Closing Keynote: Skunk Works
This is a modal window.
The media could not be loaded, either because the server or network failed or because the format is not supported.
Formal Metadata
Title |
| |
Title of Series | ||
Part Number | 2 | |
Number of Parts | 89 | |
Author | ||
License | CC Attribution - ShareAlike 3.0 Unported: You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal and non-commercial purpose as long as the work is attributed to the author in the manner specified by the author or licensor and the work or content is shared also in adapted form only under the conditions of this | |
Identifiers | 10.5446/31504 (DOI) | |
Publisher | ||
Release Date | ||
Language |
Content Metadata
Subject Area | ||
Genre | ||
Abstract |
|
RailsConf 20162 / 89
1
2
3
5
7
8
11
12
19
22
28
29
30
31
33
45
47
50
52
53
58
62
67
71
74
76
77
78
79
80
81
84
85
86
88
89
00:00
GodPerturbation theoryCrash (computing)NeuroinformatikLengthType theoryCompilation albumProcess (computing)Mobile appMereologyComputer animationLecture/Conference
01:07
Lattice (order)Crash (computing)Normal (geometry)Process (computing)RankingPlanningDegree (graph theory)Point (geometry)MereologyMatching (graph theory)Service (economics)Student's t-testProjective planeQuicksortForceFactory (trading post)Cartesian coordinate systemChainPower (physics)Revision controlLine (geometry)Event horizonForcing (mathematics)SpacetimeBoss CorporationInheritance (object-oriented programming)Multiplication signFormal grammarDifferentiable manifoldInteractive televisionRule of inferenceMechanism designDivision (mathematics)Bookmark (World Wide Web)Product (business)Dimensional analysisAreaBuildingHeegaard splittingVirtual machineStatisticsVery-high-bit-rate digital subscriber lineSign (mathematics)InformationSoftware developerPrototypeLevel (video gaming)Office suiteFile formatArithmetic meanInternetworkingSoftware bugLibrary (computing)Design by contractSlide ruleNear-ringFilm editingBitDemosceneSinc functionSoftwareAngleAdditionDistanceFirst-order logicLecture/Conference
10:30
Right angleData structureBuildingPlanningSquare numberProcess (computing)Operator (mathematics)StapeldateiMaterialization (paranormal)MereologyWeightMultiplication signComputer wormChannel capacitySet (mathematics)INTEGRALObservational studyComputer configurationEndliche ModelltheorieTheory of relativityLevel (video gaming)Device driverBitHacker (term)Configuration spaceDifferent (Kate Ryan album)Series (mathematics)Mechanism designImage resolutionExtreme programmingDegree (graph theory)ChainObject (grammar)NumberPressureExpected valueSuite (music)Row (database)Line (geometry)Service (economics)CalculationMach's principlePlane (geometry)Point (geometry)Charge carrierMatching (graph theory)Dependent and independent variablesParsingSystem callFigurate numberTurbulenceProjective planeInteractive televisionRevision controlOnline helpSoftwareUniverse (mathematics)Run time (program lifecycle phase)Electronic visual displayField (computer science)Computer-assisted translationFrame problemFamilyInstallation artOrder (biology)Chemical equationSingle-precision floating-point formatState of matterLecture/Conference
19:52
Extreme programmingOperator (mathematics)Structural loadPlanningIntegrated development environmentFamilyDegree (graph theory)PlastikkarteData structureOctahedronMultiplication signDisk read-and-write headSoftware testingDrop (liquid)TelecommunicationBookmark (World Wide Web)ChainRow (database)Device driverSatelliteConstructor (object-oriented programming)Point (geometry)Data compressionLevel (video gaming)Physical systemLine (geometry)Field (computer science)Graph coloring2 (number)VelocityImage resolutionCausalityRange (statistics)Adaptive behaviorStreaming mediaThumbnailAirfoilMetropolitan area networkEntire functionComputer wormSeries (mathematics)Term (mathematics)Supersonic speedOpen setInternetworkingGroup actionFlash memoryRiflingGame controllerSpring (hydrology)MassConcordance (publishing)Electronic visual displayLimit (category theory)CalculationMachine visionContext awarenessSound effectHacker (term)Office suiteMach's principleDecision tree learningCone penetration testLecture/Conference
29:14
Cartesian coordinate systemPlanningWaveEndliche ModelltheorieNeuroinformatikAnalytic continuationPrototypePhysicalismRange (statistics)State observerOperator (mathematics)Multiplication signoutputComplex (psychology)Food energyRule of inferenceWell-formed formulaProcess (computing)BuildingMathematicsPole (complex analysis)Electronic mailing listExpert systemInferenceDisk read-and-write headRight angleDifferent (Kate Ryan album)Office suiteDistancePoint (geometry)TheoryGame controllerPhase transitionElectronic visual displayTotal S.A.Nuclear spaceDependent and independent variablesNumberSoftware testingAreaContext awarenessStaff (military)Web pageSurfaceSummierbarkeitPower (physics)Moment (mathematics)Bit rateRhombusAerodynamicsSoftwareQuicksortArithmetic meanWeightLecture/Conference
38:37
PlanningProcess (computing)BitContext awarenessPhysical lawRule of inferenceRight angleInformation technology consultingComputer wormMereologyClient (computing)Channel capacityoutputSoftware engineeringObject (grammar)NumberNeuroinformatikGUI widgetMultilaterationMultiplication signPower (physics)Data managementDecision theoryConnected spaceComputer-aided designMach's principleSelf-organizationSound effectPoint (geometry)Hacker (term)PrototypeFactory (trading post)Volume (thermodynamics)Dependent and independent variablesCASE <Informatik>CurveSoftwarePhysical systemSurfaceCurvatureProjective planeMathematicsComplex (psychology)Boss CorporationTraffic reportingCalculationReflection (mathematics)WeightGoodness of fitAerodynamicsQuicksortCodeTwitterData miningWritingChainService (economics)Quantum stateTrailExtension (kinesiology)Figurate numberKey (cryptography)Utility softwareMusical ensembleTape driveStudent's t-testNP-hardLecture/Conference
47:59
Computer animation
Transcript: English(auto-generated)
00:03
I'm very pleased to introduce our first evening keynote for RailsConf this year. Nicholas Means did a talk at RubyConf last year that was called How to Crash an Airplane,
00:23
and I remember looking at this as it came through the CFP app and thinking, this is going to be the most depressing talk ever, because there's a talk about an airplane crash in 1989 in which 111 out of like 239 people were killed, the amazing part of course being that not all 239 were killed.
00:43
But it turned out to be a really, really interesting meditation on how people and computers interact with each other. Anyway, so when I saw this talk come through the CFP for RailsConf this year, I thought, this is a talk that needs a wider audience maybe, and I was very pleased when Nicholas agreed to do this as a keynote.
01:01
So please join me in welcoming Nicholas Means. All right, so I hope you guys have seen a lot of really great talks today and met a lot of really interesting people. Like Sarah said, I gave a talk at RubyConf this fall about United Flight 232, and in
01:21
the intro to that talk, I said that I was a student of plane crashes, and that's true. I am a student of plane crashes. I'm very fascinated by what goes on in the cockpit, what chain of events causes a plane to crash, but it's not the whole story about my interaction with aviation. I'm actually a huge aviation buff in general, and I have been as long as I can remember.
01:43
I love planes as long as I can remember. I think it all started when I was eight or nine years old, and my parents took me to an air show at Dyess Air Force Base in Abilene, Texas. Now, the featured attraction that day were the Thunderbirds, the Air Force's F-16 demonstration team, and they did all sorts of high-speed acrobatics.
02:00
They flew in tight formations. It was amazing. It was incredibly impressive, but as great as they were, they weren't the thing that captured my imagination that day. The thing that really stuck in my young mind was standing nose to nose with this amazing machine, the SR-71 Blackbird. It's my favorite plane. I'm sure there's plenty of people in the audience. It's your favorite plane, too.
02:20
It's an amazing machine. You can just look at it and tell how fast it wants to go. It's got those razor-sharp leading edges, smooth curves. The engines are every bit as large as the fuselage. You can't really tell it from this angle. But seeing this plane, seeing it up close, and hearing about what this plane could do started a lifelong obsession with aircraft for me.
02:41
I went back home. I was in elementary school at the time, and this was before the internet, so I went to my school library, and I had the librarian pull every book she could find that even mentioned the SR-71 for me. And I started reading about this plane, and I really haven't stopped since. Well, years later, my career has taken a decidedly non-aviation turn.
03:01
I am the VP of Engineering at iTriage and WellMatch, and I spend my days leading teams of software engineers, but I'm still fascinated by airplanes and by stories from the world of aviation. Sometimes I even find wisdom in these stories about how we practice our craft and how we lead our teams. The story of United 232 was very much one of those stories for me, and this is one of those as well.
03:22
So if you see an SR-71 in a museum somewhere, you should look for this logo on the tail. It's not always there, but sometimes it is. The reason for this skunk is that the SR-71 was designed by Lockheed Martin's Advanced Projects Division, better known as the Skunk Works. Now, companies use the phrase Skunk Works for all sorts of things, usually some top
03:43
secret project where they need a bunch of innovation in a hurry, but Lockheed Skunk Works was the original one. And today, I want to tell you the story of some of Skunk Works' most iconic planes and the amazing engineers that built them. And to do that, I have to start with Clarence Kelly Johnson. Without him, there would be no Skunk Works.
04:02
Kelly graduated from Michigan in 1932. He applied for work at Lockheed, and he was turned down. He went back to Michigan to get his master's degree in aeronautical engineering, and after he got his degree, he went back to Lockheed, and he was hired, not as an aeronautical engineer, but as a tool designer for 83 bucks a month.
04:21
Slowly but surely, Kelly worked his way up the ranks, and the first plane he designed that you would probably know of is the P-38 Lightning. Now, if you've studied World War II aviation at all, if you've ever been to a World War II aviation museum, you have seen this plane. It's one of the most famous planes of World War II, and it was one of pilot's favorite planes to fly. It was very successful in dogfighting.
04:42
So Kelly kept himself busy working on that until intelligence started to come in that the Germans had developed a new plane, the Messerschmitt Me 262. Now what made this plane remarkable is it's the first jet fighter that was ever placed into service. It was faster than anything the Allies had. The Germans had invested in jet propulsion far earlier than anybody else, and they
05:02
were way far ahead of the Americans. Now the British had offered the de Havilland H-1B Goblin engine to the US, and the Air Force held a meeting with Lockheed and asked if they would be interested in designing a plane around this engine. The Air Force proposed that Lockheed build a single prototype, and they designated it
05:20
the XP-80. Well, all along, Kelly Johnson had been pestering his bosses at Lockheed to set up an experimental aircraft division where he could let engineers and designers and mechanics work in close proximity to each other and communicate directly, not have to go through all the bureaucratic channels at Lockheed, and this seemed to the higher brass at Lockheed to be a perfect opportunity to give Kelly Johnson that.
05:44
The only problem was Lockheed had no factory space available. This was in the middle of World War II. All of their facilities were busy manufacturing the P-38 Lightning, and so Kelly Johnson's first order of business was to rent a circus tent.
06:02
He set this circus tent up next to an existing building on the Lockheed grounds. He installed phones, air conditioning, everything you needed to make it an office. The building he set it up next to was a plastic factory, and apparently it smelled terrible. So the XP-80 was a top secret. The team had been briefed to not reveal to anybody what they were working on, even when
06:20
they answered the phone, and so because of the smell, Irv Culver, one of the structural engineers on the project with a reputation as a bit of a cut-up, took to answering the phone, skunk works, how can I help you? And the name stuck. So that's how skunk works became skunk works. The contract for the XP-80 was signed on June 24th of 1943, and the team had been
06:43
given 180 days to build this plane. The only concrete information they had was the dimensions of the engine. They didn't even have a mock-up. They had to build that themselves in-house from the blueprints, and they designed the plane around this engine. Normally, they would have mocked up the whole plane before they started in on building
07:01
the production aircraft, but not this time. Kelly Johnson decided that the plane itself would be their mock-up, that they wouldn't mock the plane ahead of time, and that his engineers would be free to design and manufacture parts on the spot to fit this plane. He also decided to do away with Lockheed's normal drawing approval process.
07:20
He decided that if they were going to bring this plane in on time, they had to work fast, and that meant doing away with all the formality they were used to working with. So he cut all the style rules and approval chains that would normally apply to airplane drawings at Lockheed, and it worked. By November the 13th, they were done. Just 143 days from when they started, they had a complete plane.
07:43
They took that plane apart, created it up, and loaded it on a flatbed truck, and drove it 70 miles east to Muroc Air Force Base in the middle of the Mojave. Now, why did they do that? Because they needed lots of room for this thing to crash. They had no idea how it was going to perform. But it performed beautifully. After New Year's, it took flight for the first time, and it flew like a dream.
08:03
The prototype they flew that day would actually go on to be the first American plane to fly 500 miles an hour in level flight, the fastest plane built to that day. The production version, the P-80 Shooting Star, would go on to be the first jet deployed by the Air Force. And it flew well into the 80s.
08:21
So they completed their mission in an unrealistic amount of time. They delivered this plane. But it started to look like maybe that would be the end of Skunk Works. This was the end of World War II. There wasn't a lot of money available for developing new aircraft. The Pentagon decided they really didn't need any new airplanes with no war going on.
08:41
But that stance didn't last for very long. This picture is of Winston Churchill, FDR, and Joseph Stalin at the Yalta Conference in February 1945. This is one of the three conferences that the big three World War II allies had to determine how they were going to govern Europe after the war. This particular conference is the one where they decided they would split Germany down the
09:01
middle and split Berlin with it. These three superpowers had united against the Axis powers during World War II, but that alliance didn't last very long after World War II. American and Russian ambitions were too much in conflict, and they quickly began ramping up military spending to make sure they kept up with each other.
09:21
We had entered the Cold War. In addition to ramping up military spending, the other thing that ramped up was reconnaissance activity. Around this time, you have to understand that 55 percent of the American population thought that it was more likely that they would die from thermonuclear war than old age. And those fears weren't unfounded. Both sides needed to know what the other was up to, and they were willing to spend
09:42
a ton of money trying to figure it out. The CIA was desperate in particular for information on this place, Kapustin Yar. This is Russia's primary secret missile development area. It's akin to Area 51 in the United States. The Air Force considered an overflight of Kapustin Yar to be far too
10:01
dangerous to do with any of the aircraft they had at the time. It was very heavily defended. They knew there was no way they could get in, take pictures, and get back. So the CIA needed a different answer. Their intelligence indicated that Russian radar couldn't see over about 65,000 feet, so they decided to spec out a plane that would fly at 70,000 feet.
10:20
Well, nothing had ever flown that high before, but they requested bids. And since they had no means of reconnaissance over Russia until this plane was ready, they needed this plane in a hurry. The bid request they put out called for this plane to be ready in eight months. So Skunk Works took a plane that they had earlier developed, the F-104 Starfighter,
10:40
which coincidentally is the first plane ever built that went Mach 2. It could go two times the speed of sound flying level. They took this plane and proposed that they modified it by dumping as much weight as they could, stretching the wings out to be as wide as they could, and changing the engine out to something that would function at 70,000 feet, because nothing had ever flown that high before.
11:00
They didn't know how to build a jet engine that would fly at 70,000 feet. Because their proposal was based on an existing plane, along with Skunk Works' proven capability on the P-80 and the F-104 to deliver on time on tight deadlines, it won over the other manufacturer's proposals of new planes. And the plane they built, of course, is the U-2.
11:22
They started work in November of 1954. They took the U-2, and they lost as much weight as they could. They took the fuselage, made it as thin as they possibly could, made it of wafer-thin aluminum. It was so thin, in fact, there's a story that an engineer accidentally bumped into this plane with a toolbox. And, you know, a normal plane, that would be no big deal.
11:40
But the U-2, it left a four-inch dent in the side of the fuselage that they had to pound out. There was some concern that this plane would never be strong enough to fly. But eight months later, in July 1955, right on time, they had a plane ready. They crated it up, loaded it into the belly of a cargo plane, and flew it out to a purpose-built airfield in the middle of a dried-out lakebed in the Nevada desert.
12:01
Why the Nevada desert? Because they weren't sure this plane would fly, and they needed lots of places to land it. This picture, taken by Kelly Johnson himself, is of the actual first flight of the U-2 on August the 4th, just a hair over eight months from when the first metal was cut. A month after the first flight, pilots were breaking altitude records
12:22
in secret almost daily over the Nevada desert. By the time they were done flight testing this plane, it had been up to 74,500 feet, well above its operational ceiling. And it had flown over 5,000 miles over 10 hours on a single tank of gas. And how'd they do it? Despite the ability to fly three miles higher than any other plane built to that point,
12:44
the U-2 was a remarkably simple plane. Weight was everything. Every pound cost the plane about a foot of altitude. So they cut weight wherever they could. This is a picture of the internal wing structure of the U-2. It weighs about four pounds per square foot. Most airplane wings weigh about 12 pounds per square foot.
13:04
So this is a third the weight of a traditional aircraft wing. To me, it looks like a cheap metal awning. There's just not a lot of material there. And of course, this introduced a lack of rigidity in the wing. So the U-2 is known for when it hits turbulence, the wings will flap like a seagull. Scared the pilots to death, but the wings never broke off.
13:24
The U-2 was also designed with tandem bicycle landing gear. If you look closely at this picture, there's no wheels under the wings. There's only two sets of wheels under the center line of the fuselage. The combined weight of this landing gear mechanism is 200 pounds. It's the lightest landing gear that's ever been deployed on a jet aircraft.
13:43
And it's easier just to show you how this works. So we're riding along in a chase car here behind a U-2. The reason they have the chase cars is because the pilot is in a bulky pressure suit and literally can't see where they are in relation to the ground. So the driver in the chase car is constantly calling out the altitude to them, and telling them how close they're getting to the ground.
14:02
And the plane, you can't land it. It wants to fly so badly, you literally have to stall it into the ground. You have to bring it down to about a foot and then stall it. And then the pilot has to fly it down the runway. He's literally flying the plane, balancing it on two wheels down the runway until he finally bleeds enough speed off to tip the plane over onto its wing.
14:24
And then they have to put landing gear under the wing so that it can taxi the rest of the way into the hangar. Now look at these guys pulling on the wing on your left here. Look how much this wing is bending as they're pulling, trying to get the other wing off the ground. Just ridiculously flexible.
14:44
They finally get the pogo gear under the wings, and this is how it takes off as well. It leaves the hangar with these under the wings, and those fall off as it reaches speed and finally lifts off the ground. It's a total hack.
15:00
And the reason is that every part of the U-2 served only one purpose. And the purpose of this plane was to get this payload to 70,000 feet over Russia. This payload, which is currently in the National Air and Space Museum in Washington, is a high-resolution camera with 36-inch focal length that could resolve an object that was two and a half feet across from 70,000 feet.
15:22
Keep in mind, this is the 1950s. This is the highest-resolution camera that has ever been built. And because that's what they cared about, they hacked the rest. They could have made the wings more rigid so they didn't flap, but it didn't matter. They could have put different landing gear on it so that it would be easier to land.
15:42
It didn't matter. This is actually a modern-day U-2. This plane is still in operation. It still has the same landing gear configuration. The wings are 20 feet wider. It has 30% more payload capability, but they never changed that crazy landing gear configuration because it works. They didn't need to, but they had a problem.
16:05
The operating assumption that Russian radar couldn't see above 65,000 feet turned out to be incorrect. Almost from the first flight U-2 took over Russia, MiGs were chasing at 15,000 and 20,000 feet below. They were firing missiles at it.
16:21
Now, nothing could get up to its altitude, but the CIA was afraid that they only had 18 months to two years of operational viability out of this plane before Russia figured out a way to shoot it down. And so they needed another answer. They needed the replacement for the U-2 almost as soon as they put it into service.
16:42
And so they designed a plane that would be faster and higher. They wanted a plane that would fly at 100,000 feet and cruise at at least Mach 2, which are crazy numbers. And so in response, Skunk Works started working on the Archangel series of design studies. This is an early model of the Archangel.
17:01
By the 11th design revision, it's starting to look a little bit more familiar. You probably think I'm about to tell you about the SR-71, but you're wrong. The plane I'm about to tell you about is the Lockheed A-12. This plane is the predecessor to the SR-71. Most people don't know it existed. The technological leap that this plane represents is almost impossible to comprehend.
17:23
It's designed to fly five miles higher than the U-2 at 90,000 feet, and it's designed to fly four times faster than the U-2 at Mach 3.25. Now, the fastest plane America has built to this date is still the F-104 Starfighter, and it can fly at Mach 2 for about a minute, minute and a half
17:42
before it either runs out of gas or the engines start overheating. This plane was intended to fly at Mach 3.25 for an hour and a half to two hours at a stretch. Now, performing at those extremes meant almost everything the team knew about traditional airplane design didn't apply, and the CIA generously gave them 22 months to figure it out
18:05
to coincide with the expected end of operational viability of the U-2. Now, aluminum was the material that they would usually build an airplane out of. It's still one of the most common materials for airframes. The problem with aluminum is that it loses its structural integrity at about 300 degrees Fahrenheit.
18:22
The calculations that they did indicated that this plane would be 800 degrees Fahrenheit at the nose and 1,200 degrees Fahrenheit at the engine cowlings. So the aluminum, if they built the plane out of aluminum, the aluminum would literally just fold up. It would have no structural integrity at those temperatures. They considered building it out of stainless because that's the obvious option when you need steel
18:41
that's gonna hold up under high heat, but that would make the plane too heavy to get to the altitudes it needed to get to. And so Henry Combs, the primary structural engineer on the A-12 project, suggested they consider titanium. Now, he had built the engine exhausts of the F-104 out of titanium, and it had worked great. The only problem with building this plane out of titanium
19:00
is that nobody knew how to build something this big out of titanium. The biggest thing that they had manufactured out of it was engine nozzles. Still, Kelly Johnson was favorable on the proposal. He said, any material that can cut our gross weight by half is damn tempting, even if it's gonna drive us nuts in the process. And he was right about it driving them nuts.
19:20
They ordered the first batch of titanium in to see what they could do with it, and they realized they had no idea how to extrude it. They had no idea how to weld it. They had no idea how to rivet it. They had no idea how to drill it. The drill bits that they used on aluminum would literally shatter when they tried to drill through titanium with them. On top of that, the US supplier that they ordered these preliminary batches from
19:41
didn't have enough capacity to supply them in the quantities they needed for the number of airplanes they thought they were gonna build. So they asked the CIA for help, and the CIA, through a series of dummy companies and anonymous third parties, set up a supply chain for the leading exporter of titanium of the day, the Soviet Union.
20:01
So the very metal to build the A-12 came from the same country it was intended to spy on. The extreme operating environment required adaptation everywhere in the plane. Early calculations they did also indicated the plane, when it got up to cruise altitude
20:20
and cruise temperature, would stretch by two to three inches. It would literally get longer because of how fast it was going. So everything in the plane had to cope with that. The control cables were made of Algyloy, which is the alloy used to make watch springs because it maintains its tensile strength at very high temperatures. The engine nozzles were made of Hastelloy X,
20:41
which is a nickel alloy, and they chose Hastelloy X because they knew it could withstand the 3400 degree Fahrenheit that the afterburners were expected to produce, and it could withstand those temperatures for the hour and a half to two hours that they would be running on afterburner. Off the shelf electronics wouldn't function because of the temperature, neither would greases, oils, hydraulic fluids, even fuels.
21:01
They had to come up with new answers for all of these things because of the operating environment of this plane. They had a custom fuel developed that wouldn't be volatile at the expected range of operating temperatures. The only problem was you couldn't get the stuff to burn. It had such a high flash point that it literally wouldn't ignite. So you had to do this.
21:21
Inject the engine with triethylborane, which is really, really nasty stuff that spontaneously combusts with this bright green flash when you expose it to the atmosphere. That was the only way to get this high flash point temperature fuel to ignite.
21:40
One of the biggest challenges was propulsion, and that's why Kelly Johnson put 32-year-old Ben Rich as the lead propulsion engineer on this plane. Young guy, not a lot of experience, but Kelly Johnson trusted him. This is one of the few places that they actually were able to adapt something off the shelf for this plane. They picked Pratt & Whitney's J58 turbojet engine. Pratt & Whitney had built this engine
22:01
for a Mach 2 Navy fighter that had been canceled, and Pratt & Whitney had about 700 hours of testing on this engine and really wanted to find a place to use it, so they were willing to go to the extremes that Skunk Works needed them to go to to make it work in this plane. They had to modify the engine to make it be able to operate continuously on afterburner
22:20
and in the thin air at 90,000 feet, but that wasn't the major innovation of the propulsion of this plane. Major innovation is the cone that you see right there. Now, that cone actually moves back into the body of the engine by about 26 inches when it gets up to cruise velocity. To understand why, you have to understand how jet engines work. Jet engines work on compression.
22:42
There's a wide opening at the front of the engine that scoops in as much air as possible, and over a series of compressors, it compresses it into a very compact stream that pushes the plane as fast as possible. You can think about what happens when you put your thumb over the nozzle on a garden hose. It's the same effect.
23:00
So these engine cones, as it got up to Mach 3, would move back into the engine 26 inches, and they were responsible for 70% of the thrust of this engine at cruise velocity. The afterburners contributed another 25%, and the engine itself only contributed about 5% of thrust. So this engine essentially converted from standard jet
23:21
to ram jet in the middle of operating. Air entering this engine at minus 65 at 90,000 feet would be 800 degrees Fahrenheit before it hit the combustion stage of the engine. This is a crazy amount of innovation. But what's just as interesting to me about this plane is the things that Skunk Works chose not to solve.
23:42
There was no fuel tank sealant that would work over the entire operating range expected of this aircraft. And so this plane would literally sit on the tarmac dripping fuel. You can see the puddle under this plane. That's jet fuel. They just didn't care. It didn't matter. When it got up to supersonic speed,
24:00
the fuel tanks would seal. It was no big deal. The other interesting thing is the plane can't even start itself. Now to start these massive jet engines, I already told you about the triethylborane. But the other thing that you have to do to get the combustion to be self-sustaining is you have to get the turbine spinning at 4,500 RPM.
24:23
You have to do that before you inject the triethylborane and get the fuel burning. Well, they thought about adding a starter motor to the plane, but it would take a very large starter motor to get the turbine turning as fast as it needed to. So they did this instead. This is the AG330 start cart, or the Buick, as the ground crews called it.
24:42
And the reason they called it that is because contained in this start cart are two Buick V8 Wildcat engines. And they would physically couple this thing to the starter shaft of these massive turbines, crank the two V8 engines up to full throttle, get the turbine turning 4,500 RPM, and light it off.
25:04
That's a crazy hack. The ground crews said that the hanger literally would sound like a stock car race when they were starting this plane up. But it didn't cost them any altitude. There are only two things that mattered
25:20
in building the A-12. It needed to go very fast, and it needed to do so very high. It went five miles higher and four times faster than the U-2. And on April 30th, 1962, one year and 100% over budget, Skunk Works gave the CIA what they wanted. This is a picture of the A-12's first flight.
25:41
Dripped fuel, couldn't even start the engines without crazy chemicals and a couple of V8 engines. It actually couldn't even take off with a full load of fuel. It had to hit a tanker almost as soon as it took off because those tiny wings wouldn't generate enough lift if you put a full load of fuel in the plane. But it didn't matter. They spent their money and their time on the things that did matter, the titanium construction, the propulsion system.
26:02
They just hacked their way around the rest. This plane went Mach 3.25 at 90,000 feet and overflew every hostile territory in the world. And it holds the distinction of being the only military aircraft never to have been shot down,
26:20
despite 3,500 missions over some of the most contested territory in the world and having hundreds of missiles launched at it. After building 15 of the A-12s for the CIA, the Air Force requested a two-seater variant with twice as much payload. That plane was the A-12's far more famous younger brother, the SR-71. It holds about every speed and altitude record there is.
26:44
It holds the record for sustained altitude at 85,069 feet. Now keep in mind, these are official records determined over an official course. The plane almost certainly flew higher than this in combat. It holds a record of sustained speed at 2,193.2 miles an hour.
27:03
It's about Mach 3.3. Now Brian Schuhl, in his book Sled Driver, he tells a bunch of stories about flying this plane. And one of them he tells is outrunning missiles in Libya. Muammar Gaddafi had launched everything in his battery at Brian's plane, and he just kept pushing the plane faster
27:20
and faster and faster, because he knew if he could just make it to his turn and get out of the country, he could miss these missiles. Well, his reconnaissance systems officer in the back seat, once they made this turn, had to remind him to slow the plane back down. When he looked at the speedometer, they were going over Mach 3.5. So we know this plane would go well faster
27:41
than this speed limit. To give you some context on just how fast this is, the muzzle velocity of a .22 caliber rifle bullet is 2,046 miles an hour. So at cruise speed, the SR-71 Blackbird can literally claim to be faster than a speeding bullet. It also set a bunch of speed records over courses.
28:03
It could fly from New York to London in an hour and 55 minutes. The Concorde, on a good day, with a heavy tailwind, could do it in 252. It could fly from Los Angeles to Washington in one hour and four minutes. And over the course of setting that record, it set another one that's one of my favorites,
28:20
because it's really easy to wrap your head around. It flew from St. Louis to Cincinnati in eight minutes and 32 seconds. Now, if you want to drive that in your car, it'll take you about five hours and 16 minutes. It's just an incredibly fast plane, and it's probably gonna hold these records forever. With the invention of high resolution satellite photography
28:43
and unmanned aerial vehicles, there's really no reason for us to ever build a plane like this again. It's just a crazy amount of innovation, especially when you consider it was built in the 60s. It would be an innovative plane if you built it today. Well, the SR-71 was Kelly Johnson's crowning achievement.
29:01
In 1975, he hit Lockheed Martin's mandatory retirement age of 65, and he passed the reins on to this man, his protege, Ben Rich. This is the same Ben Rich that had designed the propulsion system for the A-12 at 32 years of age. Now, Ben took over Skunk Works at kind of a tumultuous time.
29:20
The US appetite for defense spending was at an all-time low after Vietnam. There just wasn't much energy to spend money on new technology. Lockheed had attempted, in the wake of this, to reenter the commercial aviation market with this plane, the L-1011 Tristar, and had lost about $2 billion in the process. And keep in mind, these are 1975 dollars.
29:45
So Rich had to find significant new work, and he had to find it fast, or he was gonna have to let go most of his most expensive and most experienced engineers. Meanwhile, the Cold War continued. Leonid Brezhnev, who'd been the Russian premier for most of the Cold War, would be in power for another eight years or so.
30:02
The Soviet Union had invested around 300 billion rubles in developing radar and surface-to-air missiles like this SA-5 that were far more advanced than any attack capability the Americans had. We couldn't fly against this. To give you some context on this, in the 18-day Yom Kippur War, which was largely a proxy war
30:20
between the US and Russia, Israel lost 109 US-built, US-trained pilot aircraft against these SA-5 missiles that were operated by largely untrained, largely undisciplined Syrian and Egyptian troops. The Soviet technology was so good, it didn't even require an experienced operator to be able to shoot down our best technology.
30:42
And to maintain the mutually assured destruction that had kept the US and Russia from all-out nuclear war throughout the Cold War, the US needed to develop something that could pierce these defenses, but ideas were in short supply. Until Dennis Overholser, a 36-year-old math and radar expert on the Skunk Works staff, walked into Ben Rich's office and tossed this document on his desk.
31:02
The method of edge waves in the physical theory of diffraction. Sounds like a really engaging read, right? Well, it was so engaging that it had actually been published by Pyotr Ufimsev, excuse me, who was the chief scientist at the Moscow Institute of Radio Engineering nearly a decade earlier
31:22
before the Air Force finally got around to translating it. They just didn't think there was anything of tactical value in here. They hadn't prioritized it. Overholser, however, found something on the last page that seemed very substantial to him. It was a method for calculating the radar cross-section of the edge and the surface of a wing and come up with an accurate number
31:41
for just how visible that wing would be on radar. You have to understand that accurately determining how visible a plane would be on radar in these days was largely impossible without building a scale model and sticking it on top of a pole on a radar range, which you can see the A-12 here. Folks like Overholser, who knew something about the science behind radar, could make some educated inferences
32:00
about what might make a difference to observability, but there were no hard and fast rules and there was no way to know for certain until you actually tested it. Stealth had long been theorized as something that might be possible, but it was always written off as too difficult, too expensive to try. But in Opham's document, Overholser was convinced that he had found the formulas that would let them predict observability ahead of time
32:23
and empirically design for it. So he asked Ben Rich to let him start on some software. Five weeks later, Dennis Overholser walked into Ben Rich's office with a sketch of this thing, which the Lockheed technical staff would quickly take to calling the Hopeless Diamond because they didn't think they'd ever be able to get it to fly.
32:44
Now in the preliminary radar range test that Lockheed did in Palmdale, the radar operator, in trying to scope the radar in on this model, thought that maybe the model had fallen off of the pole on the test range. And so he asked Ben Rich to stick his head out the window and see if the plane was still on top of the pole.
33:01
So Ben did that. About the time Ben stuck his head out the door, along comes a crow and lands on the plane. And the radar operator goes, oh, never mind, I've got it now. So you couldn't see the plane, but you could see a crow on the radar. And at that moment, Ben Rich knew they were onto something big.
33:21
Around this time, DARPA was holding a competition for design of a theoretical stealth aircraft. Lockheed and Northrop won the first phase of the competition and were given 1.5 million to refine their concepts and build 38-foot models that were then gonna be tested at the Air Force's most sensitive and sophisticated radar range in White Sands, New Mexico. That's what you see in this picture,
33:41
Lockheed's 38-foot model. The only problem was that when they got this out to the radar range to test it, the model was so good that the only thing they could see on the radar was the pole. Now the Air Force had always assumed that the pole on their radar ranges was invisible. They had never had this problem before. And they didn't know what to do about it.
34:03
So Dennis Overholtzer went to work and he made him a better pole too. And that's what you see here. The pole cost somewhere around $500,000 in and of itself but it was no longer visible on radar and they could test the plane. And they came up with a really interesting way
34:21
to test the plane to see just how visible it was on radar. They knew that they could take a ball bearing and calculate the theoretical responsiveness of a ball bearing. They knew what a ball bearing should look like on radar. So they decided that they would glue ball bearings to the front of the plane and see how small they could get before they saw the airplane. This is where they started. This is a two inch ball bearing.
34:40
They couldn't see the plane. They went smaller and smaller and smaller and smaller until they got to this. That is a one eighth inch ball bearing. Most of you probably can't see that. It's smaller than a BB. They still couldn't see the plane. They only saw the ball bearing.
35:03
So needless to say, Lockheed won the competition pretty easily. The only thing that was left to be seen is A, can you get this thing in the air? And B, once you do, is it still stealthy once you add things that the model doesn't have like engines and air intakes and a pilot?
35:21
The Air Force wanted two prototype planes in 14 months and Skunk Works agreed. And 14 months later, they came up with this, the Have Blue. Now if this thing looks like it might fly to you, it's just because your brain has been conditioned to think that things that look like this are airplanes, because you've seen enough pictures
35:40
of the stealth fighter over the years. Everybody at Lockheed was still sort of in doubt until they actually saw it in the air. Now how did they get this thing built in 14 months? Well, not invented here was not a thing at Skunk Works. This thing is literally off of the surplus shelf. It uses the flight control computer from the F-16,
36:01
navigation from the B-52, the pilot seat from the F-16, the heads up display from the F-18, engines from a T-2B trainer, and the list goes on and on and on. The only original thing about this plane is the outer skin. The biggest thing that they had to solve, obviously, is aerodynamics. This thing is actually unstable
36:21
in all three axes of flight. That means that it is pitch unstable, it is yaw unstable, and it's roll unstable. The only plane that they had deployed that was unstable in any axis of flight was the F-16, and it was only pitch unstable. It didn't have to contend with all three.
36:43
So they set the flight computer up to determine what inputs they needed to send to the control surfaces to make this thing fly. And they would take the pilot's input and sum it with the things that the computer knew that it needed to do, and that would be what went to the control surfaces. The early nickname for this plane was the Wobblin' Goblin, because it took them a while to dial in that software.
37:03
But true to form, they got it to fly. Most test flights were at night to avoid prying eyes, so this is one of the few pictures we have of this plane in the air. That's why it's such a terrible picture. But now that it could fly, they needed to see if it could live up to the promise of stealthiness. And so they flew it against this,
37:20
the target acquisition radar from a Hawk missile battery, the most advanced radar technology that the US had at the time. The plane literally overflew this radar right over top. Radar never picked it up. The missiles never swung into alignment. They just pointed lazily at the mountains off in the distance. And they knew that their plane was a success.
37:42
So about five years later, the first F-117 stealth fighter detachment was in operation at Tonopah Test Range Airport. Now, Tonopah Test Range is the massive military complex in the Nevada desert that encompasses Area 51. So there's a pretty good chance that a large number of UFO sightings in the late 80s are this thing.
38:03
The American public didn't find out about this plane until the first night of Operation Desert Storm. The Air Force sent a total of 22 F-117s into Baghdad that night. And privately, they had calculated that they would lose about 30% of those planes. Because Baghdad at that point was more well defended than even Moscow had been at the height of the Cold War.
38:23
But they didn't lose any planes that first night. As the pilots were flying out of Baghdad and they re-established radio contact, they realized that everybody was present and accounted for. They didn't lose a single one of these planes over the entirety of Operation Desert Storm. This whole plane is one big hack.
38:43
They needed it to be invisible on radar. That's all they needed. And they got there. They got really close. They got it to be as visible as basically a BB. By basically not caring about aerodynamics at all. And hacking their way around the laws of physics that govern how planes fly.
39:02
Computers of the day weren't powerful enough to calculate radar reflectivity of curved surfaces. So that's why this thing is made up of a bunch of angular flat surfaces. It had nothing to do with the design of the plane. They didn't have the computers to design stealthy curved surfaces. So they just built a plane that was all straight surfaces. Kelly Johnson had a longstanding saying
39:21
that beautiful planes fly beautifully and nobody at Skunk Works thought this was a beautiful plane. But it didn't matter. It didn't matter that it wasn't a beautiful plane. It did exactly what it was designed to do. So how'd they do it? All of these amazing planes, each of which was groundbreaking in some significant way. There's plenty more that we haven't even talked about.
39:42
Well, our story ends the same place it began. A scrappy team of, at its peak, 23 designers and 105 fabricators created the P-80 around a mocked-up engine in 143 days. And that plane was in service for over 40 years. Not much about Kelly Johnson's philosophy
40:00
of how to build planes changed over the years, even when he passed the reins on to Ben Rich. He was a proponent of prototyping and learning. I tried to find a picture of Have Blue and the F-117 together on the tarmac. I figured surely it had to be out there. But I started looking at the dates, and I realized just how much of a throwaway prototype Have Blue was. They had managed to crash both of them
40:21
before the F-117 was ever built. He liked to iterate. You can see here the A-12 is on the right and the SR-71 is on the left. The A-12 could go a little faster and a little higher than the SR-71, but it turns out it didn't need to. They revised it to a two-seater with double the payload capacity for the Air Force,
40:40
gladly trading a bit of altitude and speed for more utility, more mission viability. Kelly also had some general rules about how to run his organization. If you wanna know more about those, you can Google Kelly's rules and you'll be taken to Kelly Johnson's 14 rules for Lockheed Skunk Works. But I'm gonna tell you about a couple of them that are especially applicable to us as software engineers. The first one is that the number of people
41:02
having any connection with the project must be restricted in an almost vicious manner. He's a small number of good people, 10 to 25% compared to the so-called normal systems. At its peak, there were 75 design engineers working on the SR-71. To give you some context for that,
41:20
Boeing used 10,000 engineers to build the Triple Seven, and Boeing had the advantage of computer-assisted design software. Lockheed was still doing all their drawings by hand at that point. Kelly Johnson hired smart people into his organization and he trusted them to do good work. But lots of companies run their software engineering organizations, like Henry Ford ran his assembly lines.
41:43
They measure all of the work the way that Henry Ford measured how many cars were produced, how effective a worker was. They set up all sorts of heavy processes to govern what work gets done and when, and they add a new process for every little hiccup. And all this process has the desired effect. We turn off our brains and we become factory workers
42:00
for as long as we can stand the boredom. The things we build, though, have more in common with the planes that Skunk Works built than they do the cars of Henry Ford. We're knowledge workers building unique software, not assembly line workers putting together widgets. How does this work in practice? Well, Peter Drucker, probably the most prolific writer on management in America
42:21
tells the story of a young infantry captain in Vietnam. The reporter asked this infantry captain how in the fog of war he maintained command of his troops. And the commander responded, around here, I'm only the guy who's responsible. If these men don't know what to do when they run into an enemy in the jungle, I'm too far away to tell them. My job is to make sure they know what to do.
42:42
What they do depends on the situation, which only they can judge. The decision lies with whoever is on the spot. This is how software teams work whether we acknowledge it or not. You're constantly making decisions as you're writing code. Managers can choose to either trust their teams to make good decisions, or they can smother them with process and micromanagement
43:01
and try to have a hand in every decision that their teams make. Good managers hire smart people and trust them to make good decisions as they write code. But they also focus on enabling them to make good decisions by making sure they understand the context and the overall goals of what they're working on. Kelly Johnson handled this
43:21
by lightening up his systems. A very simple drawing and drawing release system with great flexibility for making changes must be provided. Now I told you about this earlier on the P80. He got away from the complex drawing systems that were required elsewhere in Lockheed because his small teams just didn't need them. They didn't need that much process. They were able to get the work done
43:41
with a much lighter amount of documentation. Now this lightweight process wouldn't have worked at the Lockheed main plant because they were trying to build a far higher volume of airplanes with far less skilled workers. But Kelly Johnson had a small team and he trusted them.
44:02
Sarah May actually had a really great tweet on this the other day. She said that team pathology is always either hanging onto processes suited to a smaller team or early adopting processes suited to a larger team. It's very true. 10 years ago a friend of mine and I decided to start a boutique software consultancy. So think about the first thing that you would do
44:21
if you were gonna start a boutique software consultancy. It's probably the same thing that we did. We went out and spent 1,000 bucks on a Jira license and spent the better part of a week getting it stood up on a VPS so we could track our work. Now keep in mind it was two of us and we had one client at the time. We didn't last very long.
44:43
You need enough process so that everyone has the context they need but not so much that people turn off their brains and blindly do what they're told. Your process is there to serve you, not the other way around. This is what Kelly Johnson got so right. He couldn't have delivered all this innovation on his own. It wasn't in his brain
45:01
but the processes he put in place allowed his teams to set the right priorities and make the right compromises at the right point in time when they made decisions. His most important rule was that there shall be only one object to get a good airplane built on time. What made a good airplane? Delivered the value the customer needed,
45:21
hit the key specs and compromised wherever necessary to hit those key specs. He was a pragmatist. Every decision he and his team made was around how to deliver the most value in the shortest amount of time for the customer while bringing out the best in his team because of the freedom and the trust that he gave his teams and because of how clearly he laid out the goals for each project, they were able to deliver
45:41
some of the most amazing planes ever built. The U-2 landed on terrible landing gear. Pilots said it was the easiest thing in the world to fly from 60,000 feet down to six inches. They hated landing it. But the team decided it was worth it to save the weight in favor of the altitude and the pilots learned to work around it. It was a good compromise.
46:00
It was a great hack. The SR-71 is the fastest plane ever built but it couldn't even start itself because it didn't have a starter motor. It would have added too much weight. It sat on the tarmac dripping fuel. They just didn't care. The team spent their time figuring out how to do the important things, how to build a plane out of titanium, how to make it go Mach 3.25 and they hacked their way around the other stuff.
46:22
The F-117 violates every law of aerodynamic design and they worked around it to make it invisible on radar. It bucks conventional wisdom in almost every way possible because of the trust that Ben Rich put in Dennis Overholser. Kelly's and later Ben's teams had unprecedented input into what they were building.
46:42
They had incredible freedom and incredible trust from their bosses. You should push for that freedom in your job if you don't already have it. The process that you follow should be the right size for your team and you should know the most important things about what you're building. You should know the goals so that you can contribute to your project's success beyond just writing code.
47:01
If that's not the case for you, push back hard, change it. If you're in a leadership role, you have a responsibility to give that freedom to your team. You need to push as many decisions and as much responsibility down to your team as you can. You need to make sure you're clearly communicating the two or three most important things that your team needs to be building at any given time
47:21
so that, like those frontline soldiers in Vietnam, they can make the right decisions based on what they're seeing and the code they're working on at that instant. You have to give them the context to make those decisions and you have to trust them to make the right decisions. If you do these things, if you trust your team to innovate and don't just trust yourself,
47:42
if you trust your coworkers to build amazing things, there's no telling what amazing stuff you're gonna be able to build together. Thanks a lot.