Speech given by Jari Arkko at the ISOC panel on Internet Localization at the Freedom Online Coalition 2014 meeting in Tallinn, Estonia. Opening question: The Internet is a global network of networks that was not designed to recognise national borders. Do you see localisation proposals as a potential threat to an interoperable and global Internet, or do you rather see some of them - such as proposal to build more cables and IXPs - as opportunities to make the Internet more resilient? Jari Arkko: Both, really. There are some good things in localization that would be beneficial, but also some that do threaten the global nature of the Internet. But before I explain, let me go back to looking at the surveillance issues and what we can do about them in general. There are political and legal aspects of this, such as laws and transparency. We at the technical community have been mainly looking at it from the perspective of what can we do under the current situation. The IETF perspective, for instance, has been to think of surveillance just as any other attack on Internet traffic. We've launched efforts, such as fixing specific vulnerabilities, and making our next generation versions protocols as secure as possible. More generally, there has been a trend to turn on more security for Internet traffic in the world, and the surveillance discussions have accelerated this trend. So one technical answer to the surveillance issues is more communications security. Of course this helps only with communications, it does not help if you do not trust your peers in the communication. Trusting your cloud provider is what I think has fuelled some of the discussion on Internet localization. But the localization topic is difficult, because a lot of the value of the Internet is in its global nature. Innovations spread globally. The people who you have to communicate with are often spread around the world. Think of Metcalfe's law. The value of the network comes from our ability to connect with as many others as possible. I still remember the time from early 1990s when many large companies had internal mail systems. Today that would be unthinkable, because, obviously, we need to also communicate with our partners and customers. Many of which are global. The same with your friends in private life. But there are many good aspects in localization proposals that I've seen. In general, adding more Internet exchange points is very useful. It typically results in higher communication speeds, greater resiliency, and ensures that local content businesses can be easily reached by their customers. And helps ensure competition among various providers. The same with adding more communication paths, such as new sea cables. And having more choice in various services and cloud providers is also good, and many of those businesses today compete on how securely they can offer their services. But network localization is not a solution to all problems, either. I hope we agree at least that anything we do needs to be capable of global communications. The other thing is that network connectivity does not follow national borders. The Internet is a network of networks, built out of voluntary connections between parties. And it is a complex network, has many separate and overlaid parts. For instance, global operators or large enterprises with their networks. And the networks reacts to changes, such a problems in some connections, and re-routes traffic accordingly. And I fear that we are trusting the national Internet model far more than we should. If the communications are open, who is to say there isn't some party somewhere in the network that is leaking the information anyway? I would advocate building more exchange points and cables and setting up local cloud providers for all the right reasons, like faster communications. But for the rest, I would actually like to propose a different approach. First, increase awareness of where your data is. Second, select the parties that you trust your services and data to be at. Third, use encryption technology to protect the communications so that how it is routed does not matter.