Metainformation II: Law produces metainformation

It is possible to think about law as a kind of metainformation. Law structures information, concepts and actions and describes consequences of an action assuming that it conforms to criteria set up. Law consists of a set of rules that describe how information should be treated, what it is and how it can be transmitted. Law thus produces metainformation in a very real sense.

Take copyright. Copyright information is metainformation about how certain information may move in a market and under what conditions it can be shared, transmitted, stored…the legal status of a certain piece of information determines its use in several different situations, and we see this explicitly in so-called rights-management information systems. In watermarking metainformation even becomes explicit: any work can be assessed as to its legal status by reading the “meta-tag” provided with the work. In intellectual property law we also see even metainformation being regulated. The most obvious example of this is the regulation of rights management information in the European directives on copyright. The prohibition against manipulating or removing electronic rights information (or metainformation as it were) is a legal rule pertaining to metainformation. The RMI is not protected because of its own value, but because of its value as metainformation.

Conversely, law can also require that we know metainformation about information. In Sweden a recent legal reform prohibited the downloading of information from the net if the user knew or should have known that the information was put there without the consent of the rights holder. That rule, in essence, is a requirement that the user contextualize and assess metainformation about a work before using or accessing it (can we use work and information as synonyms? Well, perhaps we should be careful: works are information sets that present with a certain cohesion). In this case an act becomes illegal if the metainformation is not accessed or at least probably determined.

One argument against a law like this could actually be based on this: that in order to maximize the sharing of information we should never require users to assess or try to determine metainformation about any work before they use it. The logic behind such an argument could then be that any other rule would impose costs on information sharing equal to the determination of the probable metainformation applicable to a work, and that such costs will rise with the information explosion we are living through.

A related question is whether law should really be tasked with determining metainformation. That is an interesting question, and fairly complicated. It requires that we think about points of regulation and points of determination. A point of regulation is where a rule applies its force, the target for a rule. Where we place the point of regulation determines where a potential crime is committed. The point of determination is where we assess if the potential crime is an actual crime by comparing the actions we have observed with the rule we have put in place.

Assume that we want to stop access to pornography. We can then do this in two different ways. The first is to force everyone to use a standard – say PICS – and then simply require a) that all browsers implement a forced PICS-analysis and b) that all pictures published on the internet be marked up with PICS (or not visible in browsers, then). We then exclude some categories of “sexual content” in law. That places the point of regulation in the browser and it places the point of determination in the classification of pictures according to the PICS standard, where no classification means that an image will not be displayed in a browser. The potential crime here is for the browser maker not to include the PICS-based filtering, or for the user to circumvent it. The determination of whether a crime is committed then becomes the determination of whether a piece of software has been written according to the demands or if it has been tampered with in any sense.

The other alternative is to legislate and to prohibit individuals from accessing pornography. If they do so they are liable to be fined or jailed according to the severity of punishment we wish to apply. If a work falls under the prohibition in the law is determined by a legal valuation in court. This places the point of regulation at the user and the point of determination in court. The potential crime here is accessing the image, the determination of if this is an actual crime is made by looking at the image and determining if it fulfills the stated criteria for pornography that is prohibited.

One way to understand this difference is to say that the first piece of legislation is architectural and the second value-oriented. In the first case we determine if a crime has been committed by looking at the design or tampering of a piece of software. In the second we try to determine the ethical content of an action.

There is something here that needs our attention. Laws that target architecture are in a very real sense post-ethical laws, they rob legal rules of moral content. A legal rule that targets the design or tampering of software says nothing about the ethical nature of watching pornography. A legal rule that asks if a picture is pornographic requires a value-judgement to a much higher degree (is not design, ultimately, about values too? Yes, of course, and tampering even more so, but there is a difference here that still qualifies as a difference in kind rather than a difference in degree, I believe).

In the first case the image becomes pornographic when the metainformation is applied to it, and this is a process not necessarily performed in court (not probable to be performed there, as a matter of fact, but more likely to be performed by the publisher of the image). In the second case the image is deemed pornographic after a court has looked at it and the determination is made under rule of law.

There seems to be a possible hypothesis here: the more consequential the application of a piece of metainformation, the more important the judicial nature of the process, the transparency and the openness.

Metainformation I: Information about information

Senare i januari ska jag på en konferens med det fina namnet Global Leaders 2010 i Singapore. Ämnet är metainformation och arrangören, Viktor Mayer-Schönberger har redan signalerat att han tror att denna fråga kommer att bli en av de mest intressanta under 10-talet. Jag skall leda en workshop och litet plenarer och så, men framförallt ser jag fram mot diskussionerna. Som en föreberedelse har jag samlat litet tankar och anteckningar om metainformation som jag också publicerar här. De är på engelska.

Metainformation is information about information. There is probably no single point where metainformation suddenly emerges, but we could imagine that information sets of certain diversity and size automatically create a demand for metainformation. One example would be a library. With a few books we can find what we need quickly, but when the size of the book collection grows it quickly becomes impossible to locate books without a systematic description of where the information is. In fact, we could argue, this is the point when the book collection becomes a library. This point, the library point, will differ for all kinds of information sets, but it will always be signified by the same qualities: the emergence of metainformation and attempts to structure the information in the information set to reduce search costs.

Now, when we construct metainformation we can do it in different ways. Constructing metainformation sets is a difficult task since it seems to require that we envision possible uses of the information sets, possible searches that we may want to perform on the material. Metainformation, in a very real sense, will be the thing that enables searches in vast information sets. Without an “index” and relevance indicators search engines would not know how to handle a search query. Or? Let’s examine that idea closer. What is the relationship between search and metainformation? One observation is that the space of possible searches is determined by the metainformation available. When you enter a library you will be able to perform searches according to the metainformation available to you. Imagine two searches: the first is “all relevant books on Picasso” and the second is “all books set in Garamond type”. The first is answerable by the index. The second is not. Metainformation thus delineates the knowable in a very real sense. Wittgenstein used to say that the boundaries of my language are the boundaries of my world (misquoted and misconstrued here, but to illustrate a point). Well, the boundaries of our searching capacity are set by the boundaries implied in metainformation.

What, then, does this mean? I think one answer to that question is that it means that the ways of producing metainformation (in this narrow sense) are ways of producing the boundaries of search space. Innovating the production of metainformation is expanding the set of possioble questions we can ask. It also implies a power relationship. He who controls the metainformation controls the search.

It is worthwhile here to make a slight detour into thinking about different ways of creating power out of information. The first point I think we need to make is that we have moved from an economy where owning information was a viable way of creating value sustainably. It is still possible, where asymmetries of information are stable or at least monetizable at very quick rates, to generate value by owning and transferring ownership of information to other parties. But another form of value creation has become much more interesting, and that is the organization of information. It is well known that Google’s vision is NOT to “own the world’s information and make it universally accessible and useful”. It is to organize the world’s information. But exactly how is value produced when we organize information? I think that it may be here that we should begin to examine the notion that if value is not produced by owning information, it may be produced by owning metainformation. I am not sure this is true, as there is another possibility as well. And that is that what we see is not value accruing to metainformation as such, as much as to the means of producing metainformation.

If this is the case it lends itself to a Marxian analysis. Marx noted that capitalism would collapse under its own victories. We would see capitalism make the means of production available to all, even to the workers, by constantly lowering the cost of the means of production. This is, essentially, what has happened to the content industry. In many cases, though not all (there is an argument here pertaining to quality and investments that I will not address here, but which I think is problematic at best), this means that films, photos, music and other forms of content can easily be produced by users. This phenomenon – user created culture – has shifted the power away from the previous owners of the means of producing content or information to users. This is a well-known analysis presented by Yochai Benkler, Lawrence Lessig and others. But what has happened at the same time is that as the information avalanche grows, we see the value shifting to another set of means of production: the production of metainformation.

The argument that we are beginning to shape here is this: in an age where anyone can make a pop song or take photos, the ability to produce metainformation is still an ability where costs prohibit the emergence of wide-scale user created metainformation.

But is this true? Look at folksonomies and tagging as phenomena. Are these not phenomena that seem to indicate the opposite? If we argue from search engines, then, yes, it seems trivial that not everyone can produce their own index and relevance algorithms on a global scale (here is an interesting question: what will happen when the prizes of indexing and relevance structuring become so low as to allow users to create metainformation on a global level?). But that is only saying that not anyone with a digital film camera can produce Lord of the Rings, right?

Well, yes and no. I think that there are reasons to think about the means of producing metainformation as a new and unevenly distributed source of value. The anatomy of the new source of value may offer interesting fields of exploration, not least when we think about how metainformation is best produced. We seem to need two things: relevance producing mechanisms (or algorithms) as well as vast data sets to test them. Now, one crucial question here seems to be if producing qualitative metainformation (in a sense, producing relevance) is positively correlated with the size of the information sets available to the party producing the metainformation. If this is the case – if those with larger data sets produce better metainformation – it would seem that there is still value in owning information. At least in owning vast information sets.

The larger the information sets I command, in that case, the better the metainformation I can produce. Think about the library. Which library would you predict has the better catalogue? A large library or a small library? Here is a thought: may it be that once we pass the library point where metainformation emerges we see the quality of the metainformation grow with the set of information it is being produced from? Is the quality of metainformation a function of the size of the set it is being generated from? Some studies – specifically of translation technology – would imply that this is indeed the case.

Indeed, beyond the library point, where metainformation emerges, we may imagine another point, the general search point, where the marginal utility of adding information to the set for the value of the metainformation being produced explodes. In such a scenario we should expect the producers of metainformation to try to access all kinds of information and accelerate this process as their information sets grow. Of course, we need to qualify this scenario, by thinking about how different kinds of information add value. One thing we can see is that personal information seems to be rich with metainformation values. The value of work shifts to the value of attention, and the value of the traces of our attention in vast information sets, used to structure metainformation, may very well be the main source of competitive advantage in the metainformation markets.

We are all, in a very real sense, librarians engaged in structuring the world.