An Observation On Net Neutrality

I was out on Los Angeles last Thursday and happened to tune in to KFPK during Beneath The Surface. They were talking about Net Neutrality with Sascha Meinrath, the founder of the Champagne-Urbana Wireless Network, and managing to totally make a hash of the discussion. That's when I realized that its pretty much impossible to talk about Net Neutrality with the lay public.

The problem is that the issue of Net Neutrality is inseparably bound up with the technical minutia of how traffic gets passed back and forth on modern networks. I presume that Mr. Meinrath, being the founder of a wireless network, has some familiarity with how they work, but during the on-air discussion he didn't even attempt to separate out the various threads. Rather, he approached Net Neutrality as a monolithic issue, completely ignoring some important technical distinctions.

The larger concept of Net Neutrality embraces (at least) three separate sub-issues:

• Anti-competative practices
• Quality of service guarantees
• Censorship

During the discussion these were all sort of smeared together even though they should be treated differently. I think its pretty non-controversial to say that service providers shouldn't engage in certain kinds of anti-competitive practice, for example de-prioritizing traffic originating from/going to a competitor's network. I see no benefit accruing to consumers from such a practice; if we're going to regulate how ISPs do their thing then we should focus on that sort of behavior.

Then there's censorship, the notion that ISPs might block particular sites or the use of particular applications. I didn't listen long enough to know if Mr. Meinrath addressed this issue specifically, but I can guess that he's against it. First, I'll point out that censorship is different from anti-competitive practice, even though both concepts fall under the greater umbrella of Net Neutrality, so the subjects should be treated individually. More importantly, "censorship" as it applies to an IP network isn't the same thing as censorship of media like books or TV. If I as an ISP decide to block inbound SMTP traffic to residential address ranges am I engaging in censorship or am I taking a reasonable precaution to reduce the spam caused by open relays? Its also possible to argue that censorship is a value-added product; some customers might pay to have their ISP block all access to porn sites, for instance. I think that is, in many ways, equivalent to requesting that your cable provider disable adult channels. Blocking all porn is a much more difficult to do well, and ISPs might be tempted to block things without being asked (probably bad), but such arguments don't undermine the fundamental premise that ISP-based content filtering is a service that consumers should be allowed to purchase.

Lastly there's Quality-of-Service; Mr. Meinrath seemed to take great umbrage with the notion that businesses could/should pay to get better quality service. He claimed that QoS amounts to double-charging business for their bandwidth. From a technical standpoint this is a gross misstatement; Mr. Meinrath may have glossed over it for the sake of general discussion, but in doing so I believe that he misrepresented the issue. Bandwidth refers to the size of the pipe to a particular business, but just because you have a 100-Mbps connection doesn't mean you're always going to get 100-Mbps throughput since your traffic has to compete with everyone else who shares your routers and switches. Guaranteed delivery/bandwidth is a separate service above and beyond the standard package; you've got to set up QoS policies in your routers etc. etc. etc. So, from a technical standpoint, ISPs are justified in charging additional money for a value-added service.

But, of course, to understand the arguments I've made above you've got to understand concepts like "de-prioritized traffic" and "open relay" and "best effort delivery" vs. "guaranteed delivery" and all sort of other network esoterica. Which is where conversation with the lay public breaks down.