Way back in the olden days, folks decided that cities should invest lots of money in public transportation systems. The reasons were many fold, including reducing the number of individual vehicles being driven in too, and parked in, congested “downtown” areas, and increasing traffic to businesses in those areas, increasing their commercial viability. Many of these systems are sold to the public with the idea that they will (at least) break even against capital and operational expenses over time, but the reality is far different. This somewhat recent chart, for instance, is for a number of public transit systems in the US:

One view of the data isn’t always useful, of course, so here are two more: CityLab and Brookings.

What does this have to do with networks? Over at APNIC, Geoff Huston has a post up about the death of transit providers. He posits that as content providers move the content users want closer to the edge, the need for transit is decreased. Larger content providers can simply provide their own transit more cheaply than most transit providers, and smaller players can simply hang themselves off one of these larger cloud providers to get their content “out there.”

To tie the two situations together, public transport and transit providers lose money for the same reason—people are willing to pay for what they see and enjoy, not for the means required to get to it. Hence toll roads are a hated part of life, regardless of the access they provide to services, while no-one seems to mind paying a much steeper price at the pump in the form of taxes on fuel. Just as public transportation isn’t a compelling “sell” for most people, transit isn’t a compelling “sell.” People want to watch funny cat videos; they don’t much care about how those cat videos are delivered. There are, of course, several other problems interacting with the disappearance of transit into the inner workings of the web browser and app.

First, content providers have a much stronger business model. Most content providers take information they get for free (cat videos), and marshal it into a major lever to gather information about individual users that can then be sold. There is cost in producing content to sell, of course—the cost of doing the data mining that adds value to information about individuals, the cost of storing the data, etc.—but the cost to income ratio is far different when you can make money off the content.

Second, content providers seem to have somehow escaped much of the brunt of “network neutrality” rulings. For instance, the recent ruling that US Providers must allow consumers to “opt out” of their information gathering doesn’t seem to apply to content providers, just transit and the edge. There is validity to the way this is playing out—imagine being told the cost of your bus fare is lower because the bus company is running sensors on the bus that monitor your vital signs to sell to local doctors, and you get the general gist of the problem set—but this doesn’t make the problem any less acute for transit and edge providers.

The final result Geoff posits is a set of “cones of content,” where each content provider (and/or content caching network—a potential bright spot) pushes information as close to the edge as possible, and then some set of “last mile” players who provide the remaining illusion of a “public Internet.” What would be the impact of such a configuration, as opposed to the way the Internet works now?

First, there is no particular reason the logic that’s driving transit to the brink, should change for the last mile. Imagine a service hosted by one of the major content providers that offers low (or no) cost service to your home, using their interface, with primary access to their services, and with full data analytics of the kind the current last mile and transit providers are not allowed to perform. There have already been trial balloons in this area, and there are certain to be more in the future. If the Internet were still built out of transit, content, and edge providers, this sort of service might not be much of a threat. There would still be an “Internet” customers of such a service could reach. If the transit providers do, in fact, die, however—it suddenly seems imperative the last mile not fall into the hands of content providers, lest we return to the days of Compuserve and AOL (remember the golden days of only being able to send email through a services gateway that couldn’t transfer some sorts of traffic?).

Second, in such a world, competition and startups would be difficult, if not impossible. It would be the effective end of the tech boom in terms of networking, as there would be almost no way to start a competing service of any kind. We could pretty much call this scenario “silicon valley eats itself.”

Third, it has the potential to split society in ways we don’t see today. Not only would political/religious/social opponents simply not listen to on another, they would be on separate networks entirely, unable to even speak at all. Joining a network in such a fragmented world wouldn’t be just gaining access to a set of applications and services, it would be joining a worldview entire.

We could, of course, hope for the best. As Geoff says, maybe this is a situation that will never come to pass. Perhaps the “cones of content” will never reach the last mile, or perhaps the transit providers and IX’s will find some new lease on life, preventing the most extreme situation from coming about.

On the other hand, maybe those of us who live in the networking world need to at least pay attention to this situation, and think through what steps we can take to prevent such a fragmentation from taking place. The death of the transit provider might seem like a small blip in the day-to-day life of the average network engineer—but in the world of tech, small blips can sometimes make big impressions.