Human beings regularly work together to get things done. In particular, people frequently collaborate on the production and dissemination of knowledge. For example, scientists often work together in teams to make new discoveries. How such collaborations produce knowledge, and how well they produce knowledge, are important questions for epistemology. In fact, several epistemologists (e.g., Hardwig 1991, Thagard 1997, Wray 2002) have addressed such questions regarding collaborative scientific research.

One of the most salient features of forming beliefs in a social context is that people end up disagreeing with one another. This is not just an obvious fact about belief-formation; it raises interesting normative questions, especially when people become aware of the opinions of others. How should my beliefs be affected by the knowledge that others hold contrary beliefs? In some cases, the answer seems easy. If I have reason to think that my friend is much better informed than I am, her dissent will often require substantial revision in my belief. If I have reason to think she's mentally deranged, her dissent may require no revision at all. But other cases are more difficult. For example, how confident should I be about my views in epistemology, knowing that they are denied by philosophers at least as intelligent, sane, knowledgeable, diligent and honest as I am?

Over recent decades, computer simulations have become a common tool among practitioners of the social sciences. They have been utilized to study such diverse phenomena as the integration and segregation of different racial groups, the emergence and evolution of friendship networks, the spread of gossip, fluctuations of housing prices in an area, the transmission of social norms, and many more. Philosophers of science and others interested in the methodological status of these studies have identified a number of distinctive virtues of the use of computer simulations. For instance, it has been generally appreciated that as simulations require the formulation of an explicit algorithm, they foster precision and clarity about whatever conceptual issues are involved in the study. The value of computer simulations as a heuristic tool for developing hypotheses, models, and theories has also been recognized, as has been the fact that they can serve as a substitute for real experiments. This is especially useful in the social domain, given that human beings cannot be freely manipulated at the discretion of the experimenter (for both points, see Hartmann 1996). However, the main virtue of computer simulations is generally believed to be that they are able to deal with the complexities that arise when many elements interact in a highly dynamic system and which often evade an exact formal analysis (see, e.g., Humphreys 1991).

In “Group Testimony” (2007) I argued that the testimony of a group cannot be understood (or at least cannot always be understood) in a summative fashion; as the testimony of some or all of the group members. In some cases, it is the group itself that testifies. I also argued that one could extend standard reductionist accounts of the justification of testimonial belief to the case of testimonial belief formed on the basis of group testimony. In this paper, I explore the issue of group testimony in greater detail by focusing on one putative source of testimony, that of Wikipedia. My aim is to the answer the following questions: Is Wikipedia a source of testimony? And if so, what is the nature of that source? Are we to understand Wikipedia entries as a collection of testimonial statements made by individuals, some subset of individuals, or is Wikipedia itself (the organization or the Wikipedia community) the entity that testifies? If Wikipedia itself is a source of testimony, what resources do we have for assessing the trustworthiness of such an unusual epistemic source? In answering these questions I hope to further elucidate the nature of collective epistemic agency (Tollefsen 2006), of which group testimony is a paradigm example.

When a mans Discourse begineth…at some saying of another, of whose ability to know the truth, and of whose honesty in not deceiving, he doubteth not; and then the Discourse is not so much concerning the Thing, as the Person; and the Resolution is called Beleefe, and Faith: Faith in the man. (1651/1991, Ch. 7; p. 48)

This paper focuses on the question of how to resolve disagreement and uses the Lehrer-Wagner model as a formal tool for investigating consensual decision-making. The main result consists in a general definition of when agents treat each other as epistemic peers (Kelly 2005; Elga 2007), and a theorem vindicating the “equal weight view” to resolve disagreement among epistemic peers. We apply our findings to an analysis of the impact of social network structures on group deliberation processes, and we demonstrate their stability with the help of numerical simulations.

I treat you as a thermometer when I use your belief states as more or less reliable indicators of the facts. Should I treat myself in a parallel way? Should I think of the outputs of my faculties and yours as like the readings of two thermometers the way a third party would? I explore some of the difficulties in answering these questions. If I am to treat myself as well as others as thermometers in this way, it would appear that I cannot reasonably trust my own convictions over yours unless I have antecedent reason to suppose that I am more likely than you to get things right. I appeal to some probabilistic considerations to suggest that our predicament as thermometers might not actually be as bad as it seems.

This paper develops a formal framework to model a process in which the formation of individual opinions is embedded in a deliberative exchange with others. The paper opts for a low-resolution modeling approach and abstracts away from most of the details of the social-epistemic process. Taking a bird's eye view allows us to analyze the chances for the truth to be found and broadly accepted under conditions of cognitive division of labour combined with a social exchange process. Cognitive division of labour means that only some individuals are active truth seekers, possibly with different capacities. Both mathematical tools and computer simulations are used to investigate the model. As an analytical result, the Funnel Theorem states that under rather weak conditions on the social process, a consensus on the truth will be reached if all individuals possess an arbitrarily small capacity to go for the truth. The Leading the pack Theorem states that under certain conditions even a single truth seeker may lead all individuals to the truth. Systematic simulations analyze how close agents can get to the truth depending upon the frequency of truth seekers, their capacities as truth seekers, the position of the truth (more to the extreme or more in the centre of an opinion space), and the willingness to take into account the opinions of others when exchanging and updating opinions.

I develop a general framework with a rationality constraint that shows how coherently to represent and deal with second-order information about one's own judgmental reliability. It is a rejection of and generalization away from the typical Bayesian requirements of unconditional judgmental self-respect and perfect knowledge of one's own beliefs, and is defended by appeal to the Principal Principle. This yields consequences about maintaining unity of the self, about symmetries and asymmetries between the first- and third-person, and a principled way of knowing when to stop second-guessing oneself. Peer disagreement is treated as a special case where one doubts oneself because of news that an intellectual equal disagrees. This framework, and variants of it, imply that the typically stated belief that an equally reliably peer disagrees is incoherent, and thus that pure rationality constraints without further substantive information cannot give an answer as to what to do. The framework also shows that treating both ourselves and others as thermometers in the disagreement situation does not imply the Equal Weight view.

The paper develops some of the conclusions, reached in Floridi (2007), concerning the future developments of Information and Communication Technologies (ICTs) and their impact on our lives. The two main theses supported in that article were that, as the information society develops, the threshold between online and offline is becoming increasingly blurred, and that once there won't be any significant difference, we shall gradually re-conceptualise ourselves not as cyborgs but rather as inforgs, i.e. socially connected, informational organisms. In this paper, I look at the development of the so-called Semantic Web and Web 2.0 from this perspective and try to forecast their future. Regarding the Semantic Web, I argue that it is a clear and well-defined project, which, despite some authoritative views to the contrary, is not a promising reality and will probably fail in the same way AI has failed in the past. Regarding Web 2.0, I argue that, although it is a rather ill-defined project, which lacks a clear explanation of its nature and scope, it does have the potentiality of becoming a success (and indeed it is already, as part of the new phenomenon of Cloud Computing) because it leverages the only semantic engines available so far in nature, us. I conclude by suggesting what other changes might be expected in the future of our digital environment.

Conciliatory views of disagreement maintain that discovering a particular type of disagreement requires that one make doxastic conciliation. In this paper I give a more formal characterization of such a view. After explaining and motivating this view as the correct view regarding the epistemic significance of disagreement, I proceed to defend it from several objections concerning higher-order evidence (evidence about the character of one's evidence) made by Thomas Kelly (2005).

I compare the epistemic culture of Wikipedia with the epistemic culture of science, with special attention to the culture of collaborative research in science. The two cultures differ markedly with respect to (1) the knowledge produced, (2) who produces the knowledge, and (3) the processes by which knowledge is produced. Wikipedia has created a community of inquirers that are governed by norms very different from those that govern scientists. Those who contribute to Wikipedia do not ground their claims on their reputations as knowers, for they stand to lose nothing if and when their contributions are found to be misleading or false. And the immediacy of the medium encourages gossip and jokes. Hence, though we have some reason to believe that an invisible hand aids scientists in realizing their epistemic goals, we cannot ground our confidence in what is reported on Wikipedia on the fact that an invisible hand ensures quality. Nor is the information on Wikipedia aptly justified in a manner similar to the way testimony can be justified.

In recent years, various computational models have been developed for studying the dynamics of belief formation in a population of epistemically interacting agents that try to determine the numerical value of a given parameter. Whereas in those models, agents’ belief states consist of single numerical beliefs, the present paper describes a model that equips agents with richer belief states containing many beliefs that, moreover, are logically interconnected. Correspondingly, the truth the agents are after is a theory (a set of sentences of a given language) rather than a numerical value. The agents epistemically interact with each other and also receive evidence in varying degrees of informativeness about the truth. We use computer simulations to study how fast and accurately such populations as wholes are able to approach the truth under differing combinations of settings of the key parameters of the model, such as the degree of informativeness of the evidence and the weight the agents give to the evidence.

Wikipedia has challenged traditional notions about the roles of experts in the Internet Age. Section 1 sets up a paradox. Wikipedia is a striking popular success, and yet its success can be attributed to the fact that it is wide open and bottom-up. How can such a successful knowledge project disdain expertise? Section 2 discusses the thesis that if Wikipedia could be shown by an excellent survey of experts to be fantastically reliable, then experts would not need to be granted positions of special authority. But, among other problems, this thesis is self-stultifying. Section 3 explores a couple ways in which egalitarian online communities might challenge the occupational roles or the epistemic leadership roles of experts. There is little support for the notion that the distinctive occupations that require expertise are being undermined. It is also implausible that Wikipedia and its like might take over the epistemic leadership roles of experts. Section 4 argues that a main reason that Wikipedia’s articles are as good as they are is that they are edited by knowledgeable people to whom deference is paid, although voluntarily. But some Wikipedia articles suffer because so many aggressive people drive off people more knowledgeable than they are; so there is no reason to think that Wikipedia’s articles will continually improve. Moreover, Wikipedia’s commitment to anonymity further drives off good contributors. Generally, some decisionmaking role for experts is not just consistent with online knowledge communities being open and bottom-up, it is recommended as well.

In this paper, we investigate various possible (Bayesian) precisifications of the (somewhat vague) statements of “the equal weight view” (EWV) that have appeared in the recent literature on disagreement. We will show that the renditions of (EWV) that immediately suggest themselves are untenable from a Bayesian point of view. In the end, we will propose some tenable (but not necessarily desirable) interpretations of (EWV). Our aim here will not be to defend any particular Bayesian precisification of (EWV), but rather to raise awareness about some of the difficulties inherent in formulating such precisifications.

At the very end of the 19th century, Gabriele Tarde wrote that all society was a product of imitation and innovation. This view regarding the development of society has, to a large extent, fallen out of favour, and especially so in those areas where the rational actor model looms large. I argue that this is unfortunate, as models of imitative learning, in some cases, agree better with what people actually do than more sophisticated models of learning. In this paper, I contrast the behaviour of imitative learning with two more sophisticated learning rules (one based on Bayesian updating, the other based on the Nash-Brown-von Neumann dynamics) in the context of social deliberation problems. I show for two social deliberation problems, the Centipede game and a simple Lewis sender-receiver game, that imitative learning provides better agreement with what people actually do, thus partially vindicating Tarde.

Evidentialism is the thesis that a person is justified in believing a proposition iff the person's evidence on balance supports that proposition. In discussing epistemological issues associated with disagreements among epistemic peers, some philosophers have endorsed principles that seem to run contrary to evidentialism, specifying how one should revise one's beliefs in light of disagreement. In this paper, I examine the connection between evidentialism and these principles. I argue that the puzzles about disagreement provide no reason to abandon evidentialism and that there are no true general principles about justified responses to disagreement other than the general evidentialist principle. I then argue that the puzzles about disagreement are primarily puzzles about the evidential impact of higher-order evidence–evidence about the significance or existence of ordinary, or first-order, evidence. I conclude by arguing that such higher-order evidence can often have a profound effect on the justification of first-order beliefs.

Journals regulate a significant portion of the communication between scientists. This paper devises an agent-based model of scientific practice and uses it to compare various strategies for selecting publications by journals. Surprisingly, it appears that the best selection method for journals is to publish relatively few papers and to select those papers it publishes at random from the available “above threshold” papers it receives. This strategy is most effective at maintaining an appropriate type of diversity that is needed to solve a particular type of scientific problem. This problem and the limitation of the model is discussed in detail.

Given the fact that many people use Wikipedia, we should ask: Can we trust it? The empirical evidence suggests that Wikipedia articles are sometimes quite good but that they vary a great deal. As such, it is wrong to ask for a monolithic verdict on Wikipedia. Interacting with Wikipedia involves assessing where it is likely to be reliable and where not. I identify five strategies that we use to assess claims from other sources and argue that, to a greater of lesser degree, Wikipedia frustrates all of them. Interacting responsibly with something like Wikipedia requires new epistemic methods and strategies.

Experts take sides in standing scholarly disagreements. They rely on the epistemic reasons favorable to their side to justify their position. It is argued here that no position actually has an overall balance of undefeated reasons in its favor. Candidates for such reasons include the objective strength of the rational support for one side, the special force of details in the case for one side, and a summary impression of truth. All such factors fail to justify any position.

The justification of induction is of central significance for cross-cultural social epistemology. Different ‘epistemological cultures’ do not only differ in their beliefs, but also in their belief-forming methods and evaluation standards. For an objective comparison of different methods and standards, one needs (meta-)induction over past successes. A notorious obstacle to the problem of justifying induction lies in the fact that the success of object-inductive prediction methods (i.e., methods applied at the level of events) can neither be shown to be universally reliable (Hume's insight) nor to be universally optimal. My proposal towards a solution of the problem of induction is meta-induction. The meta-inductivist applies the principle of induction to all competing prediction methods that are accessible to her. By means of mathematical analysis and computer simulations of prediction games I show that there exist meta-inductive prediction strategies whose success is universally optimal among all accessible prediction strategies, modulo a small short-run loss. The proposed justification of meta-induction is mathematically analytical. It implies, however, an a posteriori justification of object-induction based on the experiences in our world. In the final section I draw conclusions about the significance of meta-induction for the social spread of knowledge and the cultural evolution of cognition, and I relate my results to other simulation results which utilize meta-inductive learning mechanisms.