Net Neutrality: who decides what traffic gets what priority?
The paper written by M. Yuksel, K. K. Ramakrishnan, S. Kalyanaraman, J. D. Houle, and R. Sadhvani: Value of Supporting Class-of-Service in IP Backbones brings mathematical rigor to a fairly intuitive observation. Put simply, it is the fact that undifferentiated networks (networks without packet priority schemes) require more available bandwidth to handle the high priority data in a timely fashion than a network that can separate classes of data into priority levels.
I don't think anyone would argue that allowing networking equipment to process e-mail with a lower priority than VOIP traffic is a bad trade-off, so long as the e-mail reaches the destination within a few minutes. E-mail is not expected to be a real time transfer while VOIP connections are.
The router here in the office has been tuned with SMTP and FTP at the lowest priority, VOIP and SSH connections at the highest priority and everything else in the middle. These settings work for me because I am most interested in fast response of packets in the high priority class while I'm more than willing to wait a few extra minutes if needed to get bulk transfers completed if that ensures they don't ruin the responsiveness of my high priority applications.
However, these decisions were made to optimize the connection performance according to this office's needs. Where the loss of net neutrality can harm the consumer of data services is when that decision is in the hands of the ISP. An ISP who has a VOIP product to sell would be highly motivated to reduce any competitors performance to ensure use of their own profit center product. We have already seen once instance of such a play via port blocking which was shot down by the FCC; other such plays in Guyana and Qatar are a common thread in Vonage's Forums.
So far these have been full port blocks; packet priorities would allow for more subtle acts that would make it far more difficult to point the finger at the ISP for the difficulties encountered. The ISPs claim that they need the freedom to open additional bandwidth above and beyond the existing speeds that would be dedicated to specific purposes. It is absurd to think that once some companies are offering high speed options for specific services that some ISPs won't see that borrowing the existing bandwidth for these services is cheaper than building out additional bandwidth.
More ominous is the rumbling that has been heard where these bandwidth providers want to be able to charge companies like Google for network access to the Internet cloud. While admitting that the consumer has paid for their connection, somehow the ISPs want to forget that a company like Google already pays enormous fees for their bandwidth as well. By being able to single out packets from a given source to ride on the low priority packet stream, the ISPs can effectively select the companies that consumers can reach with the expected performance, and those they can not. By avoiding port blocking and instead degrading those who won't pay the piper via packet priorities, they can be subtle in the punishment of those who rebel against them; potentially avoiding FCC intervention while forcing capitulation by the content providers.
So while this paper makes it clear that less bandwidth is required when packets can be prioritized, the underlying question of who does the prioritization is not impacted by this result. With most areas of the United States served only by handful of providers, the impulse to maximize profitable services sold could easily exceed any competitive threat of carrier switching, especially if the problems caused are simply the result of poor latency and bandwidth due to use of non-preferred software, protocols or servers.