The Long View: Downsizing Is Easy; Government Is Hard

When I turned 18, I registered as a libertarian. Then I found out how weird most libertarians are, and I switched back to mainstream American politics. Joking aside, a big philosophical difference that I have with libertarianism is the idea that government, as such, is evil. This is an idea with a long pedigree in American politics, and lots of examples. The kind of governments that really took this idea to heart are known mostly for the poverty of their citizens and the corruption of their civil servants.

A libertarian critique of the wastefulness and inefficiency of government is often on point. The failure of the FAA Advanced Automation System that John discusses here is an excellent example. The challenge is that libertarian solutions have been tried, and found wanting. Privatizing the FAA, or at least it's air traffic control function, probably would work. But there isn't any reason to think that would have made the development of satellite based air traffic control any faster. 

Downsizing Is Easy; Government Is Hard


Why We Failed to Upgrade the Air Traffic Control System

In this campaign year of 1996, the airwaves and the newspapers are full of criticisms of big government and promises to return power to the people and to the states. Since the downfall of the Nixon Administration, American politics has been increasingly dominated by the theme that government itself is an evil. Responding simply to the polls, politicians echo what people say about the corruption and incompetence of public officials, even while seeking to occupy public offices themselves. Almost the only changes they can bring themselves to recommend are to make the organs of civil government smaller, more cautious, less able to affect the lives of ordinary citizens. Apparently, the chief duty of public officials is to restrain the ferocious ambitions of government bureaus, especially on the federal level.

This state of mind is lunacy. The chief responsibility of public officials is to govern. Government is hard. It requires the full attention of very smart people just to maintain those functions of the state that are necessary for civilized life. When officials are concentrating on something else, whether their own careers or some ideological nostrum, then the streets are not fixed, kids are not taught to read, and the country loses control of its borders. On the federal level in recent years, the textbook case of disaster-through- inattention was the savings-and-loan scandal of the 1980s. The cause was simple. The Federal Savings and Loan Insurance Corporation continued to insure the deposits at savings-and-loans as it always had, but it stopped doing serious inspections of the savings- and-loans to see if they merited the insurance. All the FSLIC had to do was maintain the inspection standards it had maintained for fifty years. The minds of the agency and of Congress were elsewhere, however, and the inevitable happened.

Now it seems that we have another scandal fit to set beside S&L Hell. The effort to create a new air traffic control system, called the Advanced Automation System, lasted 15 years and was originally supposed to be completed in 1990 at a cost of $3.6 billion. It was cancelled two years ago. It wasted half-a-billion dollars without improving the existing system. Now the FAA is working on a more modest upgrade for only $898 million, which it hopes to begin in 1998. This project is supposed to be on schedule. It, too, is an outrage. The problem with planning to simply upgrade the existing air traffic control system is that the one we have now is fundamentally obsolete. The matter goes beyond the old hardware and software. The system is based on controllers passing off aircraft from one imaginary point to another, following a system of radio beacons that goes back to 1941. Today, of course, when you can tell where you are to within a few meters by using the satellite technology of the Global Positioning System, there is no need for this clumsy relay. The upgrade now in the works makes no provision for this technology, though it might be adapted to use it in the future.

How did this Third World humiliation come about? Most of the information for this piece comes from a story by Matthew Wald that appeared on the front page of The New York Times on January 29, 1996. He proposes various causes, from the Reagan Administration's profligate ways with research and development money in the era of Stars Wars to the FAA's unrealistic safety requirements. However, it seems to me that the most important cause is more basic and more frightening. It increasingly menaces the legitimacy of our public institutions. The problem is that our elected officials have lost the ability to discern the core functions of government, the ones that must be maintained at all costs. They have lost the ability to discern these vital functions because the people who elect them have lost this ability, too.

The impetus to begin the overhaul of the air traffic control system was the air controllers strike of 1981. The controllers themselves were striking in part because even then they knew the system was inadequate to the demands placed upon it. In any event, the strike converted the FAA to the need for a more highly automated system, one that would be less vulnerable to walkouts. Most important of all, the explosion of air travel that accompanied deregulation clearly required a fundamental upgrade in the system.

Now think about this. The basic predicate of everything else the FAA might hope to accomplish is reliable air traffic control. There is no function, none at all, more central to the FAA's mission. However, having decided in 1982 to build an improved system, the FAA still managed to take until 1988 to choose a contractor for the software (the heart of the project) and to write the specifications. After all that effort, they chose IBM. This was like pouring over a menu at an ice cream parlor for fifteen minutes and then ordering vanilla.

You might think that control of the FAA would rank in importance with the chairmanship of the Federal Reserve or the directorship of the FBI in terms of the attention it receives from Congress and the President. It is a job that deals with life and death issues in an industry that just about everybody has contact with at least a few times a year. This sort of job needs incumbents who serve for long terms so they can attend to long-term policy. This is not the case with the Administrator of the FAA. Since the Administrator who began the Advanced Automation System project left office in 1984 until the current Administrator, David Hinson, killed it two years ago, the office has had three other occupants. It was also vacant for twenty months. This is the sort of indifference that make-weight jobs like Surgeon General get.

Political attention at the level of long-term policy is necessary for an important organ of government, because it lets the bureaucracy know what is politically possible. The staff at the FAA has not had this advantage, to the cost of us all. The original plan of the Advanced Automation System was to consolidate the controllers in its 250 Terminal Radar Control facilities, called "tracons," into 20 major control centers. This would have required not just new hardware but a dramatically increased level of computing power, since the enhanced control centers would have far more work do. Long after the project was underway, the FAA realized that consolidation of this sort was as politically difficult as closing an Army base; no member of Congress likes to have a federal facility closed in the home district. So the FAA told IBM to forget about the consolidation. The current upgrade plans involve only the control centers. This decision, of course, had the incidental effect of making the years of work already done largely irrelevant.

We know that it is possible to do complicated research and development in a reasonable period of time and at a predictable cost. While IBM was working simply on its proposal for the Advanced Automation System, for instance, Apple Computer conceived, developed and marketed the Mac, operating system and all. In times past, the government too knew how to coax good research and development from contractors. A good example is the contract that the Army made with the Wright brothers in the early years of this century to develop a spotter plane. It had few specifications. The ones it did have were enforceable and made perfect sense, such as that the machine had to be operable by a person of normal intelligence. The contract came to under ten pages.

The contract between the Department of Translocation and IBM for the new air traffic control software, in contrast, ran to hundreds of pages. Not only was it the longest contract in the DOT's experience, it was the longest in IBM's. Now, as anyone knows who has ever dealt with a complicated set of rules, the more rules you have, the more likely you are to overlook the important ones. The FAA, for instance, wanted the software to be upgradable while the system was in use. This requirement ("continuous operation") was perhaps implicit in the specifications. Unfortunately, nowhere in the hundreds of pages did it ever actually state this requirement. IBM had been working on the project for years before the engineers became aware of the requirement, which meant that hundreds of millions of dollars of programming they had already done had to be redone. By the early 1990s, confusion like this was putting the project's completion date back three and a half months for every month that passed. A new system was receding into the infinite future.

The Advanced Automation System suffered from that peculiar form of dissociation from reality which seems to happen only when lawyers try to tell engineers what to do. According to an one engineer working with the FAA, some of the specifications in the contract may have violated the laws of physics. The FAA asked for a system that was 99.99999% reliable. This is the actual figure; it would have meant a system that malfunctioned only about three seconds a year. Every further increment of reliability, of course, is harder to achieve than the one before. When you are talking about five decimal places, you are talking about spending an immense amount of effort (and money) to achieve almost nothing. As it happened, nothing was what the programmers achieved.

A report by independent consultants, the CNA Corporation, also lays a large share of the blame on IBM itself. The project involved dozens of programmers, whose work IBM made little effort to coordinate. (Despite the length of the contract, no one seems to have given much thought to compensating IBM in such a way that it would have been in the company's interest to spot problems beforehand.) IBM has great strengths in the design both of hardware and software, but they are not necessarily the first company you would think of if you wanted software that would have to be used by exhausted people in stressful situations. The software engineers seem at every stage to have had little interest in making the control interfaces user friendly. The flight controllers wanted to be able to move images and information around on their screens with a recessed button. IBM wanted them to type code. This is a perfect example of what is known as "corporate culture."

The National Transportation Safety Board recently affirmed that the current air traffic control system is safe, but acknowledged that the system is increasingly prone to delay. The major airlines say these delays cost them around $5 billion a year. As we have seen, there are no plans to build an up-to-date system. The best we might hope for, a few years for now, is a system that is obsolete but not actually made of junk.

The problem with the air traffic control system is not big government, over-regulation or the need for privatization. One might be forgiven for suspecting that a privatized FAA would work about as well as the Post Office and Amtrak, both of which are run by semi- independent corporations. In fact, privatizing the FAA would be just the sort of phony solution that has come to substitute for competent administration since the 1970s. Sometimes in some situations we need smaller government. What we need now is ordinary good government.


This article originally appeared in the February 1996 issue of Business Travel Executive magazine. 

Copyright © 1996 by John J. Reilly

Why post old articles?

Who was John J. Reilly?

All of John's posts here

An archive of John's site