Substrate neutrality
For a generation of intellectual property scholars, communications issues were peripheral. The telephone wires were humming along, doing their jobs — what really mattered were Hollywood's attempts to control formerly unregulated uses of content.
Now communications issues have come to be central. As we've discussed in the past, communications law is to this networked age as intellectual property law was to the information age and labor law was to the industrial age.
Intellectual property law is no longer the esoteric/arcane pursuit it was just 20 years ago. We've learned a great deal, and along the way we've thought from time to time that the choices made in the name of IP law would make or break the net. It turns out that IP law was just a proxy.
There are many different incumbents (law enforcement, content companies, telcos) who would like to see the internet made into something much more like a cell phone network. They have many strong arguments on their side, and the questions we're now confronting aren't easy.
It's my preliminary view, though, that the propertization and vertical integration of broadband networks is not a good idea, and that we should resist the new laws, new institutions, and new asymmetries of information now being sought in the name of internet “property” rights. We need a new politics for interactive networks, and an environmentally sound approach.
What's wrong with prioritization of packets? Other networks, like broadcast or newspaper businesses, have such control.
To understand what's wrong, we have to go back to the framing issues I've talked about here recently. The telcos and cablecos are used to running networks (1) optimized for a single service, (2) in which hardware and software are inextricably intertwined, and (3) in which you need permission to introduce a new element. When they say “internet” they mean infrastructure. They mean substrate. They say they built the substrate and now own it. They bring forward the romantic figure of the builder and say it would be un-American to resist his claims to ownership. They're looking to reinsert the friction/permission of the old networks they built before, by reattaching the medium to the message.
But when users say “internet” they mean relationships. We forget, because so many machines are involved, that the internet is a social world. Users don't think about transport — they're indifferent to the substrate. They care about what they do there. And what they do is create a complex adaptive system unlike any other communications network we've ever had before. The unpredictable ecology of the internet could never have been generated by a broadcaster or a newspaper. It's constantly revising itself in response to the feedback it's getting from everyone. And its value is almost wholly unrelated to the work carried out by the access valves, the gatekeepers to internet access. As I've said before, the internet is like an ocean, but formed through attention rather than nature. (And, just as we're almost totally ignorant of the life-forms beneath the waves, we don't know all that much about what's going on on the internet.) The essence of our relationship to this ecology, this complex adaptive system, is one of explanation/comprehension — at the most. We can't predict what it will do next.
The point about this ecology is that it is largely indifferent to the substrate it's carried on. The CD is not the song. The term “network neutrality” doesn't capture this — we should consider using “substrate neutrality” instead. Otherwise the network providers' arguments are so easy: “But it's our network!” they can say. (AT&T's new slogan is “Delivering your world,” as if the online experience was a visual pizza. We won't even need to rise from the couch.)
The network providers want vertical integration, and they're using strong economic arguments to get there. They need incentives to bring the US into the ranks of real broadband users; they want to internalize the benefits of the network, and argue that if they can they'll be better able to exploit the network efficiently. There's no competitor out there who is interested in providing unfettered internet access. So the existing broadband providers can comfortably call for new laws (blessing a two-tiered internet), new institutions (a greatly-expanded FCC), and new asymmetries of information (we don't really know what they're doing on their networks, and they're not telling us).
There are many parallels to the IP fights of the last decade, as I've pointed out in the past. We've got a romantic builder, where we used to have a romantic author. We're being told that the internet (like a developing nation) will remain a backwater unless we get these rules right. Indigenous content (fan fiction! flickr!) is being undervalued or ignored. As with the IP White Paper, we're being told that we need to fund this highway through the protection of property holders.
But the risks of propertizing the non-substrate parts of the internet are great. We may cut off the feedback loops for as-yet-unborn new online uses. The unpredictability of the internet may be profoundly affected, with the creativity that unpredictability fosters hurt. We may move the internet to an equilibrium state — a couch potato state. We'll never know what we've lost.
Of course, this is only a potential problem. We don't really know whether internet use will be degraded in some way, and we're hoping that these legislative changes are being sought to change what we presume is the status quo — unfettered access.
Should degradation happen, we'll need a more informed politics for the internet. We'll need to point out that economic modeling is likely to overlook what is crucial in favor of what is easier to describe (as Julie Cohen has said in the past). We'll need to address whether limits to self-interest are necessary — which is a hard question.
The first step will need to be to conceptually separate transport from internet, substrate from content, infrastructure from applications. This is where “substrate neutrality” comes in.
The second step, which is harder, will be to assert that the public interest in protecting the property interests now held in broadband pipes is not as important as protecting public access to the internet. The unpredictability, the unknown potential of the complex non-substrate internet is good for creativity and humans in ways we cannot now count up. We just can't do the tradeoff empirically between property and competition in this context. The non-substrate internet is incommensurable with the property interests now held in its pipes.
The new public interest here, the new regulatory role for the FCC, may be to set the conditions of access to this network.
There are many weaknesses in this argument. We need to study other countries and see how they've accomplished their broadband numbers. Perhaps internet access should be a public utility. We need to overcome the economic analyses of the incumbents, because they're claiming that they get clear answers from these analyses. There's an enormous amount of money at stake, and the real problem of maintaining high-speed networks that become utility services.
But I'm more and more persuaded that neoclassical economics doesn't capture the substantially non-market, dynamic elements that make up the internet. Reductionism just doesn't work here. We could, as with all environmental problems, try to do a conventional cost-benefit analysis — but we don't have any real numbers that would allow us to reach some optimum balance. We could say that this is all an institutional/policy analysis game, and look at it that way. Or we could make the hard choice of looking at what is good for everyone, as a whole, and privilege access.
It's now very fashionable to criticize the early cyber utopians. But I think they were right. The problem is that we now have an insufficient cultural commitment to the cooperation/creativity/interactivity vision of the net that they had — and so we don't have any economic (market) strings to attach to the operations of the large actors. That's why we may need government intervention for access. (But not for conditioning use of internet content in other architectural ways, in my view. I don't see the rationale for porting broadcast/cable “social policy” regulation onto the internet.)
I'm convinced that open-ended use of the internet will enhance rather than detract from the value of this network, and that extending property rights to effect vertically integrated control over this ecosystem would be destructive in the long run.
The telcos/cablecos have force — they have routers. But we have the power of ideas, and the early cyber guys had it right. Approaching this as an environmental problem (”access to the ocean”) can only help.
Comments
5 Responses to “Substrate neutrality”
Got something to say?
For individual routers: It adds additional cpu and memory requirements, along with additional code complexity. This increases router cost and reduces reliability. Further, if the router is the bottleneck, rather than the wire, it reduces traffic.
For the routing fabric: It increase routing protocol complexity, increasing the probability of misconfiguration. And adding delay to selected traffic classes may have difficult-to-predict effects, including lowering total goodput, and perhaps decreasing statistical multiplexing.
Correct. Even more problematic is that as a packet sails from source to destination, its 'priority' may or may not (probably not) get equal treatment through each network. I doubt that AT&T will give TimeWarner's 'prioritized' packets preference over their own 'prioritized' packets. After all, AT&T would have likely sold traffic prioritization to their own customers, and can't have their speedy delivery delayed by a competitor's high-priority packets.
Susan's substrate neutrality lies most in line with how the OSI layered model of networking is composed. Each layer is intended to be principally agnostic (if that term can be applied to a network layer) of the layers below and above. The layer where IP resides can have IP-V6 slide in to replace V4, and the layers above and below won't know the difference. Web browsers still fetch and display web sites, domain names still resolve, and so forth (Okay there more to it than that, but conceptually that's the idea). And, the users prinicipally don't need to know all the details of network their experiece rides on - just their experience.
At the same time, I get nervous when the policy debate superstructure gets too far removed from the nitty-gritty of the infrastructure engineering.
So, fwiw, here's a well-known vendor's perspective: Supporting Differentiated Service Classes in Large IP Networks, Juniper Networks, 2001.
I'm afraid this paper is somewhat dated, but in some of their more recent marketing literature, Juniper is still recommending this intro.
(& Btw, please note that I don't necessarily endorse the pov presented in this paper—I'm not selling equipment! It's just that I do think recognizing different perspectives is valuable.)
The paper seems aimed chiefly at people building private IP-based networks, although it does make mention of how unpredictable performance on the Internet will be, just so that people trying to use differenciated services don't scratch their heads when things don't speed up in a reliable way out on the Internet. They don't really advocate a tiered internet, just that prioritizing traffic can help reduce the need to add bandwidth.
I can't think of a scenerio other than lose-lose in trying to set up a tiered internet. Vendors will clash like never before as their 'fast-lane' services traffic slows down on competitors networks; end users lose by having their current service degraded as providers try in vain to speed up certain types of traffic. All your neighbors would get silghtly faster VOIP traffic if they pay for it, and your traffic slows down.
I'm not saying that selling prioritization is a bad thing in and of itself, just that it will result in very inconsistant results, even within a vendor's network. And, it seems to me that it will open up the door for tier 1 providors to lock out vendors like Skype or Vonage. And, it would effectively lock out innovation.
[…] is created through new laws, new institutions, and new asymmetries of information. (This trio has come around […]