ABSTRACT:
The right to free speech underpins democracy, furthers the development of ideas into progress, makes human relationships flourish, and therefore promotes societal well-being. Free speech is enshrined in constitutions, invoked in the amphitheaters of parliaments, and hallowed in the halls of the academe. It is also not infrequently honored in the breach. A variety of sanctions limit free speech. These include legal, normative, social, and organizational limitations. We will perhaps agree that some of these are unavoidable if the society and its institutions are to function: Free speech, in its egregious variations, may conflict with other human rights. However, like other fundamental values, the right to free speech requires constant and informed cultivation and protection; it is not a static phenomenon, it is in a dynamic development. What are the proper boundaries? And why should we in the information systems (IS) research community concern ourself with researching the subject?
In the recent decades, our very understanding of the benefits and drawbacks of absolutely free speech has been challenged in its essence by two developments that fall squarely into our research and practice domains. Social media (SM), a great boon to society at large and to each of us, have brought the first major challenge. It has been commonly recognized that owing to the massive and active access, the expressions on SM by bad actors can metastasize and erupt into such phenomena as societal instability, victimizing segments of society, or just bullying schoolchildren. Hate speech and fake news are the usual designations for two of the most wide-spread categories of the abuse of SM. Societal polarization and crowd-induced vilification of the “other” at whatever level are on everyday display, with SM to blame for the reinforcement.
The next challenge is just coming in. With the advent of the generative artificial intelligence (AI), machines acquire the ability to manipulate and deceive humans. Whether we consider these actions of machines speech, and whether we know how to allocate agency for the machines’ actions (or speech acts), danger looms. Without our (humans’) ability to align the actions of AI systems with the values, or simply intentions, of the minders of these systems, we can lose control, and any conversation about free speech can become moot. Research on agency assignment and alignment actualization is of the crucial moment. The developments are taking place in the context of geopolitical shifts and no-holds-barred competition that may lead to irreversible outcomes. As a scholarly field we need to dedicate ourselves to pursuing this existentially important issue.
The three papers opening this issue of JMIS pursue the detection of hate speech, and the reduction of the spread of these attacks and of fake news. The first, by Seong-Su Kim, Seongbeom Kim, and Hee-Woong Kim, presents a novel deep-learning model aiming to detect hate speech on the online news platforms. The comments sections of news platforms have become loci of hate speech, debasing or perverting the value of the platforms. The model of Kim et al. is theoretically grounded, and the paper offers both an artifact and a theoretical advancement. Since the approach is set within the specific context of comments on news platforms, the authors are able to use agenda-setting theory to capture the deleterious user comments as related to the news being commented on. The researchers demonstrate empirically the superiority of their model. The paper illustrates well what we can do to help return free speech to its rightful place in the human intercourse.
The rapid spread of fake news on social media is another major ill affecting them—and us. Are corrective messages effective? Kelvin K. King and Diego Escobari present their theoretically and empirically based investigation in the context of a very large number of tweets. The authors are able to dichotomize the corrections into the effective ones and the actually counter-effective corrections. They also present nuanced findings regarding the potency of the messages that affects the herding and echo-chamber behaviors of platform users. Platform sponsors have much to learn about freeing the speech from abusers and their followers.
Social media firestorms are massive barrages of messages containing negative opinions about their subject, be it a person, a group, or a brand. In their virality, they are highly corrosive, with deleterious effects not only to their subject, but—in the limit—to the societal trust. As we know, trust is a foundational value of a flourishing society and its diminishment leads to the erosion of not only economic well-being but overall political and social fabric. Here, Tommy K.H. Chan, Zach W.Y. Lee, Meizhi Pan, and Ke Sun focus on the antecedents, conduct, and outcomes of brand-oriented firestorms capable of damaging the reputation and lasting image of the affected brand. They examine the ideologically based form of firestorms, rooted in sociopolitical values of the participants in a clash concerning the brand. The research is couched within the social learning theory and meaningfully contributes to it.
Amrit Tiwana and Stephen K. Kim empirically study the boundary between a platform and the apps that complement its functionality. More specifically, they investigate boundary morphing: the addition of the proprietary code to the apps versus the increase of the platform’s own functionality. This morphing contributes over time to the competitive gain or loss of the platform as well as to the competitive standing of the apps that expand the platform’s functionality. Relying on the longitudinal data from a major platform and the lens of the transaction-cost economics, the authors determine the factors of the boundary contraction and expansion, respectively. The results have high relevance to the understanding of our platform-based organization of IS services and to the strategic steering of its architectural components. While the AndroidOS ecosystem studied here is safe in the marketplace, the lessons learned are applicable to the service pricing and to the competitive posture of platforms in general.
Online platforms are the subject of the next paper as well. Mareike Möhlmann, Robert Gregory, and Ola Henfridsson also study the boundary between the platform and its stakeholders. In this case, the investigation concerns content platforms and three key categories of the stakeholders: creators, consumers, and advertisers. The platform, YouTube in the researchers’ study, contains algorithmic rules for balancing the concerns of the parties at interest. Human intervention does not scale, and algorithmic control is necessary (as well as consistent and modifiable). In the interpretive paper, the authors use grounded theory to develop a model of platform governance to address free speech (yes), information diversity, and content safety. The authors’ perspective is that of the negotiation among the stakeholders regarding the rules of engagement on the platform, with the lead role played by the platform sponsors. This governance regime decentralizes to a degree the governance of the platform, lending it a greater stability and satisfaction of the stakeholders, while also displaying the disadvantages surfaced in the paper.
Gamification has become a popular means to evoke user behavior helpful toward the desired outcomes. Here, Dong-Heon (Austin) Kwak, Asher John, Jose Benitez, Soomin Park, Yu (Audrey) Zhao, and Merrill Warkentin study the gamification of online learning. In this mode, the learners self-regulate in the environment of (frequent) self-isolation and a multitude of distractions. The gamification components may include leaderboards or badges and other types of more or less symbolic rewards that aim to increase engagement. Gamification aims to deliver psychological recovery experiences (PRE) of achievement, mastery and control, that restore the learners cognitive and emotional resources after periods of mental effort. The authors’ empirics and analysis show up the role of PRE in the relationship between gamification affordances, and the mastery and control experienced by learners. With the highly dynamic jobs trajectories that can be foreseen in the face of AI progress, online learning may be expected to experience a dramatic gain in importance, and so will its gamification.
Cyberattacks, as we know well, are on a dramatic rise in number, strength, and the diversity of sources. In the next paper, Jong Seok Lee investigates—just as his paper’s title says—the risk of cyberattacks arising from strategic alliances with Big Tech companies. Considering that such alliances are common, the understanding of these risks and of the potential countermeasures is of great interest to the individual firms and to the economy at large as cascading effects are a plausible possibility. The author finds the system-of-systems approach fruitful in both integrating the alliance’s IS and in studying the integration. The intensity of intangible assets is identified as a key factor in extent of the risk incurred. The results are helpful to the executives considering an alliance, structuring it with cybersecurity in view, insuring some of the risks determined here, and jointly monitoring the operation.
Two subsequent papers contribute significantly to our understanding of the effects of online expressions on reputations, trust, and the consequences of these effects in the job market. In the context of crowd-work platforms, Florian Schneider, Pavel Dykmann, and Timm Teubner study the consequences of the inconsistency in online reputational signals on the platforms’ users. In an eye-tracking experiment, they find increased cognitive load and trust reduction owing to such inconsistencies. Furthermore, the researchers establish the differences between the effects of the signals received from the primary platform and those imported from others. The lack of reputation portability or any other integration with the primary platform serves poorly the crowd workers. In the second of the papers, Sherae L. Daniel, Renzhi (Fred) Zhao, and Likoebe M. Maruping address a closely related issue, that of the promotion of open source software (OSS) developers as influenced by the digital traces. As we know, open source code is a now a part of organizational software infrastructure, and, in general, many large organizations are involved in opening some of their code or using—under a variety of constraints and with a variety of modifications—the OSS developed by others. The authors find that the revelation on the platforms of the technical skills of the developers has a different effect from the revelation of interpersonal skills. They also empirically establish the effects of the contest (meritocratic) mechanism versus the sponsored (relationship-based) one operating in the advancement within the OSS communities.
Across the world, the vaunted access to the more prominent physicians is a limiting resource in healthcare. There are only so many highly-reputed physicians, and their practices are concentrated in big cities. IS can help, and the authors of the concluding paper, Sijia Zhou, Yanzhen Chen, and Xin LI, map out one way how this can be done. The authors research the possibility of offering non-diagnostic services on online healthcare platforms (OHP). Such services include, for example, medicine prescribing based on online consultations, which can meet the simpler patient conditions. Yes, you will deal with a more junior physician in a virtual consultation, but you will not wait three months for an encounter with an overburdened authority, and a local physician can be of a greater help to you in the future. The research questions the authors ask have not been asked before. They concern the reallocation and rebalancing of demand for physician services, and they are asked in the context of the organic rebalancing by patient actions rather than the promulgation of new regulations. With a large database on a major OHP and the use of econometrics, the authors answer the questions they ask. This work can lead to increased societal welfare and offers a pragmatic solution to a healthcare problem that otherwise can only get worse. Notably, the solution can be applied, mutatis mutandis, in other contexts of demand imbalance.