I don’t make a point of following Clay Shirky’s writings but every now and then they cross my path since it seems he has a sizable following on the Internet that enthusiastically makes sure that everyone else knows when he has written something new. Given his popularity on the net, I’m all the more frustrated by his tendency to suggest that his writings are grounded in well-established theory and then go on to casually mangle those theories beyond recognition with his limited understanding of them. His recent article, “Institutions, confidence, and the news crisis“, carries on his trend of rending theories almost unrecognizable, the victim this time being Coase’s economic theories about institutions and organizations.
A couple of days ago I came across a reference to a new article by Shirky in which he responds to Dean Starkman’s criticism of what he calls the “Future of News” (FON) movement, represented by Shirky and others. In Shirky’s article, “Institutions, confidence, and the news crisis“, he references well-established theory before going on to demonstrate, at best, an incomplete understanding of the theory (as he all too often does – see below). In this case, the victim is Coase’s theory on institutions and organizations. Shirky makes no attempt to differentiate between institutions and organizations, a key distinction in Coase’s thinking, and even seems to treat them as interchangeable. To be fair to Shirky, Starkman makes the same mistake. However, Starkman makes no attempt to ground his use of the terminology in Coase’s theories, which Shirky does very explicitly.
Very early on in his article, Shirky says, “Institutions reduce the choices available to their members. (This is Ronald Coase’s famous argument about transaction costs.) This reduction allows better focus on the remaining choices they face.” The first part is a very common characterization of Coase’s thinking, i.e. that institutions are constraining, in a certain sense. But, to fully grasp the significance of what is being said, we need to understand what Coase means by “institution” (it’s not found in the particular paper on transaction costs that Shirky links to but is suggested and was better developed later). Coase makes a clear distinction between “institutions” and “organizations” (as do the social sciences in general). “Institutions” are the rules that apply to certain situations and practices (or as it is sometimes described, “how things are done”). Thus, we can describe the “institution” of marriage as a practice that is governed by specific norms and regulations. “Organizations” are the social units that people construct to work toward common goals. Organizations are often influenced by the institutions that govern practice in their specific domain. A Catholic church is an example of an organization, some of whose practices are governed by the institution of marriage.
Although Coase did not make a clear distinction between institutions and organizations in his famous paper on “transaction costs” that Shirky links to, it clearly illustrates the first steps in that direction (eventually culminating in Coase’s “new institutional economics”). In the linked paper, Coase is primarily focused on “the firm”, an example of an organization, and why, in certain instances, firms do not play by the rules of their principle governing institution, i.e. “the market”. What he finds is that organizations may, in some cases, perceive a benefit from playing by the rules of a different institution than that which would be expected to dominate its practice.
What makes this all a bit confusing is that the theoretical term “institution” differs from its common colloquial use. In everyday parlance the term “institution” is used to describe what academics refer to as an “organization”, and it is obviously in this sense that Starkman uses the term. That’s not problematic because he is consistent in his use of the term. Shirky could be excused for carrying on with the same usage but for the fact that he specifically references Coase and states, in the same thought, that “Institutions reduce the choices available to their members.” The thing is that that statement does not apply to organizations. Organizations, as such, do not limit individuals’ choices. It is the rulebook (i.e. the institution) that the organization chooses to adhere to that limits the choices. There is nothing problematic with an organization choosing to adhere to a different rulebook than would be considered the norm. So, while institutions limit choices, they also present opportunities to do things differently, i.e. by presenting a choice of rulebooks to organizations.
So, here is Shirky’s problem. He chooses to frame his discussion of “institutions” by referencing Coase. He then goes on to use the terminology in a manner that is not consistent with his point of reference. Finally, what carries him over the hurdle into the domain of the non-sensical; he uses phrases to set up his argument that rely on a correct interpretation of the key terminology (really, try reading the article with Coase’s intended definitions of the key terms in mind – it makes no sense whatsoever). Furthermore, he seems completely oblivious to his mistake. And what’s worst – there seems to be a pattern of these kinds of mistakes in Shirky’s work.
Is there a pattern here?
The first time I encountered this problem was in Shirky’s oft cited criticism of the semantic web, “The Semantic Web, Syllogism, and Worldview“. In the article, Shirky purports to demonstrate the limitations of syllogistic reasoning as a logical foundation for a more meaningful and “intelligent” web (i.e. the “semantic web”). All he manages to accomplish, however, is to demonstrate his very limited understanding of syllogisms (I emailed Shirky a response to his article many years ago and never received a response – I’ve included the pertinent text of the email at the end of this article).
Several months ago I again happened on one of Shirky’s articles that was being bandied about on the net. This time he was writing about the impact of technology on news publishing in an article titled, “Why we need the new news environment to be chaotic“. In it, Shirky suggests that news is a public good. He then goes on to define “public good” in a manner that would not be familiar to anyone with a reasonable knowledge of basic economics. My reaction to that article can be read here.
Excerpt of email to clay@shirky.com 29.12.2006:
“I also read ‘The Semantic Web, Syllogism, and Worldview’, where you largely base your rather devastating criticism of the Semantic Web on a criticism of syllogisms. The syllogism approach is interesting, but I see a couple of problems in this formulation. Firstly, your presentation of syllogisms is incomplete and wrong to the point where the resulting criticism is fatally flawed. Secondly, syllogisms, as a means of evaluating the validity of logical constructs, are far more common than you claim.
The key point that you omit from your presentation of syllogisms is that syllogisms are categorized into 4 different moods and 4 different figures based on the scope of the premises and their structure. To take your own example:
– The creator of shirky.com lives in Brooklyn
– People who live in Brooklyn speak with a Brooklyn accent
In a correctly formatted syllogism the scope of each premise must be clear (one of: all, none, some or some not), since this affects the validity of the argument. In your first premise, the scope is clear because you refer to a specific individual (creator of shirky.com). Therefore, the statement pertains to all instances of that reference (there is only one). Reforming that into a more syllogism-friendly statement yields (we are allowed to do this because syllogisms are not intended for analysing literal sentences, but rather what is being expressed in a literal sentence):
– All creators of shirky.com are people who live in Brooklyn
In your second premise, the scope could be ‘all’ or ‘some’ instances. You cannot determine the validity of the syllogism without knowing which pertains. As such this syllogism is incomplete. Yet, following this example, you go on to evaluate it as if the premise refers to all instances of ‘people who live in Brooklyn’ (scope = all). The syllogism friendly version is then:
– All people who live in Brooklyn speak with a Brooklyn accent
And our conclusion is:
– All creators of shirky.com speak with a Brooklyn accent
In formal syllogism representation this argument would be described as AAA-4. In fact, all syllogisms of this form are valid. But, remember that syllogisms are for evaluating the validity of an argument. Saying that an argument is valid is not the same as saying that it is true.
In the following paragraph you go on to say that ‘each [statement] is true in a different way. It is tempting to note that the second statement is a generalization that can only be understood in context, but that way madness lies.’ I assure you that there lies no madness. What you are saying here (and actually with your final sentence in the preceding paragraph) is that the second premise should be of the form:
– Some people who live in Brooklyn speak with a Brooklyn accent
which is an entirely different form of syllogism. Assuming our conclusion is unchanged, we now have a AIA-4 syllogism, which in all cases is invalid for very obvious reasons.
If you leave out the significance of the proper form of syllogisms and their premises it is quite easy to come to the conclusion that they are not relevant to our daily lives. But, I hope that my explanations above have clarified that at least to some extent. It should then be clear that we make decisions based on syllogisms all the time. For example, let’s say I search on Google for ‘syllogism’ and get thousands of hits:
– All web sites ending in .edu are authoritative
– All authoritative sites have the information I need
– All sites ending in .edu have the information I need
This is of a familiar form, AAA-4, and as I said before, all syllogisms of this form are valid. Now, I may not conciously go through a process like this, but it does seem a fair representation of how I might justify my choices if asked.
Up until now, I’ve been talking about so-called ‘categorical syllogisms’, but ‘hypothetical syllogisms’ are much more common and adhere to the same basic rules as the former. Hypothetical syllogisms aim to describe our practical decision making, like ‘how early do I have to get up to catch a 7.30 bus to get to work by 8?’. But, I’m not going to get into that now.
So, whatever might be the problem with the notion of a Semantic Web, I don’t think that it is due to a reliance on syllogisms. A far more obvious and well-known problem is the inability of computers to extract meaning from natural language, which they would need to be able to do to make effective use of syllogisms, since we rarely express our thoughts in the appropriate syllogism-friendly form.
Best regards,
Tryggvi Thayer”
