Skip navigation


At any given moment, the Internet as we know it is poised on the edge of extinction.

There are at least two ways to understand that sentence. One is pretty gloomy, suggesting a fragile Internet that must continually be rescued from doom. The other way to look at it is that the Internet is always in a state of evolution, and it’s our understanding of it that is challenged to keep up. I tend to prefer the latter perspective, and think it’s important to keep the Internet open for innovation.

At the same time, change can be scary — if it leads to an outcome that impacts us badly, from which we cannot recover, for example. That’s at least one reason why discussion of policy requirements and changing the Internet can be pretty tense.

If we want a dialog about the Internet that is as global as the network itself, we need to know how to talk about change:

  • what are the useful points of reference (hint: they aren’t particular technologies), and
  • how can we frame a successful dialog?

A couple of weeks ago, I was invited to participate in a workshop of the Global Commission on Internet Governance’s Research Advisory Network, to talk about the paper I wrote, “On the Nature of the Internet”. That paper looks at major policy issues through the lens of the unchanging or “invariant” properties of the Internet. What follows is a summary of the points I made in the panel discussion on “Fragmentation and Universality”, and some insightful follow up questions and discussion.

The purpose of the invariants is to step back from the specific thing that was built so we can understand what about it matters, even as there is change all around us. That is, rather than clinging to specific aspects as mysterious and inaccessible truths that were established by the original builders of the Internet, which set up some sort of exclusive club of technologists who can talk about what’s right. Technologists — all of them, not just the ones who built it — have the training to understand and create new directions in the technology, but that’s not the whole story when it comes to talking about the Internet as the foundation for modern communication and daily life.

For the purposes of the workshop discussion, I opted to focus on three of the invariants:

Global reach, integrity: There is one Internet. Its purpose has always been about connecting disparate resources, so universality is key. At the same time, it is a network of networks, which allows for individual control and specialization of networks that join together in common purpose. You can say that there are “two networks” today with IPv4 and IPv6 addressed networks interacting through gateways, but the reality is that’s not a stable configuration. If we want to continue to have global reach and integrity, the march must continue to IPv6 deployment globally.

Collaboration: We are all in this together! We can’t each do our own thing and hope the best works out. In fact, there is clear evidence of (competitive) industry interests still being able to come together to solve problems: World IPv6 Day and World IPv6 Launch were made real by disparate group of industry
participants coming together to do some work that none of them could do on their own.

No permanent favourites: no Internet company is “too big to fail”, no single technology defines the Internet. Not even the Internet Protocol (IP — v4 or v6) is the defining characteristic of the Internet. There were precursor technologies, and there could be others that follow — as long as they support a network that demonstrates these invariant properties.

How can we talk about change, and concerns of breaking the Internet up into unusable, unrecognizable pieces (fragmenting)? Borrowing from the Internet’s own lessons, I argue that collaboration is key — collaboration is the anti-fragmentation tool.

For successful collaboration, work is needed to take a discussion among many people and achieve a useful outcome. As we increasingly look to have “multistakeholder” processes that work, moving on from the format of multilateral negotiations, it seems some principles for successful collaborative activities might be useful. Here are some to consider:

Unified purpose of participants — getting everyone to the same
side of the table. The vision for the outcome will necessarily differ at the outset, but the desire for an outcome has to be consistent. And, there has to be some shared sense of “good”. For the IETF, good is “engineering excellence”.

Committed contribution — along the lines of it not being a matter of headcount, people involved have to be just that: involved. They make (skillful) contributions to move the discussion forward towards the desired outcome. Distractions are discouraged. Exploring alternatives is one thing, but willful distraction is not acceptable.

No single player can be unduly dominant — it’s important to listen and actually hear everyone, without being deafened by anyone (individual, corporation, interest, government). Having an overly dominant participant can “suck the air out of the room”, leaving others unable and/or uninterested to try to engage fully.

Successful outcome is based on measuring something other
than self
— it’s not even win-win, it’s about getting the
job done. This is not just about negotiations or jockeying for
individual gain. Rather, the collaboration is about mutually moving to
an optimal end state. Admittedly, “optimal” is in the sense of “least bad” sometimes, and it’s measured across the system, not the
participants. Of preference, this is measured in the light of the
“greater good” — i.e., for the good of the Internet or the
world, or whatever

That pretty much concluded my formal remarks at the workshop, but I enjoyed some follow up discussion. Workshop organizer Laura DeNardis asked me how I thought the “no permanent favourites” invariant is faring in today’s world? And, is it not a little reflexive to say that the invariants are the things that should be held fixed about the Internet? doesn’t that make them permanent favourites? Zinngggg! 🙂

First, I do believe that “no permanent favourites” holds true today, even if it may not be obvious that some of the current industry titans could be challenged. We technologists grumble about how hard it is to make changes in the underlying layers of the Internet (things seem to be stagnating). But, the point of the invariant is that we should not support impediments to change, nor should we limit plans for the future based on the tools of today. The timescales for change may be years and decades.

The original formulation of the invariant property (deemed too “geeky”) was that the Internet was about “survival of the population, not the individual”. That is, it seeks to ensure that there continues to be a network that works, globally, whether or not that forces change along the way.

Later, Eric Jardine asked “what about competition? There are times when entities are driven to new lengths because of competition and it
winds up benefiting more people (than collaboration alone)”. That’s a really fair point — indeed, the whole commercial development of the Internet is a demonstration of that. I have to say that, in my conceptualization of collaboration, the important thing is for individuals to bring their skills and qualities to bear on the issue at hand, not to simply blend everyone into one form.

Hearing this, Olaf Kolkman mentioned a phrasing he had used as far back as an OECD meeting in 2008: “Competition where possible, collaboration where necessary”. And that does seem to sum it up nicely.

One Comment

    • william manning
    • Posted March 18, 2016 at 17:24
    • Permalink

    a facet that might have been missed was the balance of simplicity v. complexity. this has been and is likely to remain as a driver in the evolution of interconnection. this is why it is rare to see duplicates in the same space, v4/v6, ospf/isis, ip/osi… one usually drives out the other. another facet is utility. the “internet” of the last 30 years provided interconnection with systems of roughly the same technology “foot print”. With the advent of IoT, we are deploying technologies with radically different footprints, so it is not clear that “one size” will still fit all. We may see a refocus on common interconnection (ALGs) between divergent transports and less on trying to hammer a single communications protocol into every device.
    Not a popular POV, not one that the IG folks want to hear, but it seems to be an emergent property.

Comments are closed.