Visit Your Local PBS Station PBS Home PBS Home Programs A-Z TV Schedules Watch Video Donate Shop PBS Search PBS
I, Cringely - The Survival of the Nerdiest with Robert X. Cringely
Search I,Cringely:

The Pulpit
The Pulpit

<< [ The Walt Within ]   |  Stupid Net Tricks  |   [ BYOB ] >>

Weekly Column

Stupid Net Tricks: Congress Thinks It Can Force Technical Changes on the Internet, but Congress Is Wrong

Status: [CLOSED]
By Robert X. Cringely

When this column began in the spring of 1997, the Telecommunications Act of 1996 was a new law that had changed how communication services were provided in the United States, reflecting the new possibilities of what was then popularly called the Information Super Highway. Nine years later, Congress is now in the throes of rewriting that same law to reflect new realities and lessons learned, with the result that there is a lot of jostling for political position. Telephone and cable TV companies find themselves, this time, on the same side of the discussion, pushing Quality of Service arguments they say are necessary for their own survival. Companies like Google and Yahoo, Amazon and eBay -- companies that exist solely in cyberspace as corporate citizens of these same networks -- say that what the telcos and cable companies propose will hurt their businesses and business in general. Some techies worry that a counter-reformation of sorts is taking place that will literally destroy the Internet, taking with it some or all of the gains that people have made through broader and easier communication.

Who is right?

I think they are all wrong, and here's why.

The 1996 law was mainly about deregulation, allowing, for example, cable TV companies to offer telephone service and telephone companies to offer TV. The idea was to spur competition and growth, and it worked. The law spawned a splitting of networks into national and international backbones, metropolitan area networks, and last mile connections to your household and mine. The only part most of us actually saw was the latter, and the connections back then were mainly dial-up and almost entirely provided by the telephone company, since cable modems were only just appearing and weren't standardized. What's different today and what is pushing Congress toward revising the law is something unanticipated in 1996 -- huge technical success. TCP/IP is king, and whatever services we can expect to use over whatever network 10 years from now will be IP-based. Circuit switching is out and packet-switching is in. Analog video networks are rapidly giving way to hybrid digital networks. The topological differentiators between this network and that are beginning to disappear, so companies are fighting over turf while they still can.

The telco/cable argument is a simple one and comes in two parts. Part One says that even if their networks are IP-based, they'll still be in either the voice or video content businesses (probably both), and ought to have a right to optimize their networks for those services. Customers have a right to hear Grandma's voice as clearly as ever, whether Uncle Bob is downloading pornography in the next room or not. Sister should see her "American Idol" blooper episode in glorious HDTV despite Uncle Bob, too. The companies argue for Quality of Service -- having the power to give packet priority to one type of service over another �- to achieve this goal. Part Two of the telco/cable argument says that, having spent the hundreds of billions of dollars required to make these services even possible, and presumably being allowed to apply some Quality of Service attributes, the companies ought to be able to recoup some of their costs through selling access to their new diamond lanes to others who are willing to pay. In other words, if you want to get your IP TV from someone other than Comcast or AT&T, then Comcast or AT&T would like to collect a surcharge from either you, the consumer, or from the provider of that pass-through video service, presumably a Google or a Yahoo.

The argument of most consumers and non-ISP Internet businesses is simple, too: Don't do it. They argue that the current Best Effort Internet is working just fine, thanks, and ought to be left alone. In order to keep things more or less the same, they argue that the new telecom law mandate network neutrality -- that ISPs not be allowed to favor one packet over another.

Techies look at this seeming political impasse and -- as is their habit -- seek a technical solution. How can we have acceptable network performance without favoring one packet over another? The most obvious answer is simple and -- to the nerdy mind -- elegant. All you need to have it both ways is massive excess bandwidth. Over-provision the network (turn two lanes of traffic into 16 or 600), and whether there are 10 or 100 or 1,000 cars on the road simply won't matter. But that solution isn't economically viable, retort the phone and cable companies, who are correct to a point, but are also giddy with the whole idea because this very nerdy (and very academic) solution is their single best argument FOR Quality of Service and packet prioritization. Even politicians can understand that, the telcos and cable companies decided, relentlessly pushing home the idea. And so the Amazons and Electronic Frontier Foundations began to sweat, worrying that the telcos will win and their digital freedom will be impugned. Next stop, they warn, is the end to the Internet as we know it. Next step is...Compuserve!

Those are the fundamental arguments as they are being cast for the simple minds of Congress. But the reality of the situation is quite different as too many lawyers push sloppy thinking that bolsters one side or another.

There are a lot of mistakes here, and one of the biggest has to do with a basic understanding of what Public Service means. The telcos and cable companies argue that they ought to get some special rights because what they do is a Public Service. But the true definition of Public Service as viewed by U.S. law is the right to BE served rather than the right to PROVIDE service. Over the last 130 years, where for the most part there was one phone company and one power company and one cable TV company going past your house, the right to be served and the right to provide service were effectively aligned, but they are no longer. I live in Charleston, SC, one of the most technologically backward cities in America, yet I can call my Mom right now using any one of half a dozen completely independent and parallel communication networks. So my right to phone service is no longer dependent on BellSouth's right to provide phone service. If BellSouth went out of business there would be no long-term disruption. The argument that any particular service provider ought to be given preferential rights in order to help them survive no longer has any real bearing on Public Service. That argument is bogus.

Bogus, too, is the idea that Congress is even in a position to mandate anything. Those 535 men and women can have some effect within the borders of the United States, but that is only one country with perhaps 20 percent of the world's Internet users. Readers from Qatar tell me that right now Skype won't work in their country, presumably because of exactly this sort of traffic-shaping argument. But what Qatar does isn't what the world does, and banning Skype probably hurts QATAR as a country more than it helps. Even the United States can't impose significant changes in technical standards on the overall Internet, having as far as I can see no inherent authority over the Internet Engineering Task Force. If Congress does try to impose such changes, then parity and any possible technical advantage are lost, and the United States goes from leading the Internet world to following. Congress, which may not understand this, probably wouldn't want it if they did understand it.

I asked Bob Kahn, the father of TCP/IP, and he made the point that the Internet is a Best Effort network and if you change that, well, you no longer have the Internet. Remember when Compuserve and AOL and other network providers existed in parallel, providing access to the Internet for their users but only through a kind of protocol translation gateway? Remember how terrible that was? Well, that is effectively what Congress is being asked to mandate.

This is lunacy.

And not only is it lunacy -- it is a repudiation of the very attributes of the Internet that have made it a success.

The Internet isn't entirely without Quality of Service components, but those components are vestiges of another era. TCP/IP was optimized for reliable data transfer, with acknowledgement and retransmission capabilities built in to ensure that -� no matter how long it took �- the data would eventually get through intact. So the Internet is already optimized for reliablility, just not for timeliness. Yet if we have complaints about TCP/IP, they often come down to just these QoS components, which I am not proposing we eliminate, but I AM proposing that we see as an example of something that in effect the telcos and cable companies are asking for MORE of. Not good.

Yet the Internet is used all the time for just the sort of communication services that it is supposedly not optimized to support. Where TCP was slow, UDP was faster. So too for ftp and http. The Internet is in constant flux and is improving just fine without the need for some overbearing control group to impose a new order. This very flexibility comes from the simple rules of the Best Effort network. The Internet is a stupid network that just schleps bits, and that is the major secret of its success. Where a telephone network might have engineered into it the capability for call waiting and caller ID, those functions can be programmed atop the Internet without requiring any changes in the underlying network. It's the applications, stupid, and Congress doesn't seem to understand that.

The scenarios that have been presented so far are a) either overbuild at huge expense or b) grant network providers the right to sell premium packet space. That's it. But what we ought to be doing is leaning on our real advantages and simply writing smarter applications that deal with the problems.

We have about 20 years of hard work that's been done by many people helping television to play over Best Effort networks. Mix some pipe-stuffing with a little Forward Error Correction, the Data Fountain network protocol, and multicast support, and it sure looks like cable TV to me. Why is nobody mentioning this?

If Congress wants to do something truly useful, they should force network providers to support multicast in every router. The capability is there already, just waiting to be turned on. Flicking that switch would do more to help multimedia applications (and to foster continued U.S. leadership in multimedia applications) than Congress can even imagine.

We have another 20 years of work in Voice-over-IP that now has VoIP phones offering better sound quality in less bandwidth per channel than was provisioned by the old circuit-switched network. That means MORE conversations in the same bandwidth, not less, and those conversations can take advantage of VoIP optional services beyond anything circuit switching could offer. Why is nobody mentioning this?

Why is nobody �- NOBODY -� mentioning Moore's Law in this discussion? The Internet is today 1,000 times the size it was in 1996, yet these discussions tend to view network growth as static. Can the Internet support HDTV or not? Comparing the bandwidth required for an HD signal with an SD signal and mapping that against historical and expected network growth, I'd say it is likely not to be a problem at all. Why is nobody mentioning this?

Congress is arguing about a problem that doesn't really exist.

All the telcos can really point to is Cisco and Lucent's ability to keep prices up on OC-192 line cards. China already sees that as a business opportunity, so I'd say that problem is well on the way to being solved no matter what Congress does.

All the backbone providers can say is that bandwidth prices have gone so low they can no longer operate at a profit.

So go out of business, then. And if you don't go out of business, explain to us why not.

These companies want protection yet offer little or no reason why they ought to be protected �- especially since the very protection they seek hurts the network they ostensibly see as the source of their future success.

The real question is whether the Telecommunications Act of 1996 even needs re-writing and that comes down to something Congress probably doesn't want to consider �- that they aren't all-powerful after all. The Internet has grown beyond Congress' ability to control it. So they can bow to special interests, change the law, and ultimately prove their own impotence. Or they can do nothing and fight again another day.

Comments from the Tribe

Status: [CLOSED] read all comments (0)