This paper examines new challenges to governments brought on by the evolving state of the human knowledge store. We contend that the nature of information is changing, creating tremendous opportunities for society, and to effectively capitalize on these opportunities new support systems, methods and attitudes; referred to here collectively as infostructure, should be considered.

The use of the term infostructure indicates the casting of informational standards and knowledge stores, in the role of public goods, capable of increasing aggregate welfare, and consequently competing with the schools, roads and bridges of traditional public infrastructure expenditure.

As the economies of scale known to networking-in-general, take on new significance for digitalized networks in the form of nested externalities, it might become increasingly difficult to defend non-collaborative infostructure strategies.

Using a schema for government information policy defined by Carl Shapiro and Hal R. Varian[SH97] , we will intersect the concept of infostructure with each of the following three categories developed by those authors.

A problem endemic to our subject matter is keeping arguments within the technical realm, after all this paper is being given at a technical conference. But determining exactly how much government should do or not do with revenue collected from its citizens can never be merely a technical decision. One can hardly say that some sort of technology is technically suitable as a national investment, while other technology is technically not. Great technology could after all prove to be a disastrous public investment. The issue must be, whether in the light of prevailing political norms, governments should or should not make technologically related investments, or institute regulatory control mechanisms, when the calculated outcome of such investments and regulations meets some economical cost benefit criterion.

We definitely do not wish to define or advocate a political norm for the appropriate levels of public spending and regulation, but it goes without saying that modern governments do collect taxes and spread them around, and if one can work out that spending money on a suspension bridge will result in x added value to the community as a whole, then one might also be able to work out that spending money on an info-bridge will give y added value to the community as a whole. Once such aggregate welfare ratios are calculated, it should be easier to interpolate them with political norms. In a society such as Robert Nozick's Ultraminimal State or David Friedman's Anarcho-Capitalism, no matter how positive the CBA ratio, there would be no public investment made, but in a typical EU member state such as Sweden, one might anticipate other outcomes.

Uninitiated readers should know that there is an ongoing debate in the economic community concerning, network effects, network externalities and path dependencies, much of which is currently in the limelight due to the Microsoft trials. In brief:

There is some confusing semantics here; for example network effects do not necessarily have to come about in, what we in the IT community would call, a network environment. Non interacting parties could be networked merely through a similarity of choices such as buying the same product. To make this distinction clear, Liebowitz and Margolis have introduced a further term; synchronization value; ["The additional value derived from being able to interact with other users of the product."] [lm]

Even though economics is the third element in our techno-political-economical equation that will determine the future of infostructures, we will try to treat the subject as we do politics (avoid it), with the exception of one observation: The economists writing about networks effects do not seem to have a clear grasp of the technologies involved, as least not for the sort of networks in our domain. Here is a quote from Liebowitz and Margolis [lm] ["... it is sometimes assumed that the network type that offers the greatest surplus when network participation is large is the one that offers the smallest surplus when participation is small.] [...] [This condition, however, is not likely to be satisfied, since synchronization effects are likely to be uniform. For example, if there is value in a cellular telephone network becoming larger, this should be equally true whether the network is digital or analog." ]

But as we shall see, there is all the difference in the world whether a network is digital or analog.

Suppose that all practicing of law is carried out in Latin and all philosophizing is done in German. If you wish to be a lawyer or law-maker you must learn Latin, and if you wish to philosophize you must learn German. In learning either language you will profit by the network effects they both offer, you will gain access to other channels of knowledge also sharing either of these languages, and you will be able to converse with native German and Latin speakers. Conversely the existing population of both knowledge domains and both languages will gain by your entry into their networks.

On the downside, lawyers and philosophers will not, without going to the trouble of learning one more language, be able to interact with each other.

If both disciplines produce literature, such as books, then they would collectively contribute to the network of book publishing, which would not happen if only philosophers published books while lawyers exclusively choose television as their media of discourse. Joint participation in the media of books creates economies of scale across the board, from the sharing of printing presses to the sharing of bookshelves, and that would also be the case if the commonality was instead video: we would have more use for our VCRs - all in spite of whatever language is used.

Once the books are on our bookshelves, what further network values do they represent? For sure, some level of reading skills are common to both German and Latin - if we could read both languages. Yet even if we could, or even if by now all the books had been translated into some common language such as English, what do a bunch of books sitting on our bookshelf have to offer each other.

What does the Short History of Western Legal Theory ( by J.M. Kelly) have to say to Hegel´s Phenomenology of Spirit (edited and translated by A.V. Miller) on this paper's author's bookshelf. What do any of his books in English (or Swedish for that matter) have to say, more than mutely pointing at one another: The Short History of Western legal theory points to the Phenomenology of Spirit, Dawkins points to Popper, Popper points to Russel, Russel points to Hegel. Yet there they sit; the Philosophy of History and the History of (western) Philosophy and act as if they would rather not know each other, like they had nothing at all in common.

Books offer convenience, they are convenient to read and convenient to have, and if a reader possesses them in the right combinations she can follow their various pointers herself, by picking that which is pointed to down off her bookshelf and reading it. This reading is the culmination of the network of books, as the reader internalizes their contents. Though she may in turn write a book of her own, the books on her shelf will never come alive like the toys in the toy store, closed for the evening, and begin to mingle and carry on. And this parallels the end-of-the-highway status of all of our analog information stores. This is where so much of our individual or collective efforts terminate. Of course one should protest that the book is not a dead-end. Qualified readers can partake of this information. It can be understood and reiterated. It can be used to infer, calculate, produce or influence new results.

But what if the contents of these books or the information store in general is redeployed within the digital channel? Could then the "Philosophy of History" and the "History of Philosophy" begin to carry on and argue and agree to disagree with each other and so on. And in other information stores, would the statistical measurements of 1957 be able to compare results with the measurements of 2003 and would the genetic code of the Geobacillus thermodenitrificans voice an opinion about the genetic code of the Wheat streak mosaic virus.

One might argue that information itself does not communicate, but perhaps applications or agents might be able to accomplish something of the above if information was not only digital, but also well structured.

We said in our thought experiment that the commonality of knowing Latin could offer lawyers some externalities to internalize upon, two or more knowers of a shared language can benefit from each other - those who can not communicate at all can't. We will find that every network effect has a commonality that defines it. Some medium or channel such as a language, preference, shared trust, etc. and that these channels seldom come alone but rather in aggregate mixes. The Liebowitz-Margolis phone example, has the obvious commonality of a shared use-technology - the phone, but that technology in turn might be part of another network; as they write, an analog or digital network.

If two lawyers, both knowing Latin, participate in the same network, say in the use domain of Insurance Law, they might still run into problems of incommensurability since one belongs to a network of insurance law specific to Japan and the other belongs to a network specific to Europe. The advantages of knowing Latin might not amount to much in this case. There could still be an interoperability problem.

Analog technologies of representation and communication tend to be stickier to the domain specific incommensurability of their origins, while digital technology tends to flow with universal solutions and therein a higher degree of interoperability. It is this interoperability that makes belonging to digital networks valuable.

Digitalization creates a network effect of its own. For example, if a significant amount of music is produced within -, or transferred over to the digital realm (using some standard), this will cause a network effect, just as it would for digitally created-, or transformed engineering designs (CAD and CAM). But when digital audio and digital engineering share the same storage and transportation systems, a network of network effects begins to unfold. One network is nested inside another. And while nested networks will be able to exploit each others gains, they will be immune (if properly aligned to each other in a frictionless manner) to some of the drawbacks of network standardization defined by Brynjolfsson and Kemerer[BK].

Like racing bicyclers who gather in formations to combat wind resistance, frictionlessly coupled network layers, are free to break rank whenever adjacent layers falter, just as digital audio coding survived the demise of the DAT recorder.

Digitalization is not a prerequisite for the interoperability of information. Unambiguous coding of any sort can also facilitate interactions in a semiautonomous manner. A scientist can aggregate non-digital productions in an equation, trusting in their correctness. An economist can assume that an economic report is true and add it to her calculation, without checking the veracity of every atomic item. But the combined values of nested networks through digitalization are unparalleled since so many layers can be encompassed.

Beyond the value of sharing a nest, what more interoperability should one expect between the network of digital music and the network of digital engineering (computer aided design and manufacturing) of our example above? In reality the digital paths of these two fields have benefited increasingly from one another for the last 30 years, beyond common processing, storage and transmission, beyond common graphical interface conventions. The similarities of a music program such as Audio Logic and a CAD program such as AutoCad can not escape even the most casual observance.

There is nothing new about regarding information as infrastructure. Governments do look upon certain information stores as national resources, and do make them public, either by choice or obligation: sometimes totally free; redistributing the cost to the general taxpayer, sometimes in accordance with non-profit cost recuperation, and occasionally at a profit. What is changing, is the accelerating impact of network effects and increasing returns made possible by digitalization and Internet related technology, and consequently the scope and depth of what might be considered a national (or global) informational resource.

There are technological arguments as to the optimal size and scope of systems comparable to arguments as to the optimal size and scope of firms and markets. Infostructures are both markets and systems simultaneously. Wide scale networking forces technology to act as a market - within a market. Complexity and diminishing returns are limit setters for size and scope in networks. Technology affects these limits.

We maintain that information has a quality-state, albeit a dynamic and evolutionary one, now experiencing a period of remarkable development, but this phenomena is upstaged in the public eye by the attention grabbing evolution of tools.

When conceptualizing Moore's Law and the tremendous progress being made in speed, power and cost-performance ratios, we tend to visualize those gains primarily as the advancement of contraption technology (computers, mobile phones and other digiphernalia) and secondarily as the refinement of operating systems and programs: Linux, Java, Windows, Photoshop, Office, Flash etc., with little very thought left over for the change-process of information itself.

The thread of this series of conferences is the markup of information. The markup in focus was initially SGML. Whether generalized or extensible, a markup language is about strengthening information, by reinforcing it with machine readable capabilities and mechanisms of referentiality, advancing the quality state. This is the infocentric approach to information technology and can be contrasted with the toolcentric approach, which has traditionally domintated the digitalization process.

Internet globalization and modern information technology has visibly upset the process of international standards, giving cause for governments to revaluate their traditional positions on standards issues.

Prior to the Internet, the international community had developed a functional (if sometimes feisty) network of standards bodies. Publication- and subscription financed organizations served effectively the interests of government and industry alike. Government involvement was often in the form of direct participation in standard's work or via tender specifications for public expenditures.

While on the surface, the emergence of impromptu standards bodies such as the IETF, the W3C, and ECMA can be seen as a reaction to the time-to-market sluggishness of ISO, ANSI and other internationally recognized standards organizations, voices have been raised in general as to the effectiveness and desirability of publication financed standards bodies in the information age.