21 Rules of Thumb for Managing Software People and Teams
Like this article? We recommend
Like this article? We recommend
Like this article? We recommend
We all manage according to rules of thumb. Sometimes we create rules ourselves—perhaps after squeezing through some hard challenge (or failing to) that teaches a big lesson. Sometimes the lesson comes to us already formed, and so succinct that we can communicate it in a phrase or a sentence. At other times, we struggle to boil complexity down to its pith.
Early in our careers as programmers, we both read Fred Brooks' landmark 1975 book The Mythical Man-Month: Essays on Software Engineering. [1] Full of wisdom that's still relevant today, and widely regarded as a definitive work in the art of software project management, the book became an instant classic among programmers. Like many others who read it, we found Brooks' one-line nuggets of wisdom both memorable and highly useful—such as this one, now known simply as Brooks' Law:
Adding manpower to a late software project makes it later.
It's impossible to count the number of times we've used this quote when managing software projects.
The desire to find and share other such memorable rules was the inspiration and driving force behind writing our book Managing the Unmanageable: Rules, Tools, and Insights for Managing Software People and Teams, in which we collected over 300 rules of thumb that we and our colleagues have used to manage programmers and teams.
Benjamin Franklin collected and published one of the first well-known sets of maxims, proverbs, and rules of thumb, known as Poor Richard's Almanack. [2] Around the same time, Thomas Fuller published his own book of proverbs that included this ubiquitous rule for much of life: [3]
A stitch in time saves nine.
Such rules that we learned while growing up and along life's way are often as applicable to managing software development as to the rest of life. Even though programming code is probably the most mutable of materials, more than one software development manager has applied this traditional principle:
Measure twice, cut once.
Nearly everyone has failed to follow this rule at some point, and most of us have lived to regret it. Perhaps none quite so noticeably as the NASA scientists whose Mars Climate Orbiter spacecraft, programmed in metric units, received instructions in English units instead from the ground-based system on Earth, causing the craft to leave its Mars orbit trajectory and enter the atmosphere of the planet, where it disintegrated.
Another traditional rule we like, though less well-known, is drawn from the many that predate our profession by hundreds of years and yet still apply to us:
Life is simpler when you plow around the stump.
Sometimes bad practices are so deeply rooted that you just need to route learning and the introduction of better practices around them.
In the rest of this article, we'll consider eight major areas of management challenges, with a sampling of the 300-plus rules from our book that address these challenges:
• Problem solving
• Hiring
• Communication
• Teamwork and collocated teams
• Delivering projects/products
• Requirements
• Development process
• Managing yourself
Problem Solving
It's bad enough that our business partners pose problems approaching solving world hunger or curing cancer—the vision thing is their job. But when we take on problems—whether theirs or ones of our own devising—without breaking down the complexity into steps and ordering those steps so we can tackle them a few at a time, we do ourselves a disservice.
Ron: At Charles Schwab, my colleagues and I were tackling the challenge of transforming what had been the world's leading discount brokerage into the premier name in online financial services. We quoted this rule frequently:
Rule #1: Don't boil the ocean.
This rule in no way diminished the size of the vision, but it reminded us not to tackle everything at once.
When I shared it with Mickey, he liked it so much that he made it a mantra at Gracenote as well. But he often followed it up with another oft-quoted principle:
Rule #2: Divide and conquer.
This is the key to solving any seemingly intractable problem. This rule is especially appealing since it can be used to tackle any problem, large or small. Keep dividing until you find a part of the problem you can handle, and solve it. Then "pop the other half off the stack where you put it" and solve it, or divide it into two halves and recurse.
Hiring
Shortly after Pixar Animation Studios was spun out of Lucasfilm in 1986, and while Steve Jobs was actively involved in Pixar as CEO, he discussed his hiring philosophy at a management meeting:
Rule #3: A's hire A's. B's hire C's.
Jobs reinforced how imperative it is to hire the right people (the "A's"), because allowing a "B" into the organization erodes the organization much more than just that one wrong hire.
We believe that hiring the right people is a manager's most important responsibility. But how can you make sure you find the right people? Dave Wilson, a software architect who has guided development for Apple, Sun, HP, Portal, and ParcPlace, shared a rule for helping to find great programmers, and both of us have validated it independently:
Rule #4: Hire people who built stuff on their own time, just for fun.
Though we've never limited ourselves to hiring only people who built stuff on their own time, it's a great litmus test for hiring the right people.
Communication
One of the important lessons about managing—and often a hard lesson to learn—is that communication is as much about listening as speaking. Many companies provide manager training in "reflective listening," which focuses on truly listening and then responding with a message that acknowledges the speaker's point of view.
Rules that remind us to listen can be very helpful. Ron first heard this rule of thumb from Kimberly Wiefling, one of the co-chairs of the Engineering Leadership SIG of SVForum, the San Francisco Bay Area's nonprofit association of technology professionals:
Rule #5: We have two ears and one mouth. Use them in this ratio.
Succinct words for a thought that has been around a bit longer: Epictetus, a Greek-born Roman slave and Stoic philosopher (55–135 A.D.), is quoted as saying, "We have two ears and one mouth so we may listen more and talk the less."
In addition to being a good listener, communication is about making time to meet and talk with your staff daily. As Dave Wilson says:
Rule #6: People drift away from project goals quickly if left alone. Good project leaders talk to people almost every day. "What did you do? Why? Did it work?"
Mickey: The long-admired Hewlett-Packard best practice of "managing by wandering around" is one of our favorite management tools. Toward the end of the day, I often found myself wandering among my programmers and stopping to talk to those who weren't "heads-down." These short "one-on-one moments," though public, helped me to connect with how people were doing, what they were working on, what road blocks they were encountering, and even how they were feeling (physically, mentally, emotionally)—all without the overhead of scheduling a meeting and sitting down formally. These shared moments were some of my favorite times—and hopefully my staff's as well.
Ron: One of my managers at Apple, Tim Swihart, said this:
Rule #7: If you're a people manager, your people are far more important than anything else you're working on.
He continued, "If a team member drops by at an awkward time and wants to chat, set aside what you're doing and pay attention. They may be building up the courage to tell you something big. I've noticed this to be especially true when the sudden chatter isn't somebody who normally drops by for idle conversation. It might be as simple as being unable to get a piece of information needed to complete a task or might be as big as an impending divorce, the death of a loved one, or something equally devastating on a personal level to them […] things that can throw a real monkey wrench into your carefully laid schedule. If they know you'll make them your top priority when they drop by, they're more likely to drop by sooner when things are about to go very wrong."
Teamwork and Collocated Teams
How do you get a team to work hard to deliver what you need? We hold as gospel a rule we learned from Agile and Scrum thought leader Mike Cohn: [4]
Rule #8: A collocated team will always outperform the equivalent distributed team.
We've seen this rule hold true time and time again, and we've managed in-house teams accordingly. It's always surprising how much "ambient" information just flows between members of a collocated team. It also breaks down barriers that might otherwise surface between team members. Product marketing managers and programmers are notorious for having "issues" with each other; collocating them can make fast friends even from these creatures of a different ilk. Similarly, Douglas Crockford, the inventor of JSON, and an entrepreneur as well as a software architect, states the following: [5]
Rule #9: I've had companies where there was an antagonism between the development teams and the testing teams. […]It worked much better when we put the two teams together and made the testers responsible for helping the developers to make their programs better.
Unfortunately, in today's world of geographically dispersed virtual teams, collocation may be a luxury you can't afford or a direction you can't foster. To address the additional challenges of time, distance, and culture that such teams will likely have, you need to use every communication method you can muster to make your teams feel less distant and more closely bonded. Skype, IM, email, and regular conference calls can help, but in the end you must redouble your efforts to ensure that communication flows both up and down to succeed.
Delivering Projects/Products
How often have you seen programmers get carried away with what should be done rather than what needs to be done? Delivered regularly and in a good-natured way, this statement from Jamie Zawinski, an early Netscape lead developer, can help to keep teams on task: [6]
Rule #10: You're not here to write code; you're here to ship products.
Say it with a smile, but back it up by "looking underneath the covers" to see the state of the code.
Focus needs to be accompanied by the right level of intensity to avoid burning out your team. If you're pushing them to work nights and weekends, week in and week out, you may be better served by adopting this rule from Pixar CTO Ed Catmull:
Rule #11: Projects should be run like marathons. You have to set a healthy pace that can win the race and expect to sprint for the finish line.
Mickey: At Pixar, Ed encouraged me to manage projects this way, and I've practiced this rule ever since. Ron has done the same since I shared the rule with him. When interviewing new candidates, we find a way to bring up this principle, so that the proper expectations are set: Sometimes the programmer will need to work very hard, sprinting for the finish line and project completion, but we don't expect programmers to sprint all the time.
In addition to managing down (your team), you also have to manage up and out (your boss and your peers) by setting appropriate expectations for delivery. This traditional rule has served us both well:
Rule #12: Underpromise, overdeliver.
It beats any other approach we know. On the other hand, you can't underpromise by very much, or you'll be labeled a piker.
Requirements
In our experience, the first place software development falls down is in requirements. Requirements are almost always ambiguous, and they can be overwritten or over-scoped, run to hundreds or even thousands of pages—or be missing entirely. Where programming management is inadequate or overwhelmed, product management is often equally challenged.
Where requirements are missing or spotty, we like to remind our product-envisioning colleagues of Louis Srygley's rule of thumb:
Rule #13: Without requirements or design, programming is the art of adding bugs to an empty text file.
Identifying the importance and priority of features has long been the bane of many product managers. In the days of waterfall processes, product managers avoided choosing which features to include by writing hundreds of pages of requirements that included everything they could think of. That resulted in projects that took years to finish. Their colleagues quickly realized that to have any hope of seeing their pet features come to life, they needed to shoehorn even more into the next round. All of which ignored the fact that few programmers could read and retain all those pages of features, let alone their interrelationships, dependencies, and nuances. That's part of what led to Agile approaches.
But change can be dangerous; it provides another excuse to duck decision-making. In the era of Agile, some product folks avoid committing by claiming that Agile means never having to provide requirements. Nothing could be further from the truth. Nasos Topakas, VP of Schwab's Mobile Trading Systems, who then headed development at Ofoto, StubHub, and SendMe Mobile, put it this way:
Rule #14: It doesn't matter what software methodology you plan to use; you always need to know the requirements of what you're trying to engineer.
The issues of requirements don't all lie on the product management side. Programming teams and managers too often demand that requirements be frozen and immutable before they begin coding. Ron Mak, middleware architect of the Mars Rover mission's collaborative information portal, shared yet another truth of requirements—at least those based on customer input:
Rule #15: Iteration is critical: Customers don't know what they want until they see it.
Updating a very old nugget of wisdom, "the proof is in the pudding," we take Leverage Software CTO Joseph Kleinschmidt's rule of thumb to heart every day:
Rule #16: In the beginning, everyone will talk about scope, and budget, and schedule, but in the end, nobody really cares about any of those things. The only thing they care about is this: People will love your software, or they won't. So that's the only criterion to which you should truly manage.
Development Process
We have both seen organizations with too little process—and too much:
• Too little process results in nothing getting done because so much effort is required to overcome the chaos.
• Too much process results in nothing getting done because so much effort is required to overcome the process.
We've both learned to surf the line between too little and too much process.
Ron: I had long called myself an advocate and implementer of "just in time" and "just enough" process. Then one of my Schwab colleagues, John Steele, by then a chief QA architect, shared one of his QA rules:
Rule #17: Do as little QA as possible and no less.
I quickly recognized the applicability to process in general, and I paraphrased the rule this way:
Rule #18: Do as little process as possible and no less.
Even as programmers, we had both accepted responsibility from the first, not just for coding products but also for making sure that they worked well. We rail against the mentality that defines "done" as anything short of happy (if not delighted) customers, which we tend to express in terms of customer satisfaction.
That mindset doesn't always get through to the many programmers who first coded in siloed environments, in which their first responsibilities were so limited that they never saw the connection between their work and the customer. Sometimes, with some kinds of products, it can help to let programmers watch over the shoulders of customers as they try (and fail) to use applications—or even task programmers to watch videos of such customer experiences.
When faced with programmers whose customer empathy levels are abysmally low but who "get" numbers and practices from others, we resort to the following rule of thumb from Bruce Rosenblum, CEO of Inera and previously VP of software development at Turning Point:
Rule #19: A great developer spends two hours testing for every hour coding.
We often recognize the need to post this rule prominently.
We also take to heart another of Joseph Kleinschmidt's go-to rules regarding programmers and quality:
Rule #20: The quality of code you demand during the first week of a project is the quality of code you'll get every week thereafter.
Managing Yourself
Ultimately, managing starts with yourself. You have to learn and practice time management, priority management, communications management, follow-up management, and more. We think one of the hardest lessons for new managers to learn is this one, succinctly expressed by David Dibble, Ron's division leader at Schwab, and now an executive VP at Yahoo!:
Rule #21: Don't let the day-to-day eat you up.
He made this statement to his management team to emphasize that managers have "real" work to do. Given the seemingly urgent—email, meetings, office routine—that can easily fill a day, only by being intentional about how we use our days can we managers prevent that from happening, and take charge. Only by managing ourselves can we hope to succeed at managing our responsibilities.
Summary
The rules of thumb and nuggets of wisdom that we've collected have helped us considerably over many years of managing software people and teams. We've seen them bring even the most grizzled programmers up short. We've used them to drive home our points and to deflate heated debates among programmers, as well as within seasoned management teams. When used appropriately, rules of thumb in a few words can communicate more than any amount of prolonged discussion. We hope that the rules we've shared here will help you as much as they have us, enabling you to manage software people and teams more effectively.
Mickey W. Mantle has been developing software for more than 40 years as a software and hardware product creator, manager, and executive for companies that include Evans & Sutherland, Pixar, Broderbund, and Gracenote. He currently develops mobile/tablet applications, writes, and consults.
Ron Lichty has been developing software for 30 years, most of them as a programming manager, director of development, and vice president of products and engineering for companies that include Apple, Fujitsu, Razorfish, and Schwab. He has written four previous books and hundreds of articles. He consults with startups and companies large and small to transform chaos to clarity and make their software development hum.
For more about us and our book, check out our Managing the Unmanageable website.
References
[1] Frederick P. Brooks, Jr., The Mythical Man-Month: Essays on Software Engineering, p. 25.
[2] Benjamin Franklin, Poor Richard's Almanack.
[3] Thomas Fuller, Gnomologia, Adagies and Proverbs, Wise Sentences and Witty Sayings, Ancient and Modern, Foreign and British.
[4] Mike Cohn, Succeeding with Agile: Software Development Using Scrum, p. 355.
[5] Douglas Crockford, as quoted in Peter Seibel's Coders at Work: Reflections on the Craft of Programming. Apress, 2009, p. 123.
[6] Jamie Zawinski, as quoted in Seibel's Coders at Work, p.22.