How Lockheed has embraced agile (and learned from DOD)
Although one advantage of contracting is often seen as giving agencies access to innovative practices in industry, contractors whose business is mostly with the government often mimic government practices — even dysfunctional ones. So it perhaps should not be surprising that agile software development, which began spreading in industry almost 20 years ago, has been slow to catch on among defense contractors, even though improvements in software embedded in weapons systems are crucial for the evolution of those systems’ capabilities, and there is a growing worry that adversaries will use improved software to neutralize the advantages of our platforms.
The F-22 is the workhorse of the U.S. tactical fighter fleet; the Air Force has 187 such planes, delivered from the late 1990s through 2011. The F-22 has a history of large cost overruns during development, and the number of planes purchased was cut back from an original 750 because of affordability problems.
Lockheed, like its weapons system competitors, had continued to use waterfall software development for weapons platforms, which meant that it took 5–7 years to field upgrades to embedded software (including two years to develop requirements). Agile was being tried, but only for some “isolated and small projects,” in the words of Michael Cawood, who is in charge of product development for the F-22 and the earlier F-16.
Then, in 2012, prodded by company engineers who knew about the spread of agile elsewhere in industry, the company moved to try agile for new embedded weapons software. The change was very incomplete, however — Cawood called it “water-scrum-fall.” Sprints were used to develop individual software elements, but product testing was held off until the software was completed, which didn’t really shorten the process at all.
In 2017 a group from the Defense Digital Service and Defense Innovation Unit Experimental (DIUX, now called DIU) came to visit a Lockheed software upgrade planning event. Their reaction was that the approach Lockheed was using was “crazy.” The digital specialists said they would like to help the company, but couldn’t do so unless Lockheed changed. “You’ve got to change or be changed,” Lockheed was told.
One of the central messages was that the company needed outside help. Red Hat, a company best known for its open source technology and skills, had been selling Lockheed off-the-shelf Red Hat tools for open-sourcing software, and the two firms worked a lot together. Red Hat also had an “open innovation lab,” which provided consulting services and open source technology tied specifically to change management. Red Hat suggested that Lockheed use this capability for its agile efforts.
Red Hat brought in three people to teach Lockheed how to implement continuous innovation and continuous testing. In November 2017 the Air Force F-22 program office agreed to allow the effort to deviate from many of the DoD 5000 series acquisition requirements to allow agile. The team would not be required to have an integrated master schedule, traditional deliverables, or earned value. Space was reconfigured to facilitate collaboration.
After piloting the changes for several months, the actual development effort at scale began in August 2018. Lockheed assigned 100 people to start doing agile, working with six Red Hat specialists beside them. “During this period, Red Hat was embedded with us, involved in every scrum and every activity,” Cawood told me. “It wasn’t just a bunch of PowerPoints. You have to learn a new process.”
One of Red Hat’s techniques was to “put a social contract in place, specifying days we will work individually and which days in groups. People agreed not to violate this.” As a Red Hat manager told me, “Devops means you need a variety of constituencies that don’t normally talk together to be talking during development and the build phase. These constituencies were not competitive, but they didn’t understand how to get together.”
So Red Hat needed to help the company transition to greater collaboration. “And we started to have fewer planning events, people were working in real time,” the manager said. At the beginning of 2019 the development team decided to self-regulate, so they made decisions at the team level rather than taking them up the hierarchy.
Now the latest F-22 software upgrade will be rolled out this summer, only a year after the effort started.
There are two good-news stories I see in this agile development effort. One is the role of the DOD innovation units in moving this change forward. I have suggested in earlier blogs that these organizations, contrary to my original worries that short-term kids in hoodies would be rejected as a foreign body, are becoming part of the DoD ecosystem. The large role of DDS and DIUX in jumpstarting Lockheed’s journey to agile is another example of this phenomenon.
The second piece of good news is the role of Capitol Hill on this. Many of us worry that when Congress becomes involved in DOD management issues, legislators’ instinct is to micromanage and to emphasize various blunderbuss assaults on waste, fraud and abuse. But that’s not what happened here. In late 2017 the 2018 DOD authorization bill directed the department to field 4–8 new agile projects, intelligent direction that was oriented towards results rather than box-ticking and that avoided micromanagement. This was a push that helped propel the effort. Score two for good government.
Where are Lockheed’s weapons platform competitors? When DDS and DIUX visited the company in 2017, they said that one of the top five weapons contractors was doing something, and “all the others are racing to be last” to adopt.
Now, Cawood said, “everyone is speaking the language, but not very many are actually doing agile.” Cawood cited for me the Defense Innovation Board document, “Detecting Agile BS,” on things organizations do when they say they are doing agile, but really aren’t — a document I blogged about a while ago. As in the government, it will be interesting to see if this starts to spread faster. Lockheed’s efforts will help.
Published with permission from Fcw.com. This originally appeared in fcw.com: https://fcw.com/blogs/lectern/2019/05/comment-kelman-lockheed-agile.aspx