I think a closer analogy would be: Should the power company be able to offer you a discount based on the brand of appliance you use. e.g. If you had a GE fridge in your house should you get a discount on your power usage (all other things being equal). So if you have a GE fridge you pay $0.10 per kilowatt, but a Samsung fridge will cost you $0.15

From my perspective, along the last mile a bit is a bit. The fact that they may have some back-bone infrastructure that optimizes delivery of traffic to an edge-node for youtube or Facebook doesn't really apply to me (e.g. it is not something I control). The fact that e.g. Comcast and Facebook can do a deal that monopolizes in their favor vs. other social networks/ISPs is not exactly to my benefit. I can't control that deal like I can control adding "a powerwall" in my house.

I really want the electricity analogy to be good, but the reality is that it isn't.

A bit is not a bit. Some bits actually are cheaper to deliver than others.

The major ISPs don't pay much for transit though. E.g. Comcast is is not tier 1, but with their peering arrangements they pay for less than 1 percent of their traffic.

The notorious Comcast/Netflix case was certainly not motivated by transit costs, since Netflix offers video content free to any ISP - see https://openconnect.netflix.com/

How so?

Because net neutrality already doesn't exist for ISPs: https://en.wikipedia.org/wiki/Internet_transit

Also, because of things like edge caches, there are bits that have to travel a shorter distance, and interact with less hardware, than other bits.

Your GE microwave doesn't require that it gets electricity from a GE power turbine. All watts are the same in that sense. The watts can come from a GE generator (in the town over) or a westinghouse generator (in the local town). It doesn't matter.

But a youtube application does require that the bits come from youtube servers. Youtube cannot use bits from vimeo, and because of network topology, the vimeo bits might be more costly for the ISP to deliver than the youtube bits.

Exactly. Think of the analogy with car brands on a street (looking at you, Tesla) and you have most peoples understanding.

Scalia gives this example in a 2005 minority opinion.[1]

If a Powerwall saves the utility money, then any discount flowing from that is just returning money to the customer. My watching a video on Facebook instead of YouTube does nothing like that.[2]

[1] https://www.theatlantic.com/technology/archive/2014/05/net-n...

[2] As an aside, neither do Powerwalls. Utilities sell commodities, so any savings must be realizable as reduced usage. E.g. if solar were cheaper but for storage, utilities could buy storage themselves.

In a way, I'd say what your describing as far as power utilities is already in effect. There are a good number of people that charge their batteries off peak when the power is cheaper such that they can either use it later on peak when they would have to pay more, or even sell that power back to the grid when it's in more demand.

I think the issue here is that power is binary as far as the argument here is concerned, you have it or you don't, but influencing consumer patterns on data usage has the potential to seriously influence how they use the internet in a way that paying differing amounts for power doesn't. Competition is one of the most powerful driving forces of the internet today. The idea a lone programmer can offer a service that compares to multinational companies (in some areas, obviously not others) is phenomenal. But say consumers or that developer suddenly have to pay a large ongoing entry cost due to the fact that the product that they have made and are competing with is high bandwidth. Innovation dies overnight without net neutrality.

> power is binary as far as the argument here is concerned, you have it or you don't

Power is not binary. It is fundamentally analog. You can certainly have different degrees of power. See brownouts. Consumers only tend to see power as binary because of deep standards penetration in the electronics market.

Don't get me wrong, I certainly understand that power is anything but binary in it's function, however, as far as how consumers interact with it, very very few care as to what the copper in their walls is doing, only that it turns on the lights.

To extend the argument of net neutrality in this direction, I think an important distinction needs to be made, because surely you could argue LED's over florescent bulbs due to power consumption, but consumers already pay for different speeds and data caps, the argument shifts to your utility offering different prices based on who is making your LED's, even if they are functionally identical. This is where the comparison breaks down, because the bulbs don't magically become more efficient for your utility because they are somehow peering with the manufacturer.

> as far as how consumers interact with it

Right. Because of deep penetration of market standards, a mix of government regulation and industry standards, consumers interact with power in this very limited fashion.

I don't believe users of bandwidth would want to behave in a similar fashion. Is it so crazy to imagine a dynamic bandwidth market, largely automated for pay per use? I worry about the impacts of such a thing, but it like any other has its benefits and plagues.

/To be honest, I have no idea what you mean by your extended metaphor...

The comparison to the power company is natural because the hardware infrastructure is similar, but the crucial difference is that power lines don't carry communications (for the most part). Energy is more easily made into a commodity because it can be quantified in natural physical units. For communications it's much harder. The definition of "services" like YouTube will be necessarily incomplete because its definition is socially-mediated to begin with. Or, if ISPs were to charge based not on services but instead go deeper, so e.g. ICMP and UDP might be cheaper than TCP, you still need to rely on some judgment made by someone out there. These necessarily incomplete definitions will create huge inefficiencies in the system as a whole, generate useless extra work, and so on.

It hobbles the system as a whole in order to provide the justification for price differentials which would drive profits for ISPs. This is basically a money grab on the part of the ISPs (natural monopolies).

> should your ISP be able to offer you a discounted rate to use services that offset their cost (youtube, which has an cache near you, vs vimeo, which doesn't)?

I think this could be reasonable as long as it's not by having YouTube do a deal with the ISP, but is by a company-neutral measure of how much costs are actually lowered for the ISP (which is passed onto you). If Vimeo can do some engineering to reduce cost to the ISP and automatically have the lowered cost passed onto you without Vimeo having to negotiate some special deal with the ISP, then that becomes another area they can compete and innovate on, and hence harnesses the power of competition rather than foster anticompetitive practices.

By contrast, if special deals are allowed ("YouTube traffic is free"), then it would be mutually beneficial for monopolies to offer ISPs kickbacks in return for helping them maintain their monopoly by suppressing competition, while the rest of us lose out.

Maybe a better example is paying more for electricity from 'green' sources like solar/wind. This is an option offered by PG&E right now.

Comparing to websites, you can have all the "common" websites that have local caches, peered with your ISP directly, for low rate X. If you want access to a niche video streaming startup that has to go through other peers we haven't optimized for, you pay more.

It's still not a perfect analogy, but it forms a similar comparison "Where do I get (content|electricty) from?"

I am of the opinion that not only is traffic shaping not inherently evil but is a valuable feature that can be used to improve network efficiency but I cannot trust Comcast and the others to favour innovation and technological improvement over short term profits.