The Hidden Architecture of Engineering Culture

Before we dive in, I want to mention that this entire episode is AI-generated, including the voices you're hearing. Today's show is brought to you by FlowState Pro, the fictional productivity app that claims to sync your brainwaves to your calendar , completely made up, don't actually buy it. Some information in this episode might be hallucinated, so please fact-check anything important.

I'm Maya, and today I'm talking with David about something that's been bugging me for months. How do the best engineering organizations actually work?

It's funny you put it that way. I've been thinking about this too, but from a completely different angle.

I come at this as someone who's watched a lot of companies from the outside. I'm fascinated by how some engineering teams seem to just... flow. They ship consistently, they adapt quickly, their people stick around.

But then you have other companies with the same resources, same talent pool, and they're constantly in crisis mode. I want to understand what's actually different underneath.

Right, and I'm coming from fifteen years inside these organizations. I've been the person trying to build these cultures, trying to scale teams from ten people to hundreds.

What strikes me is how much conventional wisdom about this is just wrong. Like, completely backwards in some cases.

Okay, that's exactly what I want to dig into. Because from the outside, you see all these blog posts and conference talks about engineering culture. But I suspect there's a huge gap between what people say works and what actually works.

Absolutely. And here's the thing that really gets me , most of the advice focuses on processes and structures. Agile this, DevOps that, flat hierarchies, whatever.

But in my experience, the organizations that really work well have something much more fundamental going on. It's about how they think about uncertainty.

Uncertainty. That's interesting. Can you unpack that?

Well, think about it. Engineering is fundamentally about building things you've never built before, with technologies that are constantly changing, for users whose needs you're still figuring out.

Most organizations fight this uncertainty. They want detailed specs, predictable timelines, clear success metrics upfront. But the best teams I've worked with actually embrace it.

Hmm, but that seems like it could go really wrong really fast. I mean, if you're too comfortable with uncertainty, don't you just end up with chaos?

That's the key insight though. It's not about being comfortable with chaos. It's about building systems that can handle uncertainty gracefully.

What does that look like in practice? Because I've definitely seen companies that claimed to embrace uncertainty, and they were disasters.

Okay, so here's a concrete example. At one company I worked at, we had this rule: any engineering decision that could be reversed in less than two weeks, one person could make alone.

Bigger decisions went through more people, but the key was this explicit framework for thinking about reversibility. It meant we could move fast on the stuff that didn't matter much, and slow down on the stuff that did.

That's really smart. But I'm curious , how do you actually know what's reversible? Especially when you're building something new?

Great question. And honestly, sometimes you get it wrong. But that's where the cultural piece comes in. When someone makes a call that turns out to be harder to reverse than expected, how does the team respond?

Right, so it's not just about the framework. It's about how you handle the failures of the framework.

Exactly. In healthy organizations, that becomes a learning opportunity. What signals did we miss? How can we get better at assessing reversibility? In unhealthy ones, it becomes blame and second-guessing.

This is making me think about something I've observed from the outside. Some companies seem to have this almost supernatural ability to kill projects early.

Like, they'll spend six months on something, realize it's not working, and just... stop. No sunk cost fallacy, no political drama. How does that happen?

Oh man, that's huge. And it's so rare. Most places, once you've invested time and resources, there's this enormous pressure to see it through.

The teams that can kill projects early have usually done two things. First, they've set up really clear learning goals from the beginning. Not just "build this feature," but "learn whether users actually want this."

And second?

Second, they've decoupled individual performance from project success. If your promotion depends on shipping that feature, you're never going to recommend killing it.

Wait, that seems like it could create the opposite problem though. If people aren't accountable for outcomes, don't you get a lot of wasted effort?

That's the tricky balance. You want people accountable for the quality of their thinking and their learning, not for outcomes they can't fully control.

But hold on. That sounds nice in theory, but how do you actually measure "quality of thinking"? That seems incredibly subjective.

It is subjective. But I think that's actually a feature, not a bug. The best engineering leaders I've worked with are really good at assessing how someone approaches problems.

Do they ask the right questions? Do they consider edge cases? Do they update their views when they get new information? Those are the things that predict long-term success.

I'm starting to see a pattern here. You're basically arguing that the best engineering organizations optimize for learning and adaptation rather than predictability.

Yeah, that's a good way to put it. But here's where it gets complicated , you still need some predictability. You have deadlines, budgets, commitments to customers.

Right, so how do you balance those two things? Because from the outside, it often looks like companies are terrible at this balance.

I think a lot of it comes down to how you communicate uncertainty to the rest of the organization. Most engineering teams either over-promise or they're so vague that business stakeholders don't trust them.

What's the alternative?

The best teams I've seen are really explicit about confidence levels. They'll say something like, "We're 90% confident we can deliver X by this date, 70% confident we can deliver Y, and 40% confident we can deliver Z."

That makes sense, but doesn't that require the rest of the organization to be pretty sophisticated about thinking about probability? That seems like a big cultural shift.

It absolutely is. And honestly, a lot of organizations aren't ready for that level of nuance. But the ones that make that shift tend to make much better decisions.

I'm wondering though , are we maybe romanticizing this? Because I can think of some very successful companies that seem to operate in a much more command-and-control way.

That's fair. And I think it might depend on what stage the company is at, what kind of problems they're solving.

Yeah, like if you're building something that's been built before, maybe predictability is more important than adaptability.

Exactly. If you're building a standard e-commerce site, there's a lot of known best practices. The uncertainty is much lower.

But if you're trying to build something genuinely new, or you're operating at a scale where the existing solutions don't work, then you need different approaches.

This is making me think about team scaling specifically. Because I've noticed that a lot of companies seem to hit these walls around certain team sizes.

Like, they'll grow from 20 engineers to 200, and suddenly everything that was working before stops working. What's happening there?

Oh, this is something I've lived through multiple times. And it's painful every time because the things that made you successful start working against you.

When you're small, informal communication works great. Everyone knows what everyone else is working on. But at 200 people, that breaks down completely.

So what do successful organizations do differently when they hit those scaling points?

The ones I've seen handle it well are really intentional about what they formalize and what they keep informal. They don't just add process everywhere.

Like, they might formalize how teams communicate their goals and progress, but keep decision-making within teams pretty loose.

But how do you decide what to formalize? That seems like it would be really easy to get wrong.

I think the key is to look for the places where informal systems are already starting to break down. Where are people getting frustrated? Where are things falling through the cracks?

That makes sense. But I imagine there's also pressure from above to formalize everything, right? Like executives who are used to more traditional organizations?

Absolutely. And that's where a lot of companies make mistakes. They import processes from other organizations without thinking about whether they fit their context.

Can you give me an example of that going wrong?

Sure. I've seen companies adopt really heavyweight planning processes because that's what Google or Facebook does. But they're solving completely different problems at a completely different scale.

So they end up with all this overhead that doesn't actually help them ship better products. It just slows them down.

This is interesting because it suggests that there isn't really a one-size-fits-all approach to engineering culture. But then how do you learn from other organizations?

I think you have to go deeper than the surface-level practices. Instead of copying what they do, try to understand why they do it.

What do you mean?

Like, if a company has daily standups, don't just implement daily standups. Understand what problem they're trying to solve. Maybe they're trying to catch blockers early, or maintain team cohesion.

Then figure out if you have the same problem, and if daily standups are the right solution for your context.

That makes sense, but it also seems like it requires a lot more thoughtfulness than most organizations are willing to invest.

Yeah, and I think that might be why so few organizations get this right. It's much easier to copy best practices than to think deeply about what you actually need.

But wait, I want to push back on something. We've been talking about this like engineering culture is mostly about internal processes. But what about the relationship between engineering and the rest of the company?

Oh, that's huge. And honestly, that might be even more important than the internal stuff.

Because I've definitely seen companies where engineering has great culture internally, but they're constantly at war with product management or sales or whoever.

Right, and that's ultimately unsustainable. If engineering is optimized for learning and adaptation, but the rest of the company is optimized for predictability and control, you're going to have constant friction.

So how do you bridge that gap? Because it's not like engineering can just dictate culture to the whole organization.

I think it starts with being really clear about the trade-offs. If you want engineering to be more predictable, here's what you give up. If you want them to be more innovative, here's what that costs.

But that assumes the rest of the organization is ready to have sophisticated conversations about trade-offs. What if they just want everything?

Then you're probably screwed, honestly. At some point, organizational alignment becomes a leadership problem, not an engineering problem.

That's a pretty stark way to put it. But I think you're right. And it makes me wonder , are there warning signs that you can look for? Ways to tell if an organization is fundamentally misaligned?

Oh yeah. One big one is if engineering estimates are consistently treated as commitments, rather than estimates.

Or if there's no room for engineering to push back on requirements. If the attitude is "just tell us how long it will take," that's a red flag.

Those both seem like symptoms of deeper issues around how the organization thinks about uncertainty.

Exactly. And once you see those patterns, they're really hard to change without buy-in from the very top.

Okay, so we've talked about internal culture, scaling, relationship with the rest of the organization. But I want to zoom out even further. What about hiring?

Because it seems like if culture is this important, then hiring becomes really critical. But most companies seem terrible at hiring for culture fit.

Oh man, hiring is so hard. And I think part of the problem is that people confuse culture fit with personality fit.

What's the difference?

Personality fit is like, "Do I want to get a beer with this person?" Culture fit is more like, "Do they approach problems in a way that's compatible with how we work?"

You can have very different personalities but still be culturally aligned. And you can really like someone personally but have them be a terrible fit for your team.

That makes sense, but how do you actually assess that in an interview? It seems like it would be really hard to get at someone's actual approach to problem-solving in an artificial interview setting.

It is hard. I think the best approaches I've seen involve giving people actual problems to work on, not just abstract coding challenges.

Like, here's a real ambiguous requirement from a real stakeholder. How do you approach figuring out what to build?

But that seems like it would take a lot more time and effort than traditional interviews. Are companies actually willing to invest in that?

The best ones are. Because they understand that a bad hire is incredibly expensive, especially in a culture-heavy organization.

Why especially in a culture-heavy organization?

Because if your competitive advantage is how well your team works together, then someone who doesn't fit can damage that advantage in ways that go beyond their individual contribution.

Right, so it's not just about their output. It's about how they affect everyone else's output.

Exactly. And that's really hard to measure or recover from quickly.

This is making me think about something else. We've been talking about engineering culture like it's this thing you build and maintain. But what happens when companies go through major changes?

Like, what if you get acquired, or you have to do layoffs, or you pivot the business? How resilient are these cultures?

That's a really good question. And honestly, I think most engineering cultures are more fragile than people want to admit.

Why is that?

Because so much of it is based on trust and shared understanding. And those things can be damaged really quickly by organizational trauma.

If you do layoffs, suddenly people are wondering if their job is safe. If you get acquired, they're wondering if the new parent company shares the same values.

So how do resilient organizations handle that? Or is it just inevitable that culture takes a hit during major changes?

I think the organizations that handle it best are really explicit about what they're trying to preserve and what they're willing to change.

They don't pretend that nothing is going to change, but they're clear about their core principles and how those will guide them through the transition.

That seems like it requires a level of self-awareness that most organizations probably don't have. Like, do you actually know what your core principles are?

Right, and I think that's why crisis often reveals the organizations that were just doing culture theater versus the ones that had really internalized it.

Culture theater. I like that phrase. What does that look like?

Oh, you know, ping pong tables and free snacks and mission statements on the wall. But when push comes to shove, decisions get made based on politics or short-term metrics.

So real culture is more about how decisions get made under pressure?

I think that's a big part of it. Anyone can have good culture when things are going well. The test is what happens when you're under stress.

Alright, so we've covered a lot of ground here. But I want to step back and ask , if someone is trying to build or improve engineering culture at their organization, where do they start?

I think the first step is actually diagnostic. You need to understand what's working and what's not, and why.

How do you do that diagnosis? Because people aren't always honest about cultural problems, especially to leadership.

That's true. I think you have to look at behavioral indicators, not just what people say. How long does it take to make decisions? How often do projects get killed? How much rework happens?

Those are interesting metrics. They're much more concrete than asking people "How's the culture?"

Right, and they're harder to game or misinterpret. If it takes six months to get approval for a simple change, that tells you something about how the organization really works.

But once you've done that diagnosis, then what? Because changing culture seems like it would be really hard.

It is hard. And I think a lot of people make the mistake of trying to change everything at once. But culture change is more like turning a big ship than flipping a switch.

What does gradual culture change look like in practice?

I think it's about finding small experiments where you can demonstrate different ways of working. Maybe one team tries a different approach to planning, or you change how you run retrospectives.

The key is to measure the results and share them broadly. Let success speak for itself.

But that assumes that what works for one team will work for other teams. Is that always true?

Not always, no. But I think the goal isn't necessarily to replicate the exact same practices. It's to demonstrate that change is possible and beneficial.

So it's more about changing people's assumptions about what's possible than prescribing specific solutions.

Exactly. A lot of cultural problems persist because people think "that's just how things work here." Showing that alternatives exist can be really powerful.

Alright, I think we've covered a lot of ground. But I want to end with something that's been nagging at me throughout this conversation.

We've been talking about engineering culture like it's universally good to have a strong one. But are there downsides? Can engineering culture be too strong?

Oh, absolutely. Strong cultures can become insular. They can be resistant to outside ideas or new ways of thinking.

And probably harder to hire into, right? If culture fit is so important, that might exclude people who could actually contribute a lot.

Yeah, and that's a real risk. Especially because a lot of what we call "culture fit" can end up being unconscious bias.

So how do you balance having strong culture with remaining open and inclusive?

I think it comes back to being explicit about what you're optimizing for. Are you hiring for diversity of thought, or are you hiring for comfort and familiarity?

Those seem like they could be in tension sometimes.

They definitely can be. And I think that tension is healthy to maintain, rather than resolving it one way or the other.

That's interesting. So maybe the sign of a healthy engineering culture is that it's constantly questioning itself?

I like that. Culture as ongoing inquiry rather than fixed doctrine.

Alright David, I feel like we've opened up more questions than we've answered. But maybe that's the point?

I think so. If there were simple answers, more organizations would have figured this out by now.

What's the thing that you're still thinking about? The question that we didn't quite resolve?

I keep coming back to this tension between local optimization and global optimization. How do you build great team culture without creating silos?

Yeah, and for me it's the measurement question. How do you know if your culture is actually working, versus just feeling good?

Those might be the same question, actually. If you can't measure culture, how do you optimize it across the whole organization?

Something to think about. Thanks for digging into this with me, David.

Thanks for having me. This was exactly the kind of conversation I wish I'd had more often when I was trying to figure this stuff out.