suggest that traders actively develop modes of awareness accounting for the costs of consciousness, and that the necessary ‘stupidity’ of high-frequency trading algorithms as well as competition pose limits to the full automation of financial markets. (shrink)

In recent years, a plethora of high-profile scientific publications has been reporting about machine learning algorithms outperforming clinicians in medical diagnosis or treatment recommendations. This has spiked interest in deploying relevant algorithms with the aim of enhancing decision-making in healthcare. In this paper, we argue that instead of straightforwardly enhancing the decision-making capabilities of clinicians and healthcare institutions, deploying machines learning algorithms entails trade-offs at the epistemic and the normative level. Whereas involving machine learning might improve the accuracy of medical (...) diagnosis, it comes at the expense of opacity when trying to assess the reliability of given diagnosis. Drawing on literature in social epistemology and moral responsibility, we argue that the uncertainty in question potentially undermines the epistemic authority of clinicians. Furthermore, we elucidate potential pitfalls of involving machine learning in healthcare with respect to paternalism, moral responsibility and fairness. At last, we discuss how the deployment of machine learning algorithms might shift the evidentiary norms of medical diagnosis. In this regard, we hope to lay the grounds for further ethical reflection of the opportunities and pitfalls of machine learning for enhancing decision-making in healthcare. (shrink)

‘@AP: Breaking: Two Explosions in the White House and Barack Obama is injured’. So read a tweet sent from a hacked Associated Press Twitter account @AP, which affected financial markets, wiping out $136.5 billion of the Standard & Poor’s 500 Index’s value. While the speed of the Associated Press hack crash event and the proprietary nature of the algorithms involved make it difficult to make causal claims about the relationship between social media and trading algorithms, we argue that it (...) helps us to critically examine the volatile connections between social media, financial markets, and third parties offering human and algorithmic analysis. By analyzing the commentaries of this event, we highlight two particular currents: one formed by computational processes that mine and analyze Twitter data, and the other being financial algorithms that make automated trades and steer the stock market. We build on sociology of finance together with media theory and focus on the work of Christian Marazzi, Gabriel Tarde and Tony Sampson to analyze the relationship between social media and financial markets. We argue that Twitter and social media are becoming more powerful forces, not just because they connect people or generate new modes of participation, but because they are connecting human communicative spaces to automated computational spaces in ways that are affectively contagious and highly volatile. (shrink)

The enormous scale of the available information and products on the Internet has necessitated the development of algorithms that intermediate between options and human users. These algorithms attempt to provide the user with relevant information. In doing so, the algorithms may incur potential negative consequences stemming from the need to select items about which it is uncertain to obtain information about users versus the need to select items about which it is certain to secure high ratings. This tension is an (...) instance of the exploration–exploitation trade‐off in the context of recommender systems. Because humans are in this interaction loop, the long‐term trade‐off behavior depends on human variability. Our goal is to characterize the trade‐off behavior as a function of human variability fundamental to such human–algorithm interaction. To tackle the characterization, we first introduce a unifying model that smoothly transitions between active learning and recommending relevant information. The unifying model gives us access to a continuum of algorithms along the exploration–exploitation trade‐off. We then present two experiments to measure the trade‐off behavior under two very different levels of human variability. The experimental results inform a thorough simulation study in which we modeled and varied human variability systematically over a wide rage. The main result is that exploration–exploitation trade‐off grows in severity as human variability increases, but there exists a regime of low variability where algorithms balanced in exploration and exploitation can largely overcome the trade‐off. (shrink)

The literature on algorithmic fairness has examined exogenous sources of biases such as shortcomings in the data and structural injustices in society. It has also examined internal sources of bias as evidenced by a number of impossibility theorems showing that no algorithm can concurrently satisfy multiple criteria of fairness. This paper contributes to the literature stemming from the impossibility theorems by examining how informational richness affects the accuracy and fairness of predictive algorithms. With the aid of a computer simulation, (...) we show that informational richness is the engine that drives improvements in the performance of a predictive algorithm, in terms of both accuracy and fairness. The centrality of informational richness suggests that classification parity, a popular criterion of algorithmic fairness, should be given relatively little weight. But we caution that the centrality of informational richness should be taken with a grain of salt in light of practical limitations, in particular, the so-called bias-variance trade off. (shrink)

This paper examines two prominent formal trade-offs in artificial intelligence (AI)---between predictive accuracy and fairness, and between predictive accuracy and interpretability. These trade-offs have become a central focus in normative and regulatory discussions as policymakers seek to understand the value tensions that can arise in the social adoption of AI tools. The prevailing interpretation views these formal trade-offs as directly corresponding to tensions between underlying social values, implying unavoidable conflicts between those social objectives. In this paper, I challenge that prevalent (...) interpretation by introducing a sociotechnical approach to examining the value implications of trade-offs. After providing a set-theoretic understanding of these model trade-offs, I argue that their implications cannot be understood in isolation from the technical, psychological, organizational, and social factors that shape the integration of AI models into decision-making pipelines. Specifically, I identify three key considerations---validity and instrumental relevance, compositionality, and dynamics---for contextualizing and characterizing these implications. These considerations reveal that the relationship between model trade-offs and corresponding values depends on critical choices and assumptions. Crucially, judicious sacrifices in one model property for another can, in fact, promote both sets of corresponding values. The proposed sociotechnical perspective thus shows that we can and should aspire to higher epistemic and ethical possibilities than the prevalent interpretation suggests, while offering practical guidance for achieving those outcomes. Finally, I draw out the broader implications of this perspective for AI design and governance, highlighting the need to broaden normative engagement across the AI lifecycle, develop legal and auditing tools sensitive to sociotechnical considerations, and rethink the vital role and appropriate structure of interdisciplinary collaboration in fostering a responsible AI workforce. (shrink)

In light of the structural role of computational technology in the expansion of modern global finance, this essay investigates the ontology of contemporary markets starting from a reformulation of liquidity—one of the tenets of financial trading. Focusing on the nexus between financial and algorithmic flows, the paper complements contemporary philosophies of the market with insights into recent theories of computation, emphasizing the functional role of contingency, both for market trading and algorithmic processes. Considering the increasing adoption (...) of advanced computational methods in automated trading strategies, this article argues that the event of price is the direct manifestation of the incomputability at the heart of market exchange. In doing so, it questions the ontological assumptions of “flow” and “fluidity” underpinning traditional conceptions of liquidity and challenges the notion of rationality in market behavior. Ultimately, the paper gestures toward some of the social and political consequences of this reformulation. (shrink)

In nature, gregarious animals, insects, or bacteria usually exhibit paradoxical behaviors in the form of group fission and fusion, which exerts an important influence on group’s pattern formation, information transfer, and epidemiology. However, the fission-fusion dynamics have received little attention compared to other flocking behavior. In this paper, an intermittent selective interaction based control algorithm for the self-organized fission-fusion behavior of flocking system is proposed, which bridges the gap between the two conflicting behaviors in a unified fashion. Specifically, a hybrid (...) velocity coordination strategy that includes both the egalitarian and selective interactions is proposed, where the egalitarian interaction is to maintain the flock’s order and achieve the fusion behavior while the selective interaction strategy is for the response to external stimulus information and generates the fission behavior. Numerical simulations demonstrate that the proposed control algorithm can realize the self-organized fission-fusion behavior of flocking system under a unified framework. The influences of the main control parameters on the performance of the fission-fusion behavior are also discussed. In particular, the trade-off parameterαbalances the exploration and exploitation behaviors of flocking system and significantly enhances its movement flexibility and environmental adaptivity. (shrink)

Contemporary financial markets have recently witnessed a sea change with the ‘algorithmic revolution’, as trading automats are used to ease the execution sequences and reduce market impact. Being constantly monitored, they take an active part in the shaping of markets, and sometimes generate crises when ‘they mess up’ or when they entail situations where traders cannot go backwards. Algorithms are software codes coding practices in an IT significant ‘textual’ device, designed to replicate trading patterns. To be accepted, (...) however, they need to comply with regulatory texts, which are nothing else but codes of conduct coding accepted practices in the markets. In this article, I draw on ethnographic fieldwork in order to open these black boxes, while trying to describe their existence as devices encapsulating several points of views. I address the question of a possible misalignment between those visions, and more specifically try to draw the consequences raised by such discrepancies as regards the future of financial regulation. (shrink)

This special issue introduces the study of financial technologies and finance to the field of philosophy of technology, bringing together two different fields that have not traditionally been in dialogue. The included articles are: Digital Art as ‘Monetised Graphics’: Enforcing Intellectual Property on the Blockchain, by Martin Zeilinger; Fundamentals of Algorithmic Markets: Liquidity, Contingency, and the Incomputability of Exchange, by Laura Lotti; ‘Crises of Modernity’ Discourses and the Rise of Financial Technologies in a Contested Mechanized World, by Marinus Ossewaarde; (...) Two Technical Images: Blockchain and High-Frequency Trading, by Diego Viana; and The Blockchain as a Narrative Technology: Investigating the Social Ontology and Normative Configurations of Cryptocurrencies, by Wessel Reijers and Mark Coeckelbergh. (shrink)

continent. 1.4 (2011): 239. In 1998 the Securities and Exchange Commission authorized electronic exchanges. Not only did this give day traders access to buy and sell securities from their desktops, it also made it possible for high powered Wall Street traders to program algorithms to make trades at speeds on the order of milliseconds.(1) The advent of automatic algorithmic trading, now known as high-frequency trading, has vastly accelerated the already increasing speed and volume of trading. This (...) project was an attempt to capture in an ephemeral and environmental way the dizzying acceleration of networked markets. The change in the daily closing value of the Dow and daily trading volume were each set to control a sine wave oscillator between 20 and 800hz. The oscillators were combined to ‘sonify’ the data, creating a changing noise reflecting the daily change and trading volume of the market. The resulting sound is somewhat tedious until the trading volume begins increasing in the 60s and then becomes increasingly complex as high-frequency trading and market volatility set in during the 1990s and 2000s. NOTES 1. Charles Duhigg. “Stock Traders Find Speed Pays, in Milliseconds.” New York Times . July 23, 2009. (shrink)

The paper demonstrates that social alignment is distinct from value alignment as it is currently understood in the AI safety literature, and argues that social alignment is an important research agenda. Work provides an important example for the argument, since work is a cooperative endeavor, and it is part of the larger manifold of social cooperation. These cooperative aspects of work are individually and socially valuable, and so they must be given a central place when evaluating the impact of AI (...) upon work. Workplace technologies are not simply instruments for achieving productive goals, but ways of mediating interpersonal relations. They are aspects of a cooperative interface i.e. the infrastructure by which we engage cooperative behavior with others. The concept of the cooperative interface suggests two conjectures to foreground in the social alignment agenda, motivated by the experience of algorithmic trading and social robotics: that AI impacts cooperation through its effects on social networks, and through its effects on social norms. (shrink)

Social media commentary, satellite imagery and GPS data are a part of ‘alternative data’, that is, data that originate outside of the standard repertoire of market data but are considered useful for predicting stock prices, detecting different risk exposures and discovering new price movement indicators. With the availability of sophisticated machine-learning analytics tools, alternative data are gaining traction within the investment management and algorithmic trading industries. Drawing on interviews with people working in investment management and algorithmic (...) class='Hi'>trading firms utilizing alternative data, as well as firms providing and sourcing such data, we emphasize social media-based sentiment analytics as one manifestation of how alternative data are deployed for stock price prediction purposes. This demonstrates both how sentiment analytics are developed and subsequently utilized by investment management firms. We argue that ‘alternative data’ are an open-ended placeholder for every data source potentially relevant for investment management purposes and harnessing these disparate data sources requires certain standardization efforts by different market participants. Besides showing how market participants understand and use alternative data, we demonstrate that alternative data often undergo processes of prospecting and assetization. We further contend that the widespread embracement of alternative data in investment management and trading encourages a financialization process at the data level which raises new governance issues. (shrink)

This paper investigates if and to what point it is possible to trade on news sentiment and if deep learning, given the current hype on the topic, would be a good tool to do so. DL is built explicitly for dealing with significant amounts of data and performing complex tasks where automatic learning is a necessity. Thanks to its promise to detect complex patterns in a dataset, it may be appealing to those investors that are looking to improve their (...) class='Hi'>trading process. Moreover, DL and specifically LSTM seem a good pick from a linguistic perspective too, given its ability to “remember” previous words in a sentence. After having explained how DL models are built, we will use this tool for forecasting the market sentiment using news headlines. The prediction is based on the Dow Jones industrial average by analyzing 25 daily news headlines available between 2008 and 2016, which will then be extended up to 2020. The result will be the indicator used for developing an algorithmic trading strategy. The analysis will be performed on two specific cases that will be pursued over five time-steps and the testing will be developed in real-world scenarios. (shrink)

Both fairness and efficiency are important considerations in market design and regulation, yet many regulators have neither defined nor measured these concepts. We develop an evidencebased policy framework in which these are both defined and measured using a series of empirical proxies. We then build a systems estimation model to examine the 2003–2011 explosive growth in algorithmic trading on the London Stock Exchange and NYSE Euronext Paris. Our results show that greater AT is associated with increased transactional efficiency (...) and reduced information leakage in top quintile stocks. For less liquid stocks, manipulation at the close declines. We also document the tradeoff between reduced spreads and increased manipulation or information leakage following the introduction of MiFID1. (shrink)

Genetic Algorithms(GAs) are effective approximation algorithms which focus on “hopeful area” in the searching process. However, in harder problems, it is often very difficult to maintain a favorable trade-off between exploitation and exploration. All individuals leave the big-valley including the global optimum, and concentrate on another big-valley including a local optimum often. In this paper, we define such a situation on conventional GAs as the “UV-phenomenon”, and suggest UV-structures as hard landscape structures that will cause the UV-phenomenon. We introduce a (...) test function which has explicit UV-structures, and show UV-phenomenon caused by them. Next we analyze Fletcher and Powell function to confirm our hypothesis. Finally we propose a novel framework of GAs which can cope with UV-structures. (shrink)

We propose a formal framework for interpretable machine learning. Combining elements from statistical learning, causal interventionism, and decision theory, we design an idealised explanation game in which players collaborate to find the best explanation for a given algorithmic prediction. Through an iterative procedure of questions and answers, the players establish a three-dimensional Pareto frontier that describes the optimal trade-offs between explanatory accuracy, simplicity, and relevance. Multiple rounds are played at different levels of abstraction, allowing the players to explore overlapping (...) causal patterns of variable granularity and scope. We characterise the conditions under which such a game is almost surely guaranteed to converge on a optimal explanation surface in polynomial time, and highlight obstacles that will tend to prevent the players from advancing beyond certain explanatory thresholds. The game serves a descriptive and a normative function, establishing a conceptual space in which to analyse and compare existing proposals, as well as design new and improved solutions. (shrink)

Online trading invariably involves dealings between strangers, so it is important for one party to be able to judge objectively the trustworthiness of the other. In such a setting, the decision to trust a user may sensibly be based on that user’s past behaviour. We introduce a specification language based on linear temporal logic for expressing a policy for categorising the behaviour patterns of a user depending on its transaction history. We also present an algorithm for checking whether the (...) transaction history obeys the stated policy. To be useful in a real setting, such a language should allow one to express realistic policies which may involve parameter quantification and quantitative or statistical patterns. We introduce several extensions of linear temporal logic to cater for such needs: a restricted form of universal and existential quantification; arbitrary computable functions and relations in the term language; and a “counting” quantifier for counting how many times a formula holds in the past. We then show that model checking a transaction history against a policy, which we call the history-based transaction monitoring problem, is PSPACE-complete in the size of the policy formula and the length of the history, assuming that the underlying interpreted func-. (shrink)

ABSTRACT: One important criticism of algorithmic systems is that they lack transparency. Such systems can be opaque because they are complex, protected by patent or trade secret, or deliberately obscure. In the EU, there is a debate about whether the General Data Protection Regulation (GDPR) contains a “right to explanation,” and if so what such a right entails. Our task in this chapter is to address this informational component of algorithmic systems. We argue that information access is integral (...) for respecting autonomy, and transparency policies should be tailored to advance autonomy. -/- To make this argument we distinguish two facets of agency (i.e., capacity to act). The first is practical agency, or the ability to act effectively according to one’s values. The second is what we call cognitive agency, which is the ability to exercise what Pamela Hieronymi calls “evaluative control” (i.e., the ability to control our affective states, such as beliefs, desires, and attitudes). We argue that respecting autonomy requires providing persons sufficient information to exercise evaluative control and properly interpret the world and one’s place in it. We draw this distinction out by considering algorithmic systems used in background checks, and we apply the view to key cases involving risk assessment in criminal justice decisions and K-12 teacher evaluation. -/- The link below is to an open access version of the chapter. (shrink)

Evolutionary multi-objective optimization algorithms have been successfully applied in many real-life problems. EMOAs approximate the set of trade-offs between multiple conflicting objectives, known as the Pareto optimal set. Reference point approaches can alleviate the optimization process by highlighting relevant areas of the Pareto set and support the decision makers to take the more confident evaluation. One important drawback of this approaches is that they require an in-depth knowledge of the problem being solved in order to function correctly. Collective intelligence has (...) been put forward as an alternative to deal with situations like these. This paper extends some well-known EMOAs to incorporate collective preferences and interactive techniques. Similarly, two new preference-based multi-objective optimization performance indicators are introduced in order to analyze the results produced by the proposed algorithms in the comparative experiments carried out. (shrink)

The use of algorithms and automated systems, especially those leveraging artificial intelligence (AI), has been exploding in the public sector, but their use has been controversial. Ethicists, public advocates, and legal scholars have debated whether biases in AI systems should bar their use or if the potential net benefits, especially toward traditionally disadvantaged groups, justify even greater expansion. While this debate has become voluminous, no scholars of which we are aware have conducted experiments with the groups affected by these policies (...) about how they view the trade-offs. We conduct a set of two conjoint experiments with a high-quality sample of 973 Americans who identify as Black or African American in which we randomize the levels of inter-group disparity in outcomes and the net effect on such adverse outcomes in two highly controversial contexts: pre-trial detention and traffic camera ticketing. The results suggest that respondents are willing to tolerate some level of disparity in outcomes in exchange for certain net improvements for their community. These results turn this debate from an abstract ethical argument into an evaluation of political feasibility and policy design based on empirics. (shrink)

This article presents the first, systematic analysis of the ethical challenges posed by recommender systems through a literature review. The article identifies six areas of concern, and maps them onto a proposed taxonomy of different kinds of ethical impact. The analysis uncovers a gap in the literature: currently user-centred approaches do not consider the interests of a variety of other stakeholders—as opposed to just the receivers of a recommendation—in assessing the ethical impacts of a recommender system.

An Artificial Intelligence algorithm trained on data that reflect racial biases may yield racially biased outputs, even if the algorithm on its own is unbiased. For example, algorithms used to schedule medical appointments in the USA predict that Black patients are at a higher risk of no-show than non-Black patients, though technically accurate given existing data that prediction results in Black patients being overwhelmingly scheduled in appointment slots that cause longer wait times than non-Black patients. This perpetuates racial inequity, in (...) this case lesser access to medical care. This gives rise to one type of Accuracy-Fairness trade-off: preserve the efficiency offered by using AI to schedule appointments or discard that efficiency in order to avoid perpetuating ethno-racial disparities. Similar trade-offs arise in a range of AI applications including others in medicine, as well as in education, judicial systems, and public security, among others. This article presents a framework for addressing such trade-offs where Machine Learning and Optimization components of the algorithm are decoupled. Applied to medical appointment scheduling, our framework articulates four approaches intervening in different ways on different components of the algorithm. Each yields specific results, in one case preserving accuracy comparable to the current state-of-the-art while eliminating the disparity. (shrink)

The notion of program verification appears to trade upon an equivocation. Algorithms, as logical structures, are appropriate subjects for deductive verification. Programs, as causal models of those structures, are not. The success of program verification as a generally applicable and completely reliable method for guaranteeing program performance is not even a theoretical possibility.

Science is popularly understood as being an ideal of impartial algorithmic objectivity that provides us with a realistic description of the world down to the last detail. The essays collected in this book—written by some of the leading experts in the field—challenge this popular image right at its heart, taking as their starting point that science trades not only in truth, but in fiction, too. With case studies that range from physics to economics and to biology, _Fictions in Science_ (...) reveals that fictions are as ubiquitous in scientific narratives and practice as they are in any other human endeavor, including literature and art. Of course scientific activity, most prominently in the formal sciences, employs logically precise algorithmic thinking. However, the key to the predictive and technological success of the empirical sciences might well lie elsewhere—perhaps even in scientists’ extraordinary creative imagination instead. As these essays demonstrate, within the bounds of what is empirically possible, a scientist’s capacity for invention and creative thinking matches that of any writer or artist. (shrink)

This article describes a left-corner parser implemented within a cognitively and neurologically motivated distributed model of memory. This parser's approach to syntactic ambiguity points toward a tidy account both of surprisal effects and of locality effects, such as the parsing breakdowns caused by center embedding. The model provides an algorithmic-level account of these breakdowns: The structure of the parser's memory and the nature of incremental parsing produce a smooth degradation of processing accuracy for longer center embeddings, and a steeper (...) degradation when they are nested, in line with recall observations by Miller and Isard and speed-accuracy trade-off observations by McElree et al.. Modeling results show that this effect is distinct from the effects of ambiguity and exceeds the effect of mere sentence length. (shrink)

This Article takes the perspective of law and philosophy, integrating insights from computer science. First, I will argue that in the era of big data analytics we need an understanding of privacy that is capable of protecting what is uncountable, incalculable or incomputable about individual persons. To instigate this new dimension of the right to privacy, I expand previous work on the relational nature of privacy, and the productive indeterminacy of human identity it implies, into an ecological understanding of privacy, (...) taking into account the technological environment that mediates the constitution of human identity. Second, I will investigate how machine learning actually works, detecting a series of design choices that inform the accuracy of the outcome, each entailing trade-offs that determine the relevance, validity and reliability of the algorithm’s accuracy for real life problems. I argue that incomputability does not call for a rejection of machine learning per se but calls for a research design that enables those who will be affected by the algorithms to become involved and to learn how machines learn — resulting in a better understanding of their potential and limitations. A better understanding of the limitations that are inherent in machine learning will deflate some of the eschatological expectations, and provide for better decision-making about whether and if so how to implement machine learning in specific domains or contexts. I will highlight how a reliable research design aligns with purpose limitation as core to its methodological integrity. This Article, then, advocates a practice of “agonistic machine learning” that will contribute to responsible decisions about the integration of data-driven applications into our environments while simultaneously bringing them under the Rule of Law. This should also provide the best means to achieve effective protection against overdetermination of individuals by machine inferences. (shrink)

The problem of fair machine learning has drawn much attention over the last few years and the bulk of offered solutions are, in principle, empirical. However, algorithmic fairness also raises important conceptual issues that would fail to be addressed if one relies entirely on empirical considerations. Herein, I will argue that the current debate has developed an empirical framework that has brought important contributions to the development of algorithmic decision-making, such as new techniques to discover and prevent discrimination, (...) additional assessment criteria, and analyses of the interaction between fairness and predictive accuracy. However, the same framework has also suggested higher-order issues regarding the translation of fairness into metrics and quantifiable trade-offs. Although the tools which have been developed so far are essential to address discrimination encoded in data and algorithms, their integration into society elicits key questions such as: What kind of assumptions and decisions underlies the empirical framework? How do the results of the empirical approach penetrate public debate? What kind of reflection and deliberation should stakeholders have over available fairness metrics? I will outline the empirical approach to fair machine learning, i.e. how the problem is framed and addressed, and suggest that there are important non-empirical issues that should be tackled. While this work will focus on the problem of algorithmic fairness, the lesson can extend to other conceptual problems in the analysis of algorithmic decision-making such as privacy and explainability. (shrink)

An increasingly prevalent approach to studying human cognition is to construe the mind as optimally allocating limited cognitive resources among cognitive processes. Under this bounded rationality approach (Icard in Philos Sci 85(1):79–101, 2018; Simon in Utility and probability, Palgrave Macmillan, 1980), it is common to assume that resource-bounded cognitive agents approximate normative solutions to statistical inference problems, and that much of the bias and variability in human performance can be explained in terms of the approximation strategies we employ. In this (...) paper, we argue that this approach restricts itself to an unnecessarily narrow scope of cognitive models, which limits its ability to explain how humans flexibly adapt their representations to novel environments. We argue that more attention should be paid to how we form our cognitive representations in the first place, and advocate for pluralistic framework which jointly optimizes over both representations and algorithms for manipulating them. We identify several fundamental trade-offs that manifest in this joint optimization, and draw on recent work to motivate a unified formal framework for this analysis. We illustrate a simplified version of this analysis with a case study in social cognition, and outline several new directions for research that this approach suggests. (shrink)