Joseph Scott

Wednesday’s New York Times article – Google and Verizon Near Deal on Web Pay Tiers – claiming that Google is in talks with Verizon for preferred treatment of network traffic destined for Google servers – made folks very upset. In a nut shell, anti-net neutrality, which is a big deal given the size of the two companies involved. This caused the halt of private FCC meetings about regulation and both Google (denial) and Verizon (denial) have denied that there are in talks for such a deal. They did confirm though that they are talking, and that they have been talking for some time. This has gathered plenty of attention.

Eric Schmidt’s comment to CNBC may clear this up a bit:

Schmidt clarified that the net neutrality he advocates is not a neutrality between different types of content, but between the same type of content. He wants to make sure that there’s no discrimination between one video download over another.

If this is correct the neutrality level would only be at the content type level, not at the lower network layer. I suppose this could be considered slightly better than anti-net neutrality, but not by much. Large network providers are notoriously slow at adapting to change. Have you tried getting an IPv6 address for your home connection? Or even a co-located server for that matter? Yeah, not as fun as it sounds.

While advocating that all video content (for example) be treated the same may sound nice, what happens when someone comes up with a hot new video delivery method? Are all the big networks going to update all of their filter rules right away to detect this new type of video packet? I think it is more likely to see rain on the moon than for all of the big network providers respond quickly to such a change. As a result any development around video would self limit itself to make sure it matched the existing packet patterns for video to make sure they get the same treatment by Internet routers. Could this have a stifling impact on innovation? You can count on it.

There is something else that Google and Verizon could be talking about in this area as well; large-scale network peering (more info on Internet peering here and here). This could provide a similar benefit (increased speed/performance for traffic to Google from Verizon) by reducing the number of routers between Verizon customers and Google servers. For instance between my home DSL connection from Qwest and www.google.com there are roughly 14 routers (according to traceroute). Of those 5 are operated by Qwest, 4 by Level3, and 5 by Google (confirmed by whois lookup for each IP address in the traceroute results). A peering arrangement between Qwest and Google would likely eliminate all 4 of the Level3 run routers and perhaps one or two more between Qwest and Google. Removing 30% of the routers and a third-party entirely would likely result in better network performance between my Qwest DSL connection and Google servers. Seems reasonable that Google and Verizon would be in high level talks to discuss peering arrangements.

There is little doubt that Google already has many peering arrangements with other network carriers, given the massive popularity of Google services such peering arrangements would be mutually beneficial. Perhaps Verizon has been holding out? Maybe they want Google to pay them for the privilege? That would go against the traditional no pay arrangement for peering, but I don’t think that would stop Verizon from asking anyway.

In the end all of this is highly speculative, including the original New York Times article. It could range from nearly 100% spot on, to being no where near the mark. Until there is some sort of definitive declaration from Google and Verizon we won’t really know what is going on. That may be the biggest point out of all this, if Google and Verizon are talking net neutrality issues then they need to be out in the open about it, the impact of such a discussion go way beyond just these two companies.

For now Google says they are “committed to an open internet”. Given Eric Schmidt’s comments on content level neutrality I don’t think this is good enough, definitions of open come and go.