Tim Lee on net neutrality, spectrum policy, and software patents

by on September 7, 2010 · 2 comments

On the podcast this week, Timothy B. Lee, PhD candidate in computer science at Princeton University and fellow at Princeton’s Center for Information Technology Policy, discusses a variety of issues.  Lee parses new net neutrality nuances, addressing recent debate over prioritization of internet services.  He also discusses wireless spectrum policy, comparing and contrasting a strict property rights model to a commons one.  Lee concludes by weighing in on potential software patent reform, referencing Paul Allen’s wide-ranging patent-infringement lawsuits and the Oracle-Google tiff over Java patents.

Related Readings

Do check out the interview, and consider subscribing to the show on iTunes. Past guests have included Clay Shirky on cognitive surplus, Nick Carr on what the internet is doing to our brains, Gina Trapani and Anil Dash on crowdsourcing, James Grimmelman on online harassment and the Google Books case, Michael Geist on ACTA, Tom Hazlett on spectrum reform, and Tyler Cowen on just about everything.

So what are you waiting for? Subscribe!

  • SagatAdon

    You know, the very notion of net neutrality is so fundamental to free speech and the Constitution that I wonder why it's even debated. Sure piracy can eat into corporate profits, but spreading creative works isn't alway bad and can be pretty helpful to indie artist, e.g.: http://lawblog.legalmatch.com/2010/08/25/the-good-side-of-illegal-file-sharing-and-online-piracy/

  • http://bennett.com/blog Richard Bennett

    When y'all get into how the Internet works, at 11:00 or so, you kinda go off the rails. Let me see if I can help out.

    The Internet consists of 40,000 Autonomous Systems, which for practical purposes we can call “networks.” No two networks interconnect anywhere in the world without an agreement, either a peering agreement or a transit agreement. If any network A wants to send a packet to any network B, A and B need to have an agreement. If they don't, but each has an agreement with a network C, then they route through network C. If there is no network C that serves both A and B, then they need to find a pair of interconnecting networks such as a D that connects to A and an E that connects to B where C and D connect to each other, and so on.

    When you send an e-mail from your PC to some PC that's on another ISPs network, all the packets from your ISP to the other ISP will generally follow the same route, and all the packets in the other direction will follow the same route, but these two routes are not the same. This asymmetry is fundamental to Internet routing.

    You don't need to take my word for this, you can verify it yourself using the traceroute tool (called tracert on Windows.) Find a destination that you can ping, and run tracert to it. Wait an hour and do it again, wait a day and do it again, etc. You'll find that the packets are always following the same route. The Internet is not a free-for-all, it's a system that works according to commercial agreements and is very static. The only time you'll see a deviation in tracert is when a link is down or when you get inside a CDN that does load balancing. The Internet in general doesn't do load-balancing.

    In Europe, many networks interconnect with QoS agreements based on DiffServ or MPLS labels. Their interconnection agreements specify that this capability will be honored for a given level of traffic and for a given price. It's up to each network to police its outgoing high priority traffic to its agreement with the other network, which it does by policing the volume of high-priority packets it agrees to carry for its customers. Quota-based systems like this are common in the Internet.

    The only reason that QoS is not ubiquitous on the Internet today today is that the applications that need it are relatively new in Internet terms – telepresence is the prime example. These applications are generally supported by private wires, but that's not ideal for their users. Streaming video doesn't need inter-domain QoS because it's mainly done by CDNs who obtain QoS by locating content close to users. See: http://www.hightechforum.org/managed-services-are-cool/

Previous post:

Next post: