Previous  |  Next

Sun

03.18.07

Tim O'Reilly

Tim O'Reilly

Honoring the Skunk Works

It's easy to forget our history when we're gushing about the potential of new technology

In response to a comment on my blog entry O'Reilly Radar Executive Briefing at Etech, rektide wrote:

lockheed skunkworks logo

The lockheed martin polecat always comes to mind when talking about rapid prototyping. A spy drone with an 18 month turnaround, built with 3d printers forming composite materials. Composites are often regarded as expensive material requiring extensive tooling, but if you build the correct tooling (3d printers), there can be immense benefits.

After I forwarded Rektide's comment (which also included some fascinating background on how the boat uphostery business in his hometown has changed due to new manufacturing technologies) to the Maker back channel at O'Reilly, Dan Woods, Associate Publisher of Make wrote:

Interesting that rektide notes the Lockheed Skunk Work's polecat as a rapid prototyping example (18 months concept to launch). A fast turn indeed, and I'm sure as rektide comments, 3d printers forming composite materials was indeed an advantage. But I can't help but comment that, since its inception during World War II, the Skunk Works has been legendary for an almost eerily bare bones approach toward impossible engineering challenges. Small teams; a dearth of layers; shielded from the bureaucracy that is rampant in sister divisions; and guided by a few design principals drafted by Kelly Johnson himself - as I recall they were drafted on a single sheet of paper (one side).

In 1943, after the war department realized the Germans were about the take to the sky with the world's first jet fighter (a concept U.S. generals had repeatedly rejected), they turned to Kelly Johnson (creator of the P-38 Lightning and father of the Skunk Works). They asked him if if would be possible to design and develop a jet fighter capable of 600mph in just 6 months. In one of the more astonishing aeronautical design feats of all time, Kelly's band of just 23 design engineers and 30 shop mechanics conceived, designed, and flew P-80 Shooting Star (code named Lulu Bell) beating every design requirement in just 143 days. Rapid prototyping...with not much more than paper and slide rules.

It is indeed awesome to remember what some of the pioneers of our modern industry did with what we'd today consider to be impossibly primitive tools. They were indeed giants. We must raise our sights if we want to match up to them!

Kelly Johnson's skunkworks rules are also reminiscent of some of what has been learned by Web 2.0 startups. Compare Jeff Bezos' "two pizza teams" at Amazon with Kelly's rule that "The number of people having any connection with the project must be restricted in an almost vicious manner. Use a small number of good people (10% to 25% compared to the so-called normal systems)."

I also love Kelly's rule that "The contractor must be delegated the authority to test his final product in flight. He can and must test it in the initial stages. If he doesn't, he rapidly loses his competency to design other vehicles." This antedates the computer industry's "eat your own dog food by decades, and what's more, did so when lives were on the line. It is also reminiscent of the practice at many Web 2.0 startups of requiring product developers to be responsible for their maintenance and operations, at least in the skunkworks stage. Kelly also notes: "Push more basic inspection responsibility back to subcontractors and vendors. Don't duplicate so much inspection." This insight was also key to the Japanese manufacturing revolution in autos.

And while he does emphasize the importance of specifications, he also writes: "A very simple drawing and drawing release system with great flexibility for making changes must be provided." Revision-control systems have become a sine-qua-non of modern development.



tags:   | comments: 6   | Sphere It
submit:

 

0 TrackBacks

TrackBack URL for this entry: http://orm3.managed.sonic.net/mt/mt-tb.cgi/1885

Comments: 6

Thomas Beck   [03.18.07 09:09 AM]

Tim - Thanks for the great reminder that necessity really is the mother of invention. Agile practices aren't just something for the lab or for the Web 2.0 world. They're useful when something absolutely, positively needs to get delivered - quickly and reliably.

The common denominator between Skunk Works and successful agile software teams is the people. When great things need to get done, select your best people, empower them, and get the hell out of the way.

michael schrage   [03.18.07 12:36 PM]

as author of a books about both collaboration and prototyping, i'd like to chip in and point out that - from skunkworks to watson&crick's cavendish labs to tbl's cern, the essence of effective innovation has been economic infrastructures that facilitate 'iteration for innovation' - in other words, the cost of generating, iterating, experimenting with and testing innovative designs has declined relative to the costs of 'studying' and 'analyzing' designs for test...the shift from 'atoms' to 'bits' as a design medium is important - but the REAL design&innovation revolution has been the shift from 'bits' to 'its' - the ability to add more value faster to iterations per unit time...

Anonymous   [03.18.07 12:37 PM]

as author of a books about both collaboration and prototyping, i'd like to chip in and point out that - from skunkworks to watson&crick's cavendish labs to tbl's cern, the essence of effective innovation has been economic infrastructures that facilitate 'iteration for innovation' - in other words, the cost of generating, iterating, experimenting with and testing innovative designs has declined relative to the costs of 'studying' and 'analyzing' designs for test...the shift from 'atoms' to 'bits' as a design medium is important - but the REAL design&innovation revolution has been the shift from 'bits' to 'its' - the ability to add more value faster to iterations per unit time...

Matt Johnson   [03.19.07 10:15 PM]

As a Lockheed employee myself, I was surfing for the latest on Web 2.0 development to discuss on Lockheed's internal blogs when I ran across your post. It was exciting and ironic to see you, in turn, talking about LM development.



Thanks for tipping your hat to Skunk Works - its a tradition that we are very proud of, and an example of agile at its best.



See you at ETech!

serge   [03.25.07 11:14 AM]

Thanks for the reminder. I introduced the Kelly's rules in software development back in the early 1970s. I was also a pilot then and I obtained a written copy of the rules as they had just became public. The existenc of the skunk works projects had been secret before that.

Unfortunately, these days it appears to me that the *small group of good people* rule has become the *small group of people* rule and at some places it's become the *one developer* rule.

eweek.com has published a survey (last week, mid-March, probably) about the *dilbertization* of enterprise IT work.

To me, the downsizing of IT work teams in corporations is a major problem which is part of the dilbertization trend.

Enterprise IT is in bid trouble these days, the worst that it has ever been. Many processes are going to break down because of it. More defects will appear and they will take more and more time to fix, and this is going to cascade. When this will hit financial markets or other infrastructure systems then large scale social pain will result.

Decision making is breaking down in many areas, not just in government. Corporations are also living through a kind of management insanity.

It is quite possible that all the current news of war, barbarity, real or imaginary scarcity of vital resources, mixed with visions of Web i.0 and mounds of $$$ (e.g., Google), globalization/oursourcing (e.g., trying to use slave labor in Asia to replace North American IT pros, and usually failing), are contributing factors to the mental de-stabilization of IT people on a very wide scale. Enterprise IT managers seemed to be more affected than others. Many IT people appear to have lost their ability for good judgment.

Some of the people I work with are amazingly de-stabilized, and they are senior. They function but not coherently and not smartly. Enterprise IT has become nightmarish. Dilbert is funny on paper but when you live it on a permanent basis, it is the opposite of funny.

In a few years, we will probably remember this period as the irak-web-dot-oh time, an aftershock of the y2k-dot-com-bubble burst. I hope that we will not remember it as the big systems crashes of 200x, as in the stock market crash of the 1930s.

Matthew Thomas   [12.29.07 11:05 AM]

My grandpa John B(Jack) Wassall Aviation pioneer
Pv-1 venture, pv-2 neptune, B-17, p-38,model 49, model 89, p-80 "shooting star" p2v family,there's more, but I wanna know, is he even talked about? Can't forget the OE's of lockheed


Post A Comment:

 (please be patient, comments may take awhile to post)




Remember Me?


Subscribe to this Site

Radar RSS feed

BUSINESS INTELLIGENCE

CURRENT CONFERENCES