3.9.1 The Concept of a Model

Much of the argument in Lessig's book revolves around Yochai Benkler's idea that a communications network can be understood in terms of three ‘layers’ – physical, code/logical and content. Lessig uses this insight as a tool for communicating some of his key ideas and as a tool for analysis. It helps him communicate because it gives his readers a framework for thinking about something – the internet – that seems like an amorphous mass to most people: this is why on page 25, for example, he seeks to relate the three layers to more familiar everyday concepts like Speakers' Corner, Madison Square Garden and cable TV. And the Benkler notion provides him with a way of analysing the issue of which aspects of the Net ought to be treated as a commons and which ought not.

Table giving examples of four communications systems using the three-layers model. These are Speakers’ Corner, Madison Square Garden, a phone system and cable TV. The status of each layer is described as either free or controlled. For Speakers’ Corner all three layers - Content, Code/Logical, Physical - are free. For Madison Square Garden the Content and Code/Logical layers are free but the Physical layer is controlled. For a phone system the Content layer is free, but the Code/Logical and Physical layers are controlled. For cable TV all three layers - Content, Code/Logical, Physical - are controlled.

But the Benkler insight, though powerful, is not all-powerful. It doesn't map exactly onto all cases of real-world communication systems. This is because it is only a model. I want to explore here what that means.

A model is a simplified representation of some aspect of reality constructed for a particular purpose. A model can exist in a person's head only, invisible to everyone else. Or it can also exist external to the modeller, for example, as a diagram on paper or a piece of text or mathematics.

The key features of a model are (i) that it is a simplification, and (ii) that it has been constructed or articulated for a particular purpose.

To illustrate this, let's look at the famous map of the London Underground, which can be downloaded from www.tfl.gov.uk.

This is a model because it's a simplified representation of an aspect of reality (London and its tube network) constructed for a particular purpose. It's simplified in various ways – for example, it represents the lines as mainly straight segments of line, whereas in reality they curve and wind their way underneath the capital's streets. And it exists for a particular purpose, to help travellers find their way from one tube station to another; it's of very limited use to anyone trying to navigate through the actual streets of the city. The simplification is helpful because it clears away the complexity of the city's topography. And it's brilliantly useful for its designated purpose, but not much use for other purposes.

Benkler's layer model is similar. It provides a comprehensible simplification of a very complex reality. There are hundreds of other ways of representing the internet. For example, some scientists at University College London constructed and maintained maps of the network. You can find them at www.cybergeography.org/atlas/atlas.html.

These maps are fascinating and useful models for some purposes. But they are of little use if your purpose is to understand issues relating to intellectual property and regulation on the Net. For that we need something like the Benkler model. And if you find that it's not perfect or it's difficult to relate to other aspects of communication, then that's simply a reflection of the fact that it is a model and therefore, by definition, simplified and specific.

Let's look at another example. Suppose I want to give one of my neighbours a box of books because my children have outgrown them. This involves another simple communications system. Is it possible to decipher the three layers in this transaction?

The content layer would be the print and pictures in the books.

The physical layer would be the pages and covers of the books, as well as the box.

What about the code layer? This would be governed by the social protocols (unwritten rules of behaviour) involved in the relationship I have with my neighbour. How well do I know my neighbour? How often do we see each other? How do we greet each other (with a ‘hello’ or a handshake or both)? How will my neighbour react to the offer? and so on. This is a complex series of norms worked out over a period of time and in accordance with locally accepted social behaviour. These dictate the nature of the transaction or negotiation. It might be possible for me to do something as simple as running into the neighbour and saying ‘Would that grandson of yours like to have a look through a box of books my gang have outgrown to see if he'd like any of them?’ Or something much more involved, being careful not to upset the sensibililties of someone who may perceive the offer as an insulting handout or an attempt to fleece them for some cash. Either way, it is the code layer which is the most difficult of the three to see and understand, even when giving away a bunch of Dr Seuss books.

Since the middle code layer is the most complex to understand, primarily because it is beyond most people's experience, it should also be the layer that we are most careful about tampering with. Arguably, it is easier to see whether changes at the content or physical layers are effective or operating according to the stated intentions for putting those changes in place. When controls are introduced at the code layer, it is more difficult for us to determine whether they are effective or whether they are generating unintended consequences which need to be addressed.

Take the box of old books as an example. Suppose my neighbour phones me at the office to say that her grandson is coming round at the weekend. I've asked my secretary to hold all calls for a couple of hours because I'm working on this unit and need to do it without interruption. I've told my neighbour she can call at any time and now she feels insulted because I'm refusing to take her phone call without knowing, at that point, that she is trying to contact me. I've introduced a temporary control at the code (social protocol) layer, which has the unintended consequence of upsetting my neighbour. When I get the message that she called, I may or may not find out that I have upset her when I talk to her again. I may just be puzzled because she is slightly frosty towards me. It could get even more complicated if my secretary happens to be taken ill shortly after speaking to my neighbour. I don't get the message that the neighbour has called. Unaware that she has tried to contact me I then don't make an effort to get in touch promptly in return, adding insult to injury. The unintended consequences come from introducing an apparently simple control to the code layer.

In the case of the internet, the middle code layer and a large part of the original top content layer were free. This was sufficient to facilitate massive innovation. The ‘end-to-end’ rules of the code layer and the sharing of knowledge at the content layer are at the heart of the ‘innovation commons’ that Lessig repeatedly tells us about. It is not necessary for all the layers to be free, or part of a commons, in order to allow for innovation. It is the mixing of free and controlled resources that allows innovation to thrive.

When you are face to face with a difficulty, you are up against a discovery.

(Lord Kelvin)

3.9 The three-layers model of the internet

3.10 Phone networks, monopolies and allowing innovation