I think the idea here is that a user could prioritize their Netflix traffic over their Dropbox traffic. It would be great actually if they offered QoS configuration to end users.
Consumer please select one:
[ X ] $99/mo - make Comcast-NBC-Universal content faster, 100 GB/day
[ ] $199/mo - make all content the same speed, 1 GB/day
Because this has worked so well in the past with cable. [/sarcasm]
Seriously, you can't buy a cable package without also paying for a bunch of bundled channels that nobody wants. Why would literally the same people behave differently just because it's internet instead of cable?
Based on the text quoted in the article, I think it's relevant to note that traffic can't be prioritized "based on compensation or lack thereof by the sender to the broadband Internet access service provider." I may be reading it wrong, but to me that says specifically that ISPs can't hold content providers hostage for payments, but says nothing about throttling end users' access to content.
Theoretically the goal is to allow some forms of desired prioritization without giving these natural monopolies a huge club to beat service providers up with.
We want to be able to do prioritization of packets. QoS is not inherently evil. A smarter network can fully serve more clients for the same $$$ of infra, by understanding and prioritizing packets which are more or less sensitive to latency and/or packet loss.
The problem is we simply can't trust the ISPs, and there's no choice / no competition to switch to when they misbehave. And as we can see, it's very difficult to write a law which only allows these "acceptable" forms of QoS.
For example, if a customer wants lower latency on their VPN tunnel used for 4K HD video conferencing from home to the office, is it OK for Comcast to offer a higher priced plan to the consumer to get that? I think we want to keep these options open, and saying it has to be an opt-in service and show up on the consumers bill is an interesting middle-ground.
The other part is ensuring the baseline service is actually providing what they say it is, and if you can't get a reliable 5Mbps netflix stream to work 95% of the time on your "50Mbps" plan then they are failing their baseline SLA and customers should be entitled to refunds.
> And as we can see, it's very difficult to write a law which only allows these "acceptable" forms of QoS.
The IP header has QoS bits. It's not hard to write a law that doesn't prohibit ISPs from honoring those bits, or even from charging customers for a given amount of bandwidth at a given level of service. As long as any service/application can tag their packets however they like, and as long as the ISP is treating all equivalently tagged packets equivalently, there is no network neutrality problem with that. And it provides a clear line.
Trying to put significantly more complicated logic than that into the middle of the network in order to try to recognize specific types of traffic is inherently defective anyway. The ISP would end up picking winners and losers based on what traffic it recognizes for prioritization. And it would inherently bias the future of the internet against innovations because new protocols would be unrecognized until they became sufficiently popular and would never become sufficiently popular if they can't be prioritized but have to compete against incumbent protocols that are.
That would not work. Most users don't care/don't know about about QoS, nor they want to learn. There was/is a ton of research on user-driven QoS, micro-payments, per-flow reservations, etc. etc. The result: there is no nice or simple scheme to a) expose QoS knobs to users b) align incentives of ISPS/ContentProviders/Users for micropayments, c) affordable equipment that can do per-user crediting across the Internet.
