[freedomben]

Do you also extend this to allow for pay-for-what-you-use? I don't disagree with charging people more for how much they use, but a lot of net neutrality proponents lose their minds at the suggestion that somebody who uses 5 GB a month should pay less than the kid streaming 10 TB of torrents every month.

[charleslmunger]

I think the overlap between those is muddled for a few reasons:

1. Zero rating. Wikipedia+facebook is free, everything else counts towards the cap. This was a subject of major debate in India.

2. First party zero rating - YouTube and Vonage count towards the cap, your ISP's streaming video or VOIP service does not.

3. Negotiated zero rating. "binge on" and similar, which is basically vendor neutral but limits the type of content excluded from the cap.

4. Extremely small caps. If YouTube counts toward the cap and TV doesn't, you won't cancel your cable package. This gives cable+ISPs a big incentive to keep caps low.

I view #1 and #2 to be clear violations of NN, and #3 as borderline. #4 is just a consequence of ISP monopoly - but it's not NN.

[freedomben]

Good thoughts. That does get muddy. I wonder how long though until AWS Included service comes around that allows data in and out on partner networks to be uncounted for any AWS customers of business using that service? That's a hypothetical of course, but I could see it.

[romwell]

I agree with your comment, and propose the following test, which will yield the same conclusions.

Given a scenario, ask the following:

"If enforced, will this give the users a financial incentive to pick one information source on the Net over another to get the same content?"

If the answer is YES, then the scenario is a violation of net neutrality principles.

So for the 4 scenarios you listed, the answers would be YES, YES, LIKELY NOT, and NO*

(*only because we restrict the test to sources on the Net).

[charleslmunger]

For #3 that's not strictly true - while the policy doesn't discriminate by vendor, it does discriminate by content type and by packet sniffing capabilities. For example, unlimited video uses the same bandwidth as a VPN downloading at the same speed as the video's bitrate, and are equal from a network management perspective - but they get vastly different treatment towards your data cap.

[carry_bit]

Why is #2 a violation of NN, but #4 is not? Does it really matter if the video is transferred using IP on top of "internet" frequencies, or IP on top of "video" frequencies, or QAM on "video" frequencies?

[charleslmunger]

In #2 the service is delivered over IP and the same last mile wires as other internet traffic. It's not "the internet", but it competes directly with the internet for last mile bandwidth, just like a Netflix or YouTube cache installed on the ISP's network.

https://arstechnica.com/information-technology/2015/11/comca...

[carry_bit]

It directly competes for last mile bandwidth regardless--the frequencies on the coax can either be used for internet or a dedicated video service but not both.

Would you really require cable ISPs provide you an internet bandwidth of X and video bandwidth of Y instead of internet bandwidth of X+Y?

[charleslmunger]

There's shades of grey here - anything you lay in a fixed diameter conduit is competing for bandwidth in the sense that you could pull another wire.

Consider an alternate history, where TV was invented after the internet. ISPs propose to offer TV service (exempt from data caps of course) at the expense of last mile bandwidth, which is otherwise subject to caps. I think that would be recognized as a net neutrality violation.

It feels unreasonable because we know the cost of providing television service scales with the number of subscribers, not the usage per subscriber. Of course, that's also true for internet service outside peak hours - a fixed monthly cap does not match the unit costs of providing internet at all.

I think pragmatic regulation would exempt existing cases like this. We wouldn't permit a new case like it in the future, though.

[romwell]

It's because we see TV and Internet as different media entirely, even if they deliver the "same" content. That is outside of the scope of NN discussion currently. In the same manner, newspapers and TV are subject to different regulation even if both deliver the "same" news. It's hard to come up with rules which would apply for all media - even maxims like the 1st amendment aren't simple to implement.

In theory, the problem you refer to should be addressed by anti-monopoly laws (NN is a tool to address a different problem).

[maxfurman]

I think the difference is that customers pay for TV separately already, regardless of the transmission medium.

[thomastjeffery]

The problem with data caps is that it makes people afraid to use any new service that requires a significant amount of data, especially when they can't easily count the data that service will use, which is usually the case.

The other problem with data caps is that they aren't a good description of bandwidth use. Bandwidth isn't a bucket of water that runs out after a certain amount has been poured. Bandwidth is the maximum flow rate of pipes connected to an infinite reservoir. Capping transfer speeds makes sense, because there is only so much that can be delivered during the same period. Capping transfer amounts doesn't make sense, because that kid streaming 10TB of torrents can push most of her bandwidth usage to off hours like 3am on a Wednesday, rather than getting it all during peak hours like 8pm on a Friday.

[Toboe]

Do you mean net neutral pay-for-what-you-use?

I.e.

You can buy 5GB volume. Everything you do counts against it.

Or do you mean selective pay-for-what-you-use-and-your-ISP-doesn't-prefer?

I.e.

You can buy 5GB, the ISP chooses what does and doesn't count against that.

From NN point of view: The former is fine. The latter isn't. (The whole king-maker problems)