Skip to content
June 17, 2019 · 21 min read

An Elegant Puzzle: Systems of Engineering Management (Author Interview) with Will Larson of Digg, Uber, and Stripe

There’s a saying that people don’t leave companies, they leave managers. Management is a key part of any organization, yet the discipline is often self-taught and unstructured. Management can make the difference between fulfillment and frustration for teams, and, ultimately, the success or failure of companies.

Will Larson’s An Elegant Puzzle tackles the challenges of engineering management–from sizing teams, to technical debt, all the way to succession planning. Drawing from his experience at Digg, Uber, and Stripe, Will Larson has developed a thoughtful approach to engineering management that leaders of all levels at companies of all sizes can apply. An Elegant Puzzle balances structured principles and human-centric thinking to help any leader create more effective and rewarding organizations for engineers to thrive in.

In this episode Ledge sits down with the author to discuss key lessons from the book.

Will Larson

Software Engineer at Stripe

Will Larson has been an engineering leader and software engineer at technology companies of many shapes and sizes, including Yahoo!, Digg, SocialCode, Uber, and, since 2016, Stripe.

Read transcript

Ledge: Will, it is great to have you on today. I’m really excited for this conversation.

Will: Yeah. I’m super excited for it as well. Thank you so much for having me.

Ledge: Great. So, because you can do it better than me, if you don’t mind, give a little introduction of yourself and your work. It’s going to be exciting to talk about your current project.

Will: I am the head of Foundation Engineering at Stripe. Stripe is increasing the GDP of the internet, making it easy for folks to get online and run their businesses online and scale them. Foundation engineering is basically data engineering, infrastructure engineering, internal and external developer tools. Really an awesome set of different problems that are some of the most interesting ones, in my opinion, that folks get to work on.

Also came out with a book, An Elegant Puzzle. I’ve definitely never written a book before, although my sister has, so I’m excited to join my sister in the published authors category. A lifelong dream.

Ledge: You guys can share your authorship around the Thanksgiving table now.

Will: She’s still an successful author and I am, I would say they see me like a dilettante, but I’ll try to live up to her standard at some point.

Ledge: Stripe Press published the book, and was cool enough to send me an advance copy. Props to them for a personal note on the little bookmark that came in it. That was super pro. That was really neat to do that. That’s amazing that you have that support from a company.

I got to dive into the book a little bit, being an organization and management and leadership kind of nerd, particularly around engineering topics – everybody who listens to the podcast will know that. I was really struck by, you have created almost like a playbook of some really awesome, tactical stuff.

I have the sense that someone could pick this book up and start to implement from page one, and that’s hard to pull off particularly in the management press.

Where did all of this come from?

Will: I think when you pick up a lot of management books… There’s a lot of great management books out there, even some of the older ones like ‘The First 90 Days’. These are really good but they tend to focus on fundamentals like, how do you do one-on-ones, how do you manage your manager, how do you do these core career planning, et cetera?

But what I really wanted to share is, also, when you’re a manager there’s a huge number of decisions you have to make, and these decisions really matter. These are decisions that figure out, who will lead these important projects? Will your company have money next year to pay people?

The quality of decision-making that you make as a manager and as a leader is so influential on yourself, but also these teams you’re responsible for supporting and the company as well. We don’t really talk about how to make those decisions and how to think about them.

To me, I want to talk through dozens of different problems I’ve gotten to encounter in my career working at fast-growing companies like Stripe, fast-growing companies like Uber. But also, companies that were maybe fast-growing in a different way, like people getting laid off or leaving. How do you think through some of these problems in a structured way, do the right thing by your team and the company?

Ledge: So much of that came from your experience. I read, I believe this is true, that a lot of this is derived from you have been a disciplined blogger for what, 12 years? Actually written down all your stuff.

It’s the kind of thing that all of us wish. Like, “Wow, I really wish that I kept that log of all those things that I did.” You did it, so that speaks volume for your discipline, that’s for sure.

It makes me think that you had this managerial memoir of so much stuff. If were to do that, I would imagine – I have not done that, not had that discipline – but if I was to do it and I look back 12 years ago to the way that I thought about a given problem, contextualized, based on no experience, based on the opinions and variables that I had at the time, it would be a very different experience for a similar problem today.

I wonder, what kind of arguments did you have with rookie Will, while you made mature Will management book?

Will: That’s a really powerful question. I think sometimes when you look at your advice you gave earlier in your career, you look at it and you realize the advice is still good but maybe you’re a different member of the situation than you thought you were. You thought you were the really thoughtful person rallying against the system, and in retrospect you were the person who was thoughtlessly ignoring the constraints and the needs around you. You’re recasting from the hero to the villain a little bit, with more…

Ledge: Did you read my bio before this? I’m kidding.

Will: Yeah. Right. If you can’t look back at what you’ve done 10 years ago and be, “Wow, I didn’t approach that in a particularly effective or thoughtful way,” then probably you’re not learning that much. If you don’t have some regret, then you’re probably not growing at the rate that you could be.

Ledge: My own experience of experience, if we get meta, is that it becomes less about wisdom than it becomes about neuroplasticity and recognizing the same patterns over and over again. That there aren’t maybe that many fundamental patterns of what people and orgs do together, and that you could pay attention to that and start to predict the future.

You don’t really know why but you can say, “I’ve seen this. It’s likely to behave in some way.” It’s the same like when you cross the street. You know how far… which car can I run in front of or not. I have trouble articulating it better than that and you really have done a good job breaking out specific circumstances.

What are the ones that stand out the most to you? You had to form of book and a narrative and an outline, and I know that’s a lot of work. Which things in there are the most resonant?

Will: Just one quick thought about what you said. To me, the best way to learn is writing things down. Just this act of writing down over and over is, to me… It’s one thing to observe something and it’s another thing to write it down and try to learn from it, or try to articulate it in a way that someone else can read it and doesn’t send you a mean Twitter DM afterwards saying your writing was terrible.

Ledge: Well, you’re going to get that anyway, but you know how that goes.

Will: It is the internet. It’s hard to get away from it, but rarely happens and always deserved when it does. You could always be clearer in the writing.

The things that really resonate as recurring themes for me is that, this idea of reality-based decision making is one that I really come back to. A lot of times, it’s so hard or inconvenient to recognize the reality of situations. That people make plans they know don’t make sense, because it’s easier than having these difficult conversations.

I really find that, if you have the information and you’re willing to confront reality, coming up with a good plan that resonates with people is not that hard. And really, that’s much of what the book is.

Talking about hypergrowth, a good reality people don’t like is, if you hire too quickly you slow down. If you start hiring quickly, the most important factor is the quality of your training. If you skip on that, you slow down and it’s better not to be hiring so quickly.

There’s all these pieces where once you start modeling out the systems there is a reality, but it’s potential a reality that it’s not okay to acknowledge is there. So, having the patience to surface that reality in a way where people can hear it and don’t view you as a dissenter that is just trying to tear things down.

Finding a way to be constructive and positive and also surface the reality so you can build a plan that addresses it, that’s the super power that comes up again and again. It’s also quite hard to do well.

Ledge: Absolutely. There’s a lot of work about this, that we evangelize growth for the sake of growth. Because you want to end up on the end of the growth curve and you want to say, “Up into the right”, and “Exponential,” and all these things.

That’s not easy. The human input to that really makes for a lot of work. Management is not the accidental collection of human problems. It takes a lot of input to do that. So, you end up in these situations where you get team upon team, upon team, upon team, upon team.

I can read your book and go, “I don’t know how I would address a lot of those things. ” Having been a startup guy where, to me, a big team is 20/30 people and I’m happy like that. To think of tens of thousands of people in an operation where you couldn’t possibly know all of them, and you’re trying to articulate the average of what those people think and will behave in a group, how does that not blow your brain up?

How did you take that apart and abstract that enough?

Will: One of the things I love about Stripe is how intentional and thoughtful we are in our decision making. I think that a lot of the best thinking in this book has really been things that I’ve learned here around, how do we systematize?

To your point, there’s a certain point when trying to account for all of the individuals involved in the total planning and the total org structuring, the number of constraints is so high that it’s, I think, impossible.

The best metaphor for, or example of, how complicated management gets sometimes is seat planning. If you think about trying to figure out who’s going to sit where, this nominally doesn’t seem like a very important problem, but everyone has really complicated constraints. As you start balancing them all out it’s not solvable.

So then you’re like, “Okay. Whose constraints do I prioritize, and is that fair?” Well, probably, it’s not fair to prioritize the people who are the most upset about their seat. That’s obviously not the just thing to do. Do you prioritize people who aren’t upset? Do you prioritize people who have legitimate needs? What about efficiency? If people sit next to the people they like, they work closely with on a daily basis, that’s more efficient. Aren’t we here to be productive? How do you do these things?

You definitely have to be careful to, in that situation, design an approach and then implement it. Not try to reverse engineer from what everyone wants. There’s certain scale where individuals’ desires, if you factor them in, they compromise other individuals’ unstated desires.

What I found is, being really systematic with a designing approach that acknowledges the constraints, and then implementing it, and being careful.

If you study, take an intro class to philosophy, there’s this idea of positive and negative freedoms. There’s the freedom to do something and the freedom from something. I think, really, as you start looking at not just ensuring the freedoms of people who are vocal about their needs but also ensuring the freedoms of people who are not vocal but the freedom from having their potential to have a window seat taken away or something. Even in this contrived example of seating, which nonetheless people really get filled with feelings about seating, you start to see interesting microcosm of how to do fair, just, management. Balancing all these different folks and the constraints of, maybe there’s only a of couple window seats to begin with. Maybe there’s not…

Ledge: Fair and just in management. That is not a set of terms that you find in a lot of the literature. Management is about efficiency and productivity and systems and documentation.

I completely agree with you that the human stuff is what makes that difficult. We are not pushing around cogs and sticks and boxes.

Will: People have so much energy they have to bring to work each day. If you have a system that they trust, that behaves in predictable ways and consistently behaves in those predictable ways, they’ll spend a lot less energy fighting that or looking of injustices. Instead, they’ll spend their energy on things that are more valuable, like the project work, et cetera.

In one of the chapters I talk about ‘work the policy, not the exceptions’. Generally speaking, there’s two different strategies for dealing with the person who wants the window seat. One is to be like, “Well, I’m tired of arguing about this. I’m just going to give this person what they’re asking because it’s been really frustrating to deal with them.” I call that working the exceptions because you basically try to massage these frustrating components.

But then, I think this is a failure to adjust or to recognize the negative freedoms of the other folks involved.

So, I really try to take this approach of, how do it take this feedback which is that window seats are prestigious, or perceived as prestigious – even though the sun’s coming in and bouncing off your monitor and they’re worse – and work it into the policy in a way where we can then apply it where people who don’t get rewarded for advocating a lot. Instead, people feel that even if they don’t spend all their time advocating for themselves, the system is going to do right by them.

Seating? Who cares about seating? That’s the small stuff. But getting the important projects that are going to get you promoted, getting promoted, compensation, recognition. All those really core things.

Ledge: And you’re a centimeter away from jumping into diversity, inclusion, all the major topics that are so important now, and that people, certainly across our tech industry but I think everywhere, are grappling with this idea of, “Hey, we need to be inclusive.” And maybe sometimes trying to be inclusive without really even thinking about what does inclusive mean in our context.

Inclusive of what, and how? How do we behave inclusively? Those are all those things that you’re talking about.

Another thing to think about with your analogy is that, maybe somebody wants to sit by the window because there’s nobody behind them and then nobody can look at their screen. I think that there are these hidden motivations of, maybe it’s privacy, maybe it’s fear.

How do you dig into all that, because you can’t get feedback from somebody who doesn’t trust you in the first place?

Will: To your point, even in this seemingly simple example there’s also people who have been made to feel uncomfortable around peers they work with and they don’t want to sit near that person. How do you factor in something like that? Which can be incredible uncomfortable for them to raise but might be the hidden factor. Behind the window seat might be motivated by trying to get out of a situation where they feel uncomfortable, but it’s difficult to even talk about.

Ledge: Yeah, absolutely. So, talk about the whole inclusion stuff and diversity and culture, and how that fits into…

I understand what you’re saying. Creating systems and structures that can be universally trusted feels like a foundational element. What do you build on top of that to start to really impact culture?

Will: When we talk about inclusion and diversity – and, as you said, this is something that is getting more and more thought about – the really exciting thing is we’re starting to see less general, “This is a good thing.” Like grandmother’s and apple pie thinking – we want more of it because it’s good.

There’s a second phase that many companies are still in, which is that, “We want more but this is a structural problem, so there’s really nothing we can do.” Like, “I wish we could help but it’s too big for us. It’s really something we have to solve at the elementary education on up.”

Then we’re getting to this maybe third phase which is, we’re starting to talk about things you can do as a manager that really change the experience of the folks on your team. And don’t distance the problem, don’t say it’s not possible for me, but things you can do.

Two that are really the core of me, or rather the core of how I see successful, inclusive teams, are two different ideas. There’s membership, and then access to opportunity.

When we think about membership, how do you interact with folks where you feel seen? Where you feel included?

There’s lots of problems that some folks have in middle school or high school, which you don’t think exist in the workplace. One of them is having a hard time finding someone to eat lunch with every day. This is a real thing that does happen to a lot of folks in the workplace.

Early on in your management, it can be easy to be like, “Well, that doesn’t seem like something I should be working on, finding you someone to eat lunch with.” And I don’t think you should be literally scheduling someone’s lunch but it is, how do you make sure you have systems where people can meet people across the company?

Something a lot of companies do is having that automatically, randomly, select people to have pairings to go eat lunch together.

It’s great. It’s real. How do you meet people in different functions at a sufficiently large company? How do you break out of your local team? And so…

Ledge: Well, what strikes me too is, how do you create a culture that’s even receptive to automating good behaviors? That immediately speaks to me of, that could be the kind of thing where you would roll your eyes and be like, “This is total crap and nobody cares about it.” Or you could say, “I trust my company so much that I know if they implemented something of that nature, that it’s going to be effective.”

Will: So, I do want to come back to getting access to critical work in a second, but to talk a little bit about what you said, so much of people’s values come from modeling the people who are perceived to be high status.

If you look at the CEO or the head of engineering or the head of product or whatever, if they’re consistently meeting other folks, you’re going to do it. It’s very hard to be like, “This is bogus,” when you see these perceived authoritative folks modeling those behaviors.

They say a lot of leadership is showing up. It’s totally true. It is showing up. It’s going to the ERGs and then listening while you’re there. It’s having lunch with folks that you haven’t interacted with before. Showing up seems simple when you have a lot of time, but once you get really busy it turns out showing up is a real reflection of your values and what matters to you.

Ledge: Absolutely. I probably interrupted some other great thought there.

I think you’re right. It’s the kind of thing, as you have increasing demands on your leadership or management time, that it’s the easiest thing to let go.

Which has a nice analog in, you talked about organization and fast growth. What’s the easiest thing to kill off? Is training, onboarding, soft skills, making sure people feel welcome. You can quickly forget to assign value to those important, maybe not urgent, things until the train runs you over.

Will: I think sometimes that work is referred to as second-shift work. There’s the very legitimate concern that, in many situations, that work is not surfaced in performance reviews, promotions or whatnot.

We’ve done a lot of really neat experiments around that. For example, actually creating data around it and pulling into templates around the performance reviews, where that data is present. Then when we’re doing a calibration to figure out the correct leveling for folks, the second-shift data will be presented. It will be there, or it’s not possible to forget about it. Also, making sure for every career ladder you have explicit goals around this sort of work.

Ultimately, avoiding phrases like ‘second-shift’ is important as well. This is the fundamental stuff of leadership. It’s not second-shift, it’s core. If we’re not doing it, it’s not that we’re not doing extra work, it’s that we’re not doing our core work. Really making sure that’s clear.

Ledge: It’s the absolute hardest stuff to tie a metric or a number. Any kind of quantitative feedback is difficult there. You end up having to say, “Rank this person on how nice they are,” if you’re not careful. I could probably do an entire episode about how to measure the stuff that actually matters and not just measure the stuff that is readily available to you.

Will: I have found, if you try to measure, measurement is fraught in many ways. But I think there are ways to measure like the number of times you facilitated a learning group, attended ERG meeting. There are these different ways to surface this stuff.

None of them are perfect, but I do think fighting the idea of non-measurability and pushing to have something. It’s not that you want to say, “Oh, you facilitated three meetings for this management learning circle, and you facilitated for four,” but you will see someone who is like, “Oh, you facilitated literally 45 of the 50 meetings last year, and you facilitated one.” It’s not the small differences but you’ll see people who are doing extraordinary things that could be easy to miss if you’re not looking for it.

Ledge: Sure. It’s incumbent upon every manager to release the scheduling mandates to allow for that kind of thing. Because particularly in an engineering context, you can very, very easily overwhelm someone’s schedule, where they just simply would never have the ability to do something like that.

Will: I think that’s right. That segues a little bit back to, picking up a thread from a little bit earlier, around project selection. How do you make sure people have access to these opportunities?

It’s definitely true that if you don’t create time for people to get access to these important projects, they won’t have time. So, how do you create buffer for it?

As a good manager, creating that space so that you can have people apply for important projects and that you select them using some sort of rubric. Then helping them free up space.

Flipping it a little bit, sometimes you, as a leader or a manager, are in fact the most experienced leader in a situation. You might have done more than your manager at a given point.

I think, sometimes we expect our managers to set everything up for us but you’re a really active participant in your career as well. Sometimes it’s like, “Hey, this is such an important opportunity, I’m going to free up time to make sure I can do it by doing a worse job of these other things.”

I think being deliberate about the level of quality you’re going to do in different parts of your job, and negotiating with your manager either implicitly or explicitly to create that space, is also a core skill. To make sure that you’re able to focus on the work that’s either personally important to you from a mission perspective, or that you think will be fulfilling and growing for you as well.

Ledge: And I think that’s part of Section 3, right? I didn’t get that far yet but I think you did write extensively about that.

Will: Yes. That’s correct. Section 3.

Ledge: Yeah. I at least read my outline.

I know you need to wrap – and we could do this all day because I’m a huge nerd – but what have I not asked? What do you want people to know about this work? I think it’s important, in the time that I’ve spent with it already, and I’d like people to meaningfully make the time to read what you put together.

So, what did I not ask that you want everybody to know?

Will: I really believe that management is an ethical profession, and that the way to be an ethical practitioner of the profession is by being extremely deliberate and intentional.

Importantly, that this doesn’t make you a worse or slower or less efficient manager. It doesn’t make your teams less efficient. I found that this makes teams gel, I found it makes teams joyous, and I found that it works. The teams do good work.

There’s, so often, these false dichotomies or false tradeoffs, where if you focus on culture or if you focus on gelling teams, it’s a waste. But it’s just not true.

You can have a team that’s doing amazing work, rested, excited, likes each other and can be really high performing. I think that should be the thing we’re all working towards, not making this tradeoff between productivity and the feeling, the experience of being inaudible.

Ledge: I love that. I’m glad I let you wrap it up because that’s far better than what I was thinking.

Will, this is so cool. When is the book out? It’s this week, right?

Will: Yeah. It’s out in five days, on the 28th.

Ledge: 28th of May, 2019. And everybody can get it on all their favorite apps, bookstores, whatever. I will tell everybody, I’m not typically a paper-book person anymore but Will has a lot of really cool diagrams and different color sheets and stuff in there. So, this might be one that you’re going to keep on your bookshelf.

I encourage all the leadership and management and engineering folks out there to grab a copy.

Will, it’s been so cool to have you on. I hope we get to do round two after you’re a famous author.

Will: Yeah. I hope so too, even if I’m not a famous author.

Thank you so much. This has been fantastic.

Ledge: Awesome.