computational social science

April 15, 2023

Reflections on the IRTF Research and Analysis of Standard-Setting Processes Research Group

Standard setting is an essential part of the governance of networked digital infrastructure. The formulation of the Hypertext Transfer Protocol (HTTP), for example, has had ubiquitous impact. Changes in standards can lead to shifts to the distribution of wealth and power. For example, the move from HTTP to HTTPS (HTTP Secure) introduced encryption that prevented many forms of eavesdropping and tampering, which was a boon to privacy and human rights and a loss to Internet service providers, who had profited from observing unencrypted web requests. In other domains, standards setting is the site of geopolitical and economic tussle, as in the 3rd Generation Partnership Project (3GPP), the standards development organization where the 5th generation mobile network standard (5G) was defined. While standard-setting often presents a seemingly impenetrable soup of acronyms, Susan Leigh Star argued that scholars should seriously study “boring things” like infrastructure, for that is where power lies. Standard setting is a target rich environment for multidisciplinary, impactful research into the political economy of technology.

So I celebrate that at the Internet Engineering Task Force 116 meeting in Yokohama, Japan (March 24-30, 2023), the inaugural meeting of the Research and Analysis of Standard-Setting Processes (Proposed) Research Group, otherwise known as RASPRG. While there are several communities that study standards-setting processes, to my knowledge this is the first organized within a standards development organization (SDO) itself. Because the IETF is a particularly open and introspective SDO, this provides RASPRG with a kind of native reflexivity and reciprocity, and ready allies in establishing channels of data access, instrumentation, and establishment of ground truth. It is an extremely promising research area that has already attracted a diverse set of researchers that includes computer scientists, ethnographers, and many in-between, as well as interested members of the IETF community. We’ve already had a number of interesting discussions about research questions and ethics on the mailing list; the legality and ethics of the research are top-of-mind since RASPRG is accountable to the greater IETF community, and includes within it many with a research background in data ethics.

A major focus of RASPRG is deriving insights from the comprehensive open data records of the IETF itself. Two research groups in particular have joined forces through RASPRG. The “Streamlining Social Decision Making for Improved Internet Standards” (sodestream) project, out of University of London and University of Glasgow, has been a well-funded research initiative of computer scientists with deep connections to Internet governance who have been developing tools to improve internet standards setting. Structured somewhat differently, BigBang is an open source research infrastructure project for studying SDOs and other on-line collaborative settings. Both these groups have built data science tools for studying the IETF and other infrastructure governance organizations, and bring this expertise to RASPRG.

I’m excited about these new developments for several reasons. As a technology policy researcher, I am often part of debates about the regulation of platforms and “AI”. However, arguably the network protocol layer is just as important as these ‘application layer’ technologies, and is an under-studied site for impactful research. It is arguably where the most significant privacy-by-design is happening, as network protocols have direct implications for, for example, the behavior of web browsers and other user agents.

I’ve also found the IETF to be an intellectually vibrant community that is curious about the economic, political, and ethical implications of its own work. We have already had open-minded and multidisciplinary conversations about difficult questions regarding demographic representation and organizational involvement, and have seen cooperation towards worthy answers to difficult ethical questions.

On a personal level, I love seeing the maturation of BigBang into something with an infrastructural role. I launched the BigBang project in graduate school with a hairbrained idea that we could build a data science tool to study the sociotechnical process of building data science tools. It failed to earn me my doctoral dissertation, but it was used in the dissertations of others who have become contributors and core developers. Though quite quirky technically, one core developer has described the project to me as a “boundary object” for bringing together different kinds of epistemic communities and imaginaries. So be it. We seem to descending the gradient towards the most fundamental forms of digital infrastructure governance, and it’s my hope that we become embedded there as a source of reflexive insight. It’s been a slow process, but BigBang is an ongoing success raising an expanding universe of questions.

Leave a comment

December 29, 2019

Why AI ethics is a difficult problem

A cornerstone of all computer science research is the analysis of the difficulty of solving problems. As is well known, some problems, like sorting a list of numbers, are relatively easy. Other problems, like the knapsack problem, are hard. Here, “easy” and “hard” are defined by computational complexity classes: the amount of processing time it takes to solve the problem as a function of the size of the input.

Statistics has its own internal understanding of the difficulty of solving problems. When doing statistical inference properly, you cannot do better than your data and the validity of your assumptions (c.f. no free lunch theorem). You cannot solve a high dimensional problem with low dimensional data (c.f. the curse of dimensionality).

“AI”, or machine learning, or data science, in its current form is the combination of statistics and computer science. Serious researchers in either domain know that the problems they are solving are often hard. (Deep learning perhaps has allowed the AI research community to suspend their disbelief for a time.)

Consider two problems:

A: The problem of predicting Y from input data X, such that the decision whose value depends on the accuracy of the estimate of Y can be made well.
A’: The problem of predicting the consequences of deploying the system that solves A in a complex sociotechnical world.

Which problem is harder?

However hard problem A is, A’ will be harder. To solve A, you need training data for X and Y, and sound inference and optimization algorithms. To solve B, you need not only training data for X and Y (in order to understand the behavior of A), but also training data from which to learn the structure of the sociotechnical world in which the system is deployed. This will be much higher dimensional data than those used to solve A’. (Simulating the total system and getting a distribution over its outcomes may also prove to be complex in terms of runtime–more complex than the original optimization problem involved in solving A).

Considering this argument, its clear why the difficulty with computer scientist’s solving AI ethics problems is not their use of abstraction as a disciplinary problem (see Selbst et al. 2019). Rather, it’s because the AI ethics problem (A’) is, for abstractly understandable reasons, much harder than the AI problem (A).

There is a great deal of humanistic discussion of AI ethics coming from law, anthropology, and so on. Qualitative research and humanistic understanding are wonderful in part because they allow for a high-dimensional understanding of their phenomena. But they are not free from the laws of logic; rather, their powers and limitations can be better understood by showing how they fit within the formally understood mathematics of learning (Benthall, 2016). When “interpretevist” researchers write about AI ethics, they are often doing important work of raising awareness about the consequences of technical systems. This is, it must be said, somewhat easier to do after the fact. They are not solving the AI ethics problem as it confronts the technology designer originally. For these, the principles of computer science apply.

One last point: any model of a sociotechnical system, internalized within an AI component of that system, will be yet-another-AI with potentially undesirable social consequences. We have discussed problem A, and also problem A’. But we can equally consider problem, A”, the problem of predicting the consequences of deployed system A’. And A”’, A””, A^(n), on into an infinite regress. It’s an interesting question whether the complexity of the problem leaps or plateaus after multiple applications of this operation.

References

Benthall, S. (2016) The Human is the Data Science. Workshop on Developing a Research Agenda for Human-Centered Data Science. CSCW 2016. (link)

Selbst, A. D., Boyd, D., Friedler, S. A., Venkatasubramanian, S., & Vertesi, J. (2019, January). Fairness and abstraction in sociotechnical systems. In Proceedings of the Conference on Fairness, Accountability, and Transparency (pp. 59-68). ACM.

1 Comment

November 22, 2019

In search of an architecture for computational economics

The relationship between scientific research, higher education, and open source software has evolved considerably over the last several years. Today, it’s fair to say that most industrially relevant “data science” practice now depends on open source software that was originally built for scientific research purposes. This has in turn legitimized that software; universities have now placed using open source data science software libraries in their undergraduate curriculum. In computer science and by very loose extension other hard sciences, releasing a high quality software tool is a recognizable academic contribution. We’ve come a long way.

The social sciences have perhaps been slower to take the software turn for many notable reasons. One major reason for this is the broad and disparate nature of the social sciences. A related reason is the disciplinary incompatibility of many social sciences with computational modeling. Abutting this academic resistance to software-based social research, however, is the wide adoption of industrial methods for managing and learning from social data. Arguably, the main industrial drivers of data science have always been social science applications, albeit those within a narrow range. Human-Computer Interaction, Computer Supported Cooperative Work, Management Science, Operations Research, and other business-applicable fields have flourished in recent years in ways that traditional “social sciences” such as Sociology, Anthropology, and History have not.

Enter the question of Economics, widely known to be the hardest (most quantitative) of the social sciences. If there were ever a social scientific field that could make the transition over onto an effective software stack, it would be Econ. In addition to what is in principle a methodological resonance, there is also the plausible link between efficient research tools and industrial applications.

Indeed, the beginnings of an open source economics field are underway. There’s an Open Source Economics Lab at University of Chicago. There’s a NumFOCUS sponsored non-profit, QuantEcon, supporting basic economics tools and associated with Nobel-prize winner Thomas Sargent. There’s Econ-Ark, a different economics toolkit funded by the Sloan Foundation. There’s the Dolo project, and so on.

In this loose taxonomy of scientific software maturity developed at an NSF-funded workshop on Scientific Software Incubators, these projects range between Stage 1, developed by a single software team for internal use, Stage 2, developed by multiple software teams for internal use, and Stage 3, a self-governing community deliberately supporting a broader community.

http://urssi.us/blog/2019/02/25/software-incubator-workshop-a-synthesis/

These are, it must be said, so far small efforts in the field of economics. One explanation for “Why?” comes from the Charter of the nascent Journal for Open Source Economics (JOSEcon). Summarizing the motivations for the journal described in that charter, there’s a compelling argument for the need for a high impact journal that requires of submissions sound software engineering behind its computational tools.

There are computational and numerical methods in economics research with many benefits:
- More expressive than purely analytically tractable models
- Ability to support parameter estimation/model fitting
Software development practice among economics researchers is currently weak
- Mainly informal code transfer with little effective code reuse
- Publication standards are not guaranteeing reproducibility
- Lots of reinventing the wheel
- Potential of a replicability crisis
The solution is a change in incentive structure
- JOSEcon aims to be a high prestige journal that requires better software practices for submissions
- A submission includes:
  - A well-documented software package
  - Short script or notebook demonstrating functionality
  - A couple pages of prose of applicability
  - Could be new research, or a replication of existing research
- Submissions are citable for academic credit towards e.g. tenure

At the moment, there seems to be a bit of a chicken-and-egg problem. Software engineering skills are in short supply among economists. So it’s unlikely that a journal that requires sound software practices behind its submissions will quickly become prominent in the field. On the other hand, it’s possible that the infrastructure for general-purpose scientific publishing will accommodate computational research and it will be left to economists to take advantage of it after the way has been prepared ahead of them.

Current proposals may lack conceptual clarity about software engineering and its precise relationship with academic publication. The incentives and needs of the two fields are subtly different in ways besides how academic research values citations. The library and dependency structure of software depends critically on functional modularity. Arguably, research publications are organized around a more narrative structure. The logic of presentation of a research publication is rarely going to fit the most efficient architecture of computational modeling.

All this points to a fascinating intellectual problem at the core of all this: what is the right architecture for computational economics software tools? Is an economic model a functional unit of logic? Or is it a narrative for presentation? Can the logical units be efficiently decomposed and reused?

Leave a comment

July 12, 2017

Overdetermined outcomes in social science

One of the reasons why it’s important to think about explicitly about downward causation in society is how it interacts with considerations of social and economic justice.

Purely bottom-up effects can seem to have a different social valence than top-down effects.

One example, as noted by David Massad, has to do with segregation in housing. Famously, the Schelling segregation model shows how segregation in housing could be the result of autonomous individual decisions by people with a small preference for being with others like themselves (homophily). But historically in the United States, one factor influencing segregation was redlining, a top-down legal process.

Today, there is no question that there is great inequality in society. But the mechanism behind that inequality is unknown (at least to me, in my current informal investigation of the topic). One explanation, no doubt overly simplified, would be to say that wealth distribution is just a disorganized heavy tail distribution. A more specific account from Piketty would frame the problem as an organized heavy tail distribution based on the feedback effect of the relative difference in rate of return on capital versus labor. Naidu would argue that this difference in the rate of return is due to political agency on the part of capitalists, which would imply a downward causation mechanism from capitalist class interest to individual wealth distributions.

The key thing to note here is that the mere fact of inequality does not give us a lot to distinguish empirically between these competing hypotheses.

It is possible that the specific distribution (i.e cumulative density function) of inequality can shed light on which, if any, of these hypotheses hold. To work this out, we would need to come up with a likelihood function for the probability of the wealth distributions occurring under each hypothesis. Likely the result would be subtle: the difference in the likelihood functions would be about not that but how much inequality results, and whether and in what ways the wealth distribution is stratified.

Of course, another approach would be to collect other data besides the wealth distribution that bears on the problem. But what would that be? The legal record of the tax code, perhaps. But this does not straightforwardly solve our problem. Whatever the laws are and however they have changed, we cannot be sure of their effect on economic outcomes without testing them somehow against the empirical distribution again.

Another challenge to teasing these hypotheses apart is that they are not entirely distinct from each other. A disorganized heavy tail distribution posits a large number of contributing factors. Difference in rate of return on capital may be one important factor. But is it everything? Need it be everything to be an important social scientific theory?

A principled way of going about the problem would be to regress the total distribution against a number of potential factors, including capital returns and income and whatever other factors come to mind. This is the approach naturally taken in data science and machine learning. The result would be the identification of a vector of coefficients that would indicate the relative importance of different factors on total wealth.

Suppose there are 20 such factors, any one of which can be removed with minimal impact on the overall outcome. What then?

Leave a comment

July 11, 2017

Why disorganized heavy tail distributions?

I wrote too soon.

Miller and Page (2009) do indeed address “fat tail” distributions explicitly in the same chapter on Emergence discussed in my last post.

However, they do not touch on the possibility that fat tail distributions might be log normal distributions generated by the Central Limit Theorem, as is well-documented by Mitzenmacher (2004).

Instead, they explicitly make a different case. They argue that there are two kinds of complexity:

disorganized complexity, complexity where extreme values balance each other out to create average aggregate behavior according to the Law of Large Numbers and Central Limit Theorem.
organized complexity, where positive and negative feedback can result in extreme outcomes, best characterized by power law or “heavy tail” distributions. Preferential attachment is an example of a feedback based mechanism for generating power law distributions (in the specific case of network degrees).

Indeed, this rough breakdown of possible scientific explanations (the relatively orderly null-hypothesis world of normal distributions, and the chaotic, more accurately rendered world of heavy tail distributions) was the one I had before I started studying complex systems and statistics more seriously in grad school.

Only later did I come to the conclusion that this is a pervasive error, because of the ease with which log normal distributions (which may be “disorganized”) can be confused with power law distributions (which tend to be explained by “organized” processes). I am a bit disappointed that Miller and Page repeat this error, but then again their book is written in 2009. I wonder whether the methodological realization (which I assume I’m not alone in, as I hear it confirmed informally in conversations with smart people sometimes) is relatively recent.

Because this is something so rarely discussed in focus, I think it may be worth pondering exactly why disorganized heavy tail distributions are not favored in the literature. There are several reasons I can think of, which I’ll offer informally here as possibilities or hypotheses.

One reason that I’ve argued for before here is that organized processes are more satisfying as explanations than disorganized processes. Most people are not very good at thinking about probabilities (Tetlock and Gardner (2016) have a great, accessible discussion of why this is the case). So to the extent that the Law of Large Numbers or Central Limit Theorem have true explanatory power, it may not be the kind of explanation most people are willing to entertain. This apparently includes scientists. Rather, a simple explanation in terms of feedback may be the kind of thing that feels like a robust scientific finding, even if there’s something spurious about it when viewed rigorously. (This is related, I think, to arguments about the end of narrative in social science.)

Another reason why disorganized heavy tail distributions may be underutilized as scientific explanations is that it is counter-intuitive that a disorganized process can produce such extreme inequality in outcomes.

This has to do with the key transformation that is the difference between a normal and a log normal distribution. A normal distribution is a bell-shaped distribution one gets when one adds a large number of independent random variables.

The log normal distribution is a heavy tail distribution one gets by multiplying a large number of positively valued independent random variables. While it does have a bell or hump, the top of the bell is not at the arithmetic mean, because the sides of the bell are skewed in size. But this is not necessarily because of the dominance of any particular factor (as would be expected if, for example, a single factor were involved in a positive feedback loop). Rather, it is the mathematical fact of many factors multiplied creating extraordinarily high values which creates the heavy right-hand side of the bell.

One way to put it is that rather than having a “deep” positive feedback loop where a single factor amplifies itself many times over, disorganized heavy tails have “shallow” positive feedback where each of many factors has a single and simultaneous amplifying effect on the impact of all the others. This amplification effect is, like multiplication itself, commutative, which means that no single factor can be considered to be causally prior to the others.

Once again, this defies specificity in an explanation, which may be for some people an explanatory desideratum.

But these extreme values are somehow ones that people demand specific explanations for. This is related, I believe, at the desire for a causal lever with which people can change outcomes, especially their own personal outcomes.

There’s an important political question implicated by all this, which is: why is wealth and power concentrated in the hands of the very few?

One explanation that must be considered is the possibility that society is accumulated history, and over thousands of years an innumerable number of independent factors have affected the distribution of wealth and power. Though rather disorganized, these factors amplify each other multiplicatively, resulting in the distribution that we see today.

The problem with this explanation is that it seems there is little to be done about this state of affairs. A person can effect a handful of the factors that contribute to their own wealth or the wealth of another, but if there are thousands of them then it’s hard to get a grip. One must view the other as simply lucky or unlucky. How can one politically mobilize around that?

References

Miller, John H., and Scott E. Page. Complex adaptive systems: An introduction to computational models of social life. Princeton university press, 2009

Mitzenmacher, Michael. “A brief history of generative models for power law and lognormal distributions.” Internet mathematics 1.2 (2004): 226-251.

Tetlock, Philip E., and Dan Gardner. Superforecasting: The art and science of prediction. Random House, 2016.

Leave a comment

July 10, 2017

The Law: Miller and Page on Emergence, and statistics in social science

I’m working now through Complex Adaptive Systems by Miller and Page and have been deeply impressed with the clarity with which they lay out key scientific principles.

In their chapter on “Emergence”, they discuss the key problem in science of accounting for how some phenomena emerge from lower level phenomena. In the hard sciences, examples include how the laws and properties of chemistry emerge from the laws and properties of particles as determined by physics. It has been suggested that the psychological states of the mind emerge from the physical states of the brain. In social sciences, there is the open question of how social forms emerge from individual behavior.

Miller and Page acknowledge that “unfortunately, emergence is one of those complex systems ideas that exists in a well-trodden, but relatively untracked, bog of discussions”. Epstein’s (2006) treatment of it is particular aggressive, as he takes aim at early emergence theorists who used the term in a kind of mystifying sense and then attempts to replace this usage with his own much more concrete one.

So far in my reading on the subject there has been a lack of mathematical rigor in the treatment of the subject, but I’ve been impressed now with what Miller and Page specifically bring to bear on the problem.

Miller and Page provide two clear criteria for an emergent phenomenon:

“Emergence is a phenomenon whereby well-formulated aggregate behavior arises from localized, individual behavior.
“Such aggregate behavior should be immune to reasonable variations in the individual behavior.”

Significantly, their first example of such an effect comes from statistics: it’s the Law of Large Numbers and related theorems like the Central Limit Theorem.

These are basic theorems in statistics about the properties of a sample of random variables. The Law of Large Numbers states that the average of a large number of samples will converge on the expected value of the expected value of one sample. The Central Limit Theorem states that the distribution of the sum of many identical and independent random variables will tends towards a normal (or Gaussian) distribution whatever the distribution of the underlying variables are.

Though mathematically statements about random variables and their aggregate value, Miller and Page correctly generalize from this to say that these Laws apply to the relationship between individual behavior and aggregate patterns. The emergent phenomena here (the mean or distribution of outcomes) fulfill their criteria for emergent properties: they are well formed and depend less and less on individual behavior the more individuals there are involved.

These Laws are taught in Statistics 101. What is under-emphasized, in my experience, is the extent to which these Laws are determinative of social phenonema. Miller and Page cite an intriguing short story by Robert Coates, entitled “The Law” (1956), that explores the idea of what would happen if the Law of Large Numbers gave out. Suddenly traffic patterns would be radically unpredictable as the number of people on the road, or in a shopping mall, or outdoors enjoying nature, would be far from average far more often than we’re used to. Absurdly, the short story ends when the statistical law is at last adopted by Congress. This is absurd because of course this is one Law that affects all social and physical reality all the time.

Where this fact crops up less frequently than it should is in discussions of the origins of distributions of wide inequality. Physicists have for a couple decades been promoting the idea that the highly unequal “long tail” distributions found in society are likely power law distributions. Clauset, Shalizi, and Newman have developed a statistical test which, when applied, demonstrates that the empirical support for many of these claims isn’t truly there. Often these distributions are empirically closer to a log normal distribution, which can be explained by the Central Limit Theorem when one combines variables through multiplication rather than addition. My own small and flawed contribution to this long and significant line of research is here.

As far as explanatory hypotheses go, the immutable laws of statistics have advantages and disadvantages. Their advantage is that they are always correct. The disadvantage of these Laws in particular is that they do not lend themselves to narrative explanation, which means they are in principle excluded from those social sciences that hold themselves to argument via narration. Narration, it is argued, is more interesting and compelling for audiences not well-versed in the general science of statistics. Since many social sciences are interested in discussion of inequality in society, this seems to put these disciplines at odds with each other. Some disciplines, the ones converging now into computational social science, will use these Laws and be correct, but uninteresting. Other disciplines will ignore these laws and be incorrect but more compelling to popular audiences.

This is a disturbing conclusion, one that I believe strikes deeply at the heart of the epistemic crisis affecting politics today. No wonder we have “post-truth” media and “fake news” when our social scientists can’t even bring themselves to accept the inconvenience of learning basic statistics. I’m not speaking out of abstract concern here. I’ve encountered this problem personally and quite dramatically myself through my early dissertation work. Trying to make this very point proved so anathema to the way social sciences have been constructed that I had to abandon the project for lack of comprehending faculty support. This is despite The Law, as Coates refers to it whimsically, being well known and “on the books” for a very, very long time.

It is perhaps disconcerting to social scientists that their fields of expertise may be characterized well by the same kind of laws, grounded in mathematics, that determine chemical interactions that the evolution of biological ecosystems. And indeed there is a strong discourse around downward causation in social systems that discusses the ways in which individuals in society may be different from individuals random variables in a large sample. However, a clear understanding of statistical generative processes must be brought to bear on the understanding of social phenomena as a kind of null hypothesis. These statistical laws are due high prior probability, in the Bayesian sense. I hope to discover one day how to formalize this intuitively clear conclusion in more authoritative, mathematical terms.

References

Benthall, S. “Testing Generative Models of Online Collaboration with BigBang (pp. 182–189).” Proceedings of the 14th Python in Science Conference. Available at https://conference. scipy. org/proceedings/scipy2015/sebastian_benthall. html. 2015.

Benthall, Sebastian. “Philosophy of computational social science.” Cosmos and History: The Journal of Natural and Social Philosophy 12.2 (2016): 13-30.

Coates, Robert M. 1956. “The Law.” In The World of Mathematics, Vol. 4, edited by James R. Newman, 2268-71. New York: Simon and Schuster.

Clauset, Aaron, Cosma Rohilla Shalizi, and Mark EJ Newman. “Power-law distributions in empirical data.” SIAM review 51.4 (2009): 661-703.

Epstein, Joshua M. Generative social science: Studies in agent-based computational modeling. Princeton University Press, 2006.

Miller, John H., and Scott E. Page. Complex adaptive systems: An introduction to computational models of social life. Princeton university press, 2009.

Sawyer, R. Keith. “Simulating emergence and downward causation in small groups.” Multi-agent-based simulation. Springer Berlin Heidelberg, 2000. 49-67.

Leave a comment

May 22, 2017

hard realism about social structure

Sawyer’s (2000) investigations into the theory of downward causation of social structure are quite subtle. He points out several positions in the sociological debate about social structure:

Holists, who believe social structures have real, independent causal powers sometimes through internalization by individuals.
Subjectivists, who believe that social structures are epiphenomenal, reducible to individuals
Interactionists, who see patterns of interaction as primary, not the agents or the structures that may produce the interactions
Hybrid theorists, who see an interplay between social structure and independent individual agency.

I’m most interested at the moment in the holist, subjectivist, and hybrid positions. This is not because I don’t see interaction as essential–I do. But I think that recognizing that interactions are the medium if not material of social life does not solve the question of why social interactions seem to be structured the way they do. Or, more positively, the interactionist contributes to the discussion by opening up process theory and generative epistemology (cf. Cederman, 2005) as a way of getting at an answer to the question. It is up to us to take it from there.

The subjectivists, in positing only the observable individuals and their actions, has Occam’s Razor on their side. To posit the unobservable entities of social forms is to “Multiply entities unecessarily”. This perhaps accounts for the durability of the subjectivist thesis. The scientific burden of proof is, in a significant sense, on the holist or hybrid theorist to show why the positing of social forms and structures offers in explanatory power what it lacks in parsimony.

Another reason for the subjectivist position is that it does ideological work. Margaret Thatcher famously once said, “There is not such thing as society”, as a condemnation of the socialist government that she would dismantle in favor of free markets. Margaret Thatcher was highly influenced by Friedrich Hayek, who argued that free markets lead to more intelligent outcomes than planned economies because they are better at using local and distributed information in society. Whatever you think of the political consequences of his work, Hayek was an early theorist in society as a system of multiple agents with “bounded rationality“. A similar model to Hayek’s is developed and tested by Epstein and Axtell (1996).

On the other hand, our natural use of language, and social expectations, and legal system all weigh in favor of social forms, institutions, and other structures. These are, naturally, all “socially constructed” but these social constructs undeniably reproduce themselves; otherwise, they would not continue to exist. This process of self-reproduction is named autopoiesis (from ‘auto-‘ (self-), ‘-poisis’ (-creation)) by Maturana and Varela (1991). The concept has been taken up by Luhmann (1995) in social theory and Brier (2008) in Library and Information Sciences (LIS). As these later theorists argue, the phenomenon of language itself can be explained only as a autopoietic social system.

There is a gap between the positions of autopoiesis theorists and the sociological holists discussed by Sawyer. Autopoiesis is, in Varela’s formulation, a general phenomenon about the organization of matter. It is, in his view, the principle of organization of life on the cellular level.

Contrast this with the ‘holist’ social theorist who sees social structures as being reproduced by the “internalization” of the structure by the constituent agents. Social structures, in this view, depend at least in part on their being understood or “known” by the agents participating in them. This implies that the agents have certain cognitive powers that, e.g., strands of organic chemicals do not. [Sawyer refers to Castelfranchi, 1998 on this point; I have yet to read it.] Arguably, social norms are only norms because they are understood by agents involved. This is the position of Habermas (1985) for example, whose whole ethical theory depends on the rational acceptance of norms in free discussion. (This is the legacy of Immanuel Kant.)

What I am arguing for is that there is, in actuality, another position, not identified by Sawyer (2000), on the emergence of social structure that does not depend on internalization but that nevertheless has causal reality. Social forms may arise from individual activity in the same way that biological organization arises from unconscious chemical interactions. I suppose this is a form of holism.

I’d like to call this view the “hard realist” view of social structure, to contrast with “soft realist” views of social structure that depend on internalization by agents. I don’t mean for this to be taken aggressively, but rather because I have a very concrete distinction in mind. If social structure depends on internalization by agents, then that means (by definition, really) that there exists an intervention on the beliefs of agents that could dissolve the social structure and transform it into something else. For example, an left-wing anarchist might argue that money only has value because we all believe it has value. If we were to just all stop valuing money, we could have a free and equal society at last.

If social structures exist even in spite of the recognition of them by social actors, then the story is quite different. This means (by definition) that interventions on the beliefs of actors will not dissolve the structure. In other words, just because something is a social construct does not mean that it can be socially deconstructed by a process of reversal. Some social structures may truly have a life of their own. (I would expect this to be truer the more we delegate social moderation to technology.)

This story is complicated by the fact that social actors vary in their cognitive capacities and this heterogeneity can materially impact social outcomes. Axtell and Epstein (2006) have a model of the formation of retirement age norms in which a small minority of actors make their decision rationally based on expected outcomes and the rest adopt the behavior of the majority of their neighbors. This results in dynamic adjustments to behavior that, under certain parameters, make the total society look more individually rational than they are in fact. This is encouraging to those of us who sometimes feel our attempts to rationally understand the world are insignificant in the face of social inertia more broadly speaking.

But it also makes it difficult to judge empirically whether a “soft realist” or “hard realist” view of social structure is more accurate. It also makes the empirical distinction between the holist and subjectivist positions difficult, for that matter. Surveying individuals about their perceptions of their social world will tell you nothing about hard realist social structures. If there are heterogenous views about what the social order actually is, that may or may not impact the actual social structure that’s there. Real social structure may indeed create systematic blindnesses in the agents that compose them.

Therefore, the only way to test for hard realist social structure is to look at aggregate social behavior (perhaps on the interactionist level of analysis) and identify where its regularities can be attributed to generative mechanisms. Multi-agents systems and complex adaptive systems look like the primary tools in the toolkit for modeling these kinds of dynamics. So far I haven’t seen an adequate discussion of how these theories can be empirically confirmed using real data.

References

Axtell, Robert L and Epstein, J. M. “COORDINATION IN TRANSIENT SOCIAL NETWORKS: AN AGENT-BASED COMPUTATIONAL MODEL OF THE TIMING OF RETIREMENT ROBERT L. AXTELL AND JOSHUA M. EPSTEIN.” Generative social science: Studies in agent-based computational modeling (2006): 146.

Brier, Søren. Cybersemiotics: Why information is not enough!. University of Toronto Press, 2008.

Castelfranchi, Cristiano. “Simulating with cognitive agents: The importance of cognitive emergence.” International Workshop on Multi-Agent Systems and Agent-Based Simulation. Springer Berlin Heidelberg, 1998.

Cederman, Lars-Erik. “Computational models of social forms: Advancing generative process theory 1.” American Journal of Sociology 110.4 (2005): 864-893.

Epstein, Joshua M., and Robert Axtell. Growing artificial societies: social science from the bottom up. Brookings Institution Press, 1996.

Habermas, Jurgen, Jürgen Habermas, and Thomas McCarthy. The theory of communicative action. Vol. 2. Beacon press, 1985.

Hayek, Friedrich August. “The use of knowledge in society.” The American economic review (1945): 519-530.

Luhmann, Niklas. Social systems. Stanford University Press, 1995.

Maturana, Humberto R., and Francisco J. Varela. Autopoiesis and cognition: The realization of the living. Vol. 42. Springer Science & Business Media, 1991.

Sawyer, R. Keith. “Simulating emergence and downward causation in small groups.” Multi-agent-based simulation. Springer Berlin Heidelberg, 2000. 49-67.

Leave a comment

May 17, 2017

Sawyer on downward causation in social systems

The work of R. Keith Sawyer (2000) is another example of computational social science literature that I wish I had encountered ten years ago. Sawyer’s work from the early ’00’s is about the connections between sociological theory and multi-agent simulations (MAS).

Sawyer uses an example of an improvisational theater skit to demonstrate how emergence and downward causation work in a small group setting. Two actors in the skit exchange maybe ten lines, each building on the expectations set by the prior actions. The first line establishes the scene is a store, and one of the actors is the owner. The second actor approaches; the first greets her as if she is a customer. She acts in a childlike way and speaks haltingly, establishing that she needs assistance.

What changes in each step of the dialogue is the shared “frame” (in Sawyer’s usage) which defines the relationships and setting of the activity. Perhaps because it is improvisational theater, the frame is carefully shared between the actors. The “Yes, And…” rule applies and nobody is contradicted. This creates the illusion of a social reality, shared by the audience.

Reading this resonated with other reading and thinking I’ve done on ideology. I think about situations where I’ve been among people with a shared vision of the world, or where that vision of the world has been contested. Much of what is studied as framing in media studies is about codifying the relations between actors and the interpretation of actions.

Surely, for some groups to survive, they must maintain a shared frame among their members. This both provides a guide for collective action and also a motivation for cohesion. An example is an activist group at a protest. If one doesn’t share some kind of frame about the relationships between certain actors and the strategies being used, it doesn’t make sense to be part of that protest. The same is true for some (but maybe not all) academic disciplines. A shared social subtext, the frame, binds together members of the discipline and gives activity within it meaning. It also motivates the formation of boundaries.

I suppose the reification of Weird Twitter was an example of a viral framing. Or should I say enframing?! (Heidegger joke).

Getting back to Sawyer, his focus is on a particularly thorny aspect of social theory, the status of social structures and their causal efficacy. How do macro- social forms emerge from individual actors (or actions), and how do those macro- forms have micro- influence over individuals (if they do at all)? Broadly speaking in terms of theoretical poles, there are historically holists, like Durkheim and Parsons, who maintain that social structures are real and have causal power through, in one prominent variation, the internalization of the structure by individuals; subjectivists, like Max Weber, who see social structure as epiphenomenal and reduce it to individual subjective states; and interactionists, which focuses on the symbolic interactions between agents and the patterns of activity. There are also hybrid theories that combine two or more of these views, most notably Giddens, who combines holist and subjectivist positions in his theory of structuration.

After explaining all this very clearly and succinctly, he goes on to talk about which paradigms of agent based modeling correspond to which classes of sociological theory.

References

Sawyer, R. Keith. “Simulating emergence and downward causation in small groups.” Multi-agent-based simulation. Springer Berlin Heidelberg, 2000. 49-67.

Leave a comment

May 16, 2017

Similarities between the cognitive science/AI and complex systems/MAS fields

One of the things that made the research traditions of cognitive science and artificial intelligence so great was the duality between them.

Cognitive science tried to understand the mind at the same time that artificial intelligence tried to discover methods for reproducing the functions of cognition artificially. Artificial intelligence techniques became hypotheses for how the mind worked, and empirically confirmed theories of how the mind worked inspired artificial intelligence techniques.

There was a lot of criticism of these fields at one point. Writers like Hubert Dreyfus, Lucy Suchman, and Winograd and Flores critiqued especially heavily one paradigm that’s now called “Good Old Fashioned AI”–the kind of AI that used static, explicit representations of the world instead of machine learning.

That was a really long time ago and now machine learning and cognitive psychology (including cognitive neuroscience) are in happy conversation, with much more successful models of learning that by and large have absorbed the critiques of earlier times.

Some people think that these old critiques still apply to modern methods in AI. Isn’t AI still AI? I believe the main confusion is that lots of people don’t know that “computable” means something very precisely mathematical: it means a function that is calculable by a partial recursive function. It just so happens that computers, the devices we know and love, can compute any computable function.

So what changed in AI was not that they were using computation to solve problems, but the way they used computation. Similarly, while there was a period where cognitive psychology tried to model mental processes using a particular kind of computable representation, and these models are now known to be inaccurate, that doesn’t mean that the mind doesn’t perform other forms of computation.

A similar kind of relationship is going on between the study of complex systems, especially complex social systems, and the techniques of multi-agent system modeling. Multi-agent system modeling is, as Epstein clarifies, about generative modeling of social processes that is computable in the mathematical sense, but the fact that physical computers are involved is incidental. Multi-agent systems are supposed to be a more realistic way of modeling agent interactions than, say, neoclassical game theory, in the same way that machine learning is a more realistic way of modeling cognition than GOFAI.

Given that, despite (or, more charitably because of) the critiques leveled against it, cognitive science and artificial intelligence have developed into widely successful and highly respected fields. We should expect complex systems/multi-agent systems research to follow a similar trajectory.

Leave a comment

Category: computational social science

References