At QCon New York, Randy Shoup, VP Engineering at WeWork, presented "Breaking Codes, Designing Jets and Building Teams". He began the talk by quoting Mark Twain, "History doesn't repeat itself, but it often rhymes", and stated that throughout history he believes the most effective teams have focused on purpose, organisational culture, people, and engineering excellence. The remainder of talk provided a compelling overview of three effective teams, examining how each approached the four focal points, and providing some subtle (and some not so subtle) analogies to modern software development organisations and teams.
The first story of an effective team covered the World War Two code-breaking efforts carried out at Bletchley Park in the UK, in which the world's first programmable electronic digital computer was built entirely in secret in 1943. It is estimated that the intelligence produced by Bletchley Park, code-named "Ultra", ended the war two years early, and saved 14 million lives. Ultra intercepts helped the Allies in the 1940 Battle of Britain, the 1941 Battle of the Atlantic, and the 1944 D-Day Normandy landings.
Although the Bletchley Park work fell under the domain of the military, there was very little hierarchy, and the organisational style was open. The decryption was conducted using a pipeline approach, with separate "huts" (physical buildings on the campus) performing each stage of intercept, decryption, cataloguing and analysis, and dissemination. There was deep cross-functional collaboration within a hut, but extreme secrecy between each of them.
There was a constant need for iteration and refinement of techniques to respond to newer Enigma machines and procedures, and even though the work was conducted under an environment of constant pressure the code-breakers were encouraged to take two-week research sabbaticals to improve methods and procedures. There was also a log book for anyone to propose improvements, and potential improvements were discussed every two weeks.
Shoup discussed how diversity of experience is critical to code breaking, and Bletchley Park recruited linguists, mathematicians, bank clerks, crossword experts, and department store managers. The team consisted of "Boffins and Debs", and at its peak employed 10,000 people, 75% of which were women. Shoup called out the contributions of several people involved in the work, including Mavis Batey, Alan Turing, and Tommy Flowers.
The second story discussed the role of Lockheed's "Skunk Works" Advanced Developments Project group, founded in 1943, which produced generation after generation of the world's fastest, highest-flying, and stealthiest aircraft, such as the P-38 Lightning, P-80 Shooting Star, U-2 Dragon Lady, SR-71 Blackbird, and F-117 Nighthawk. The design, development, and manufacture was conducted within a single cross-functional facility, which is not typical for modern aircraft.
The aircraft were designed and built by the Skunk Works group in co-located teams, with designers and technical experts always available on site, and intense collaboration with test pilots. There was extensive use of modelling, computational simulation and wooden mockups for prototyping and verifying hypotheses. Clarence "Kelly" Johnson, the first team leader of the group, created an organisational culture focused on rapid iteration, flexibility and collective ownership, and encouraged a direct relationship between design engineers, mechanics and manufacturing. Johnson created fourteen rules that captured this doctrine. Ben Rich, the second director of the group, also believed that everyone was responsible for quality:
"We made every shop worker who designed or handled a part responsible for quality control. Any worker – not just a supervisor or a manager – could send back a part that didn't meet his or her standards."
Shoup also highlighted the work by Mary G. Ross, the first Native American female engineer, from the Cherokee Nation. She was part of the founding team of 40 Skunk Works engineers, and contributed to the P-38, Agena rocket, ballistic missiles and satellites.
The final story focused on Xerox PARC (Palo Alto Research Center), founded in 1970. The influence on the computing industry of this organisation should not be underestimated, and it produced the first graphical user interface and overlapping "windows" on a screen, object-oriented programming with Smalltalk, WYSIWYG text editing with Bravo, networking with Ethernet, and modern printing via the laser printer.
Xerox PARC had a flat organisation, with no hierarchy, and was designed as a hybrid of academia and industry. There were regularly scheduled "Dealer" meetings, in which team members took turns in presenting ideas and defending them against the questions and inquiry from the rest of the group. The facilitators of the Dealer meetings were careful to make sure that only intellectual criticism of the merit of an idea received attention and consideration. These debates helped improve products under development and sometimes resulted in completely new ideas for future pursuit. The group also practiced what they called "Tom Sawyering" -- informal and dynamic collaboration across groups and projects -- which enabled a continuous form of peer review.
Alan Kay challenged his team to create the world's most powerful language in a single page of code. "Simple things should be simple", he said, and although the ultimate output -- the Smalltalk language -- was slightly longer than two pages, its elegance and simplicity inspired the creation of subsequent "object-oriented" languages, including Java. In addition to Kay's work, Shoup also highlighted the work of Adele Goldberg, the co-developer of Smalltalk-80 and "design templates" (later referred to as "design patterns"); and his father, Richard Shoup, who developed "Superpaint", the world's first digital videographics system. The elder Shoup inspired collaborated with Pixar co-founder Alvy Ray Smith, winning an Emmy Award in 1983 and an Academy Award in 1998 for his work.
Concluding the talk, Shoup re-visited his earlier mention of the need for purpose, organisational culture, people and engineering excellence within high-performing organisations. Teams should be encouraged to think big, and they need to be laser-focused on an important and motivating goal for the entire organisation. Cross-functional, "full stack" teams are most effective for this type of work, and autonomy should be maximised, while bureaucracy and central control minimised; collaboration is key, and so is cultivating a learning culture. The best people should be hired for the job, regardless of background -- diversity of experience and perspective is highly valuable -- and they should be treated well. Engineering excellence is established through systems-thinking and looking for holistic solutions to problems. There must be a pragmatic focus on delivering, and constant iteration and feedback must be baked into all processes. Although many of the principles of Agile, Lean and DevOps may appear relatively modern, they might not be as new as many of us think.
Additional information about the talk can be found on the QCon NY website, and the slides (PDF) can be downloaded from the schedule page.