Fifty Feet Beats a Status Report

The Core Insight

Lockheed built the P-80 prototype in a rented circus tent beside a plastics factory, in 143 days against a 180-day deadline. The team was 23 borrowed design engineers and about 30 shop mechanics.

Skunk Works is Ben Rich's 1994 memoir of the Lockheed division he took over in January 1975. Under the flying stories sits an operating model, and Rich prints it whole: fourteen rules Kelly Johnson wrote in one night.

Most engineering organizations treat headcount as capacity. More people means more output, and more process to keep the people coordinated. Johnson treated headcount as the cost, and every comparison in this book is a ratio.

The U-2 was designed by about 50 engineers with their desks touching, and the prototype flew eight months later. The whole Blackbird design group was seventy-five, and Rich's propulsion staff on the hardest problem in the airplane was three. F-117 flight test, Lockheed and Air Force together, totaled 240 people. The B-2 at Palmdale carried more than 2,000 auditors, engineers and kibitzers producing a million sheets of paper a day.

Maintenance ran the same ratio. Two SR-71s in England took a 35-man Skunk Works crew, and two at Kadena took 600 Air Force personnel. At Mildenhall, Fred Carmody replaced 80 Air Force mechanics with 18, and turned eight consecutive aborted missions into eight consecutive successes.

The Framework

The fourteen rules are the spine, and each one blocks a failure mode rather than stating a value.

• The program manager holds complete control of technical, financial and operational decisions, which removes decision latency.
• Headcount gets restricted in an almost vicious manner, so coordination overhead never outgrows output.
• The drawing release system stays simple and flexible, which makes recovering schedule after a failure cheap.
• Reports stay at a minimum while important work is recorded thoroughly, so reporting never substitutes for work.
• Cost review happens monthly across spent, committed and projected, so no overrun arrives ninety days late.
• The contractor flight-tests his own product, and Johnson lost the F-16 competition rather than concede that rule.

Rule fourteen pays performance rather than headcount supervised. The main plant gave raises for supervising more people. Johnson gave the raise to the man who supervised least, because that meant he was doing more himself.

A ten-rule sheet handed to Rich on his first day in 1954 carries the physical constraint. Engineers work within a stone's throw of the airplane being built. Anyone who anticipates a delay reports it to Johnson in writing.

An unwritten fifteenth rule travelled by word of mouth: starve before doing business with the Navy. Rich broke it once, and Sea Shadow passed from a six-man team to 85 division bureaucrats and a 50-man Navy oversight team.

Key Ideas

The Designer Belongs on the Shop Floor

The spoken rule was that no engineer sits more than fifty feet from the assembly area. Johnson kept the U-2 team in a narrow hallway off the production floor. Birds dive-bombed the drafting tables, and he left the team there anyway.

Engineers went to the floor the moment blueprints were approved and lived with their designs through fabrication, assembly and test. Senior shop men were expected to confront a designer over a drawing that did not work.

Quality control was inverted. Elsewhere it reported to the head of the shop, and here it reported to Rich. Any worker on the floor, not only a supervisor, had the authority to reject a part.

The Blackbird came out of that arrangement in a day and a half. Rich unhinged the door between the aerodynamics room and the structures room and laid it across two desks. He tacked paper to it and had all ten specialists design the airplane together. That drawing became the A-12 and the SR-71.

Eighty Percent Is the Target

Rich's standard was eighty percent. The team never got a design a hundred percent right, they knew enough to nail eighty, and eighty was plenty good enough.

The Blackbird inlet is the worked example. Efficiency reached 70 percent in the first half year, and reaching 80 took fourteen more months. Rich's cost rule states the general case: the final 10 percent of striving toward perfection consumes 40 percent of total expenditure. Safety, quality and security are the only things that get a hundred percent.

Authority Beats Supervision

Rich took over on January 17, 1975 and set his terms that day. The department heads made most of the tough calls. He stated what he wanted, got out of the way, absorbed the heat from Lockheed management, and heard bad news first.

Alan Brown describes the result from underneath. Rich let people do their jobs with minimum interference, and he was never a second-guesser.

Paperwork carried the same ratio as headcount. A Skunk Works purchase order for vendor development ran three pages and drew a four-page reply with specifications. The same transaction at the main plant produced a 185-page purchase order, a 1,200-page proposal, and three further volumes.

The CIA broke rule thirteen in 1960, demanding a resident government engineer. Johnson conceded one man, Norm Nelson, who got a desk and a phone and no chair. Handed a 20 million dollar Air Force add-on for Blackbird wing tanks, Nelson calculated it extended range by eighty miles. Johnson returned the money that afternoon.

Secrecy Bought an Exemption From Bureaucracy

Skunk Works aircraft operated in secret eight to ten years before acknowledgement, so no adversary started on countermeasures. That was the stated purpose. The structural benefit ran larger.

General Larry Welch ran the Air Force side of the stealth fighter. The classification cut the bureaucracy out, and he met Perry monthly for decisions with no delays and no useless meetings. The Blackbird took two Air Force officers, three Pentagon officials and four congressional players. The B-2 had to clear every hoop at both ends.

The bill is printed too. Johnson urged Rich to cancel Have Blue rather than take the Top Secret stamp, warning that security raises costs 25 percent. Coffee mugs bearing the aircraft nose were classified and locked in a safe between breaks. About 40 auditors ended up living inside the plant, and an F-117 progress payment was cut 30 percent over a document-logging lapse.

A Soviet Paper Made Stealth Computable

Denys Overholser, 36, found the key in a forty-page paper by Pyotr Ufimtsev, chief scientist at the Moscow Institute of Radio Engineering. The paper was published in Moscow nine years earlier and translated by the Air Force.

Ufimtsev extended Maxwell far enough to calculate radar return across a surface and at its edge, then sum the two. Computers in 1975 did not handle three dimensions or curves, which is why the airplane came out faceted. Break the shape into thousands of flat triangles and add up their signatures.

The result underneath is counter-intuitive. With flat surfaces the energy returned is independent of target size. A small airplane, a bomber and an aircraft carrier of the same shape return identical cross sections.

Overholser asked six months to write the code. Rich gave him three. Overholser and his eighty-something mentor Bill Schroeder delivered Echo I in five weeks.

The design was presented in 1975, claiming a thousand times less visible than anything the division had built. In September a chamber test against the D-21 drone confirmed the factor of a thousand. In October Lockheed and Northrop each got 1,500,000 dollars and four months. At White Sands the following March, a 38-foot wooden model produced the lowest cross sections ever measured, ten times better than Northrop's.

Prototype on Borrowed Hardware

Have Blue was built to be thrown away, on wooden tooling discarded after use. The avionics were scavenged: actuators from the F-111, a flight control computer from the F-16, inertial navigation from the B-52. About 3 million dollars of Air Force equipment went into two airplanes.

Two airplanes flew for two years on 30 million dollars, against three or four times that for a normal advanced prototype. Rich went to Lockheed's board, mid-scandal and near bankruptcy, for the 10 million dollars the Air Force did not cover. Contract to first flight took twenty months.

One flight came back lit up on radar, visible from fifty miles. The cause was three screw heads standing less than an eighth of an inch proud.

The F-117 contract was signed on November 1, 1978, at 350 million dollars for five aircraft with no price-adjustment clause. Rich guaranteed the full-size machine matched the exact radar cross section of the 38-foot wooden model. Then OPEC raised prices more than 50 percent in 1979, inflation hit 16 percent, and the Air Force refused to renegotiate. He lost slightly more than 6 million dollars on the first five and made about 80 million dollars after two batches. The program brought in more than 6 billion dollars.

Production ran eight a year at 43 million dollars fly-away, on a 78 percent learning curve while building two airplanes every three months. The first squadron was combat-ready five years after go-ahead, against ten years for McDonnell Douglas to field twenty F-18s.

Desert Storm settled the bet. Thirty-seven F-117s flew from King Khalid, about 900 miles from Baghdad, from 0300 on January 17, 1991. The type was 2 percent of allied air assets and flew 1,271 missions, one percent of sorties. It accounted for 40 percent of all damaged targets at a 75 percent direct-hit rate, with zero losses.

One comparison carries more than the totals. A daylight package of 72 aircraft, including 14 attacking F-16s, scored no hits at the nuclear complex. Eight F-117s with two tankers destroyed three of the four reactors and damaged the fourth.

Practical Applications

Cap the team before you cap the scope, and give whoever runs it authority over money, technique and schedule. Half the rule produces the worst case, where a small team with no authority reports upward all day.

Put the designer within fifty feet of the build. In software that means the engineer who wrote the service owns the deploy, the pager and the customer complaint. Distance from consequence is what the rule measures.

Set the target at eighty percent and name the exceptions out loud. Safety, quality and security get a hundred. Everything else stops when it works, because the last 10 percent of perfection costs 40 percent of the money.

Build the disposable version on borrowed parts before committing to numbers. Have Blue answered one question with wooden tooling and other programs' avionics, on 30 million dollars.

Cut the reporting to a minimum and keep the record. A monthly cost review across spent, committed and projected to completion replaces most status documents. Books ninety days late are what turn a small overrun into a cancelled program.

Who This Is For

Engineering leaders running hard technical work under a deadline get the most from this book. Founders get the headcount discipline, the eighty percent target and the prototype economics. The dense chapters are two, three, nine, fourteen and fifteen.

Skip the pilot testimonial chapters and the Cold War recaps.

Read it as what it is. This is a memoir by the division's own leader, and the testimonials to Rich are printed inside Rich's own book. He assigns himself the vision on stealth, while Overholser found the paper, Schroeder wrote the math and Scherrer drew the shape. The Desert Storm statistics are Air Force-supplied and arrive without sourcing, in the chapter that vindicates his career bet.

The figures do not reconcile either. Have Blue is 28 million dollars estimated, 30 million funded, and 34 million per the test pilot. The rules are presented as the cause while the book documents their breach. Johnson broke rule eleven himself with a bank loan on the U-2. A payroll of 7,500 and forty resident auditors broke rules three and five, and the success continued anyway.

The conditions do not transfer whole. The customer was a government buying secret aircraft, and the classification that hid them also exempted the program from the bureaucracy governing everything else. A startup has no equivalent of eight years of adversary blindness, and no buyer who funds a demonstrator it cannot describe. What transfers is the headcount rule, the fifty feet, the delegated authority, the eighty percent target and the disposable prototype.

The Decision

The test runs on your own reporting and takes an hour. List every recurring report, review meeting and status document your team produced last month. Put each one next to the decision it changed.

Anything that changed no decision is a person paid to describe work instead of doing it. Rich's three-page purchase order carried the same transaction as the main plant's 185 pages, and the traceability that mattered sat inside the three.

Cut what changed nothing, and hand those hours back to the people who were writing them.

Then measure the distance between the person who designs and the person who builds. Johnson's number was fifty feet.