Imagining the space in-between

In internet policy debates it has become increasingly commonplace to not only talk about how the internet can be governed, but also about how it governs. By that I mean the possibilities that the internet and the services that are built upon it open up – or fail to open up – for their users. There is increasing awareness of how applications and services such as social media platforms frame our interactions with each other. For instance, interfaces allow users to do certain things on platforms (e.g. share content) and not others (e.g. download content), while nudging mechanisms call our attention to some information and not other. – Proceeding from the understanding that human experiences get streamlined or “flattened” in such socio-technical settings a recent NGI panel at transmediale festival proposed to reclaim some autonomy by strengthening “the space in-between, the space made of intangible relationships and embodied experiences that constitute our social fabric” in a next generation internet.

In a time when everybody is talking about efficiency gains through automation, inquiring about such spaces in-between might seem superfluous. But actually, the idea is not that far fetched. Because while it is true that technical architectures can feel deterministic when they wall us in, they can also do the opposite. They can foster an openness that offers or even requires interaction between users and that prompts collaborative sense-making. One way to achieve this is to deliberately induce aspects of ambiguity or uncertainty within a system. This means designing a system in a way that it leaves users with choice and more interactive freedom.

One place where social interaction arises as a result of architectural uncertainty is the core of the internet. According to the layered architecture model of the internet as exemplified in the protocol stack, I am referring to the network layer of the internet as the core. It is where internet engineers – who call themselves networkers – manufacture internet connectivity. Connectivity is the ability to transfer data from any endpoint of the internet to any other. It is the basic resource of the digital society. Networkers have three central concerns: they interconnect the more than 60,000 networks that together constitute the internet today, they configure and maintain the connections, and occasionally they also shut them down. Contrary to what one might expect, even today – more than 20 years after the full commercialisation of the internet in 1995 – this job is all but a plug-and-play task. In fact, the network layer of the internet is a space that is full of complexity and personal interaction.

To get an impression of the spirit of what is often called the community of networkers, please tune into the video below for a minute or two. It was recorded at a conference in Leeds, hosted jointly by the internet exchanges IXLeeds, LINX and LoNAP, and I have been permitted to use it here courtesy of the European Internet Exchange Association EuroIX. I will subsequently explain some of the issues that the singers articulate in the clip.

Half way through the song we understand that there are some negative feelings about lawyers, and I will come back to that later. But first, let me clarify some basic terms. To shut or de-peer basically means to terminate an interconnection – that is something networkers typically dislike. They want to see traffic flow. Then they sing about peering. Two networks peer when they exchange traffic that is meant for each other’s customers directly and typically at no cost. It is a sophisticated form of barter trade. Networkers also want to share routes in peace. Routes are in simple terms like the streets of the internet, the paths along which internet traffic can be forwarded from an originating network to a receiving network.

The routing system of the internet is where the notion of uncertainty plays out. Because the reason that network operators need to share routing information with each other at all instead of just looking at some kind of a map or directory is that there is no such authoritative map of the internet. It just does not exist, and that is by design. None of the network maps you may have come across captures all the available paths on the internet. Nobody has an overview of the full internet. In the absence of a general map, networkers use an internet protocol to tell each other directly, which destinations they can reach for each other. This protocol is called BGP, the Border Gateway Protocol. Internet connectivity relies on this decentral information-sharing mechanism.

But the way BGP functions is disputed. It has a characteristic that some people think of as a feature while others think of it as a bug: it has no mechanism of validation. This means that networkers cannot know if the routing information that they receive from their counterparts is legitimate and correct. They have to take it at face value. In fact, the characteristics of BGP mean that the internet actually operates with a routing system that is prone to misuse and abuse. Nobody can keep a networker from signalling faulty information to his or her neighbours – be it by mistake or for nefarious purposes like surveillance. If incorrect routing information spreads from network to network across the internet, traffic gets sent to the wrong destinations. Internet engineers have to deal with this on a day-by-day basis. And they do so, among other things, by mobilising personal trust relationships with other networkers.

To tackle connectivity issues networkers regularly work together, even when they are competitors; after all, most networks are businesses that operate for profit. In order for the system to exist, in order to have a product, these competitors have to cooperate. Often times, cooperation takes the form of backchannel communication. For instance, when there is loss of connectivity due to natural disasters or unusual routing announcements, networkers will often times try to identify and interpret the problem together. Imagine a chat like the following:

I see Network X announcing routes that I think belong to Network Y. Do you see the same? – Yes, I see it too. – Is the announcement legit? No? – Does anybody know the engineer in charge at X and can call him to stop announcing these routes? – Yes, the engineer at network Z already sent him a message. – In the meantime, let’s all shut down our connections to that network to isolate it and prevent damage from the routing system!

Many networkers work together in this highly informal way all the time, mostly without internet users taking any note. It is like a social network that underpins the internet as a network of networks.

Now let’s go back to the apparent disdain for lawyers. Why is that? Lawyers and contracts stand for a formalisation that transforms interactions into transactions. This is at odds with the informal way in which many networkers like to arrange interconnections – even though there is a commercial value to these interconnections. A recent survey by the NGO Packet Clearing House has confirmed that even today, more than 99 percent of the peering relationships are based on handshake agreements. So what the networkers in the video bemoan is not the existence of lawyers, but an eventual formalisation of the practically multi-faceted, difficult-to-measure, techno-social relationships that they maintain with each other.

Coming back to the initial proposition, I have tried to point out by way of example that technical architectures do not necessarily shape action in the sense of limiting choices. In the case of BGP, they can also be a source of uncertainty that necessitates collaborative interpretation and sensemaking. This does not always have to lead to happy socialising as in the video above. But in the case of the community of internet engineers, it sometimes does. And there is evidence that many in this community act with a shared understanding that internet connectivity is a common good. This is a valuable thing.