We need to put the lie to the just-so stories that carrier networks are "the Internet"
I wrote this in a response to a discussion thread on NNSquad claiming that the Internet depends on some bits being more special than others.
30-Aug-2009Version 2: 2023-03-18 10:09:41

I wrote this in response to a discussion thread on http://www.nsquad.org that claims that neutrality prohibits people from “differentiated services” on the assumption that such capabilities are necessary.

Very simply you can do whatever you want with differentiated services. If the market will bear it then fine. The only requirement is that you give the Internet a chance. It’s not about forcing neutral – it’s about neutral as a byproduct of best-efforts indifference to the meaning of the packets as per interpretations entirely outside the network. We are hobbled by the very idea of networks vs. infrastructure (http://frankston.com/?n=railroads).

I've already written about these issues in http://frankston.com/?n=IPMarkeyIII and http://frankston.com/?n=IPMarkeyIIIa so I don't want to repeat myself.

Carriers can provide all the differentiation and special treatment and whatever they want. It's just not the "best efforts" Internet. (As I read it) the bill doesn’t prohibit having a separate network for phone services or whatever . It's just that it's not "The Internet" by definition. The problem is that the "best efforts" has worked so well that those who want to have special networks that hold aside capacity to be resold at a higher price find they cannot afford own a network just so that they can maintain enough control to keep their promises. So they make up self-serving stories and sometimes even believe them.

The problem is that the carriers currently have enough control of chokepoints to obfuscate the issues, especially in the local communities where you are forced to buy from a provider. Thus they can continue to tell just-so stories about how the Internet works only because of them even if such stories make no sense.  They do not have enough control of the entire path to make such promises. They themselves rely on having sufficient capacity for statistics to work. This was true even in the old days of statistical multiplexing (AKA best-efforts internally). I did a reality check with Tom Evslin who offered the original backbone for VoIP traffic -- no QoS -- just capacity.

So at very least we need to force the carriers to give the Internet a chance. Their stories make no sense once you drill done on the particulars. Since each element of the system has no value out of context any assignment of cost and price to any aspect of the system out of context must, by definition, be an accounting that is not directly related to real costs or values. The purpose of accounting is to provide measures for a purpose – typically operating a corporation or for determining taxes to be paid.

In this case accounting practices are not only not aligned with societal needs they run directly counter to societal needs by depriving us of the opportunities that drive the economy. If we look at the connectivity infrastructure as a whole there is fixed cost and a variable cost as with any system (to a first approximation). We combine these into a single cost by amortizing the fixed costs of time and over the customer base (subscribers). And we get it tragically wrong in the case of telecom. We’ve never managed to get the benefits of having paid for the copper, we’ve created artificially barriers to using the zero-cost wireless capacity (the costs are created by forcing us to pay providers (http://frankston.com/?n=SpectrumDirt) and now we’re expected to foot the bill for yet another infrastructure made of glass. Yet even before applying Moore’s law the costs for the glass are $1K per home (at a loose approximation). If we amortize that over one year that’s on the order of a cable bill. And if we amortize it over all the homes we get a common infrastructure that serves all and we can cover it easily with the money we save by using that infrastructure for many purposes included saving money on healthcare and public safety and education. (As I’ve written in http://frankston.com/?n=IPTelecomCosts, http://frankston.com/?n=InternetDynamic). The effective costs are far lower if we consider glass an option but not a requirement and don’t confusing physical materials with concepts (http://frankston.com/?n=TelecomPrison).

We can then choose an accounting model which is only about the negligible variable costs thus achieve a zero marginal cost.

I’m trying to understand what is hard to understand. We’ve done the experiment – differentiated service models such as SLAs and the phone network have failed against a best efforts model to the point that many claiming to offer differentiated services ride atop best efforts. It’s like selling tap water in bottles at a premium – a fine business as long as you can get away with it. But imagine if we only had bottled water because people believed that it was better – could you imagine agriculture where you had to use bottled water to irrigate the fields. Absurd? No more absurd than being told we have to pay a premium merely to communicate because our local facilities are no longer available to the community and have to be leased back.

We also see the absurdities when we look at settlement costs among carriers as bits get routed halfway across the country and back to go six inches between Comcast and Verizon just to avoid settlement costs and to maintain multiple physical networks which then get stitched back into one logical network. If there were any reality to the accounting models such practices would not be viable – they only make seem to make sense because we don’t do the most basic reality checks.

Perhaps we need to get past the semantic loading of “networks” and “The Internet”. I like the term “the bit commons” which emphasis that there is a physical infrastructure that becomes “The Internet” due to how we use it. Copper, glass and fiber are just sitting there (http://frankston.com/?name=OurCFR). They become the Internet when we “light it up” and then do our own networking.

Those who blithely defend current carrier practices need to explain how it can work in practice across disparate networks with no ability to assure a relationship between the promises and physical networking elements or why we should or would want to pay a premium when we could get so much capacity at zero marginal cost? They need to provide more about their accounting model and the purposes it services. They need to answer for the cost to our health, safety and education by forcing a billing relationship merely to communicate. Why is using copper, fiber or glass different from sidewalks (http://frankston.com/?n=Sidewalks)?