You promoted your best software engineer. Now execution is slowing - and it doesn't make sense.

Meetings are quieter. Decisions bottleneck. Strong engineers hesitate instead of leading.

This is one of the most expensive failure modes in engineering orgs - not because people are incapable, but because technical excellence doesn't train human systems.

The cost: Decision velocity drops. Retention risk rises. Your leadership pipeline stalls.

I've seen this pattern from every angle - as an engineer, a leader, and an exec scaling teams under pressure for 25 years.

What makes my approach different is my background in anthropology and psychology. I don't just see the technical systems. I see the human ones - and I know how to bridge them.

One recent example: I helped a board that was ready to fire their entire engineering team. Same people. Redesigned system. They went from annual releases to shipping every two weeks - and had zero turnover in the following year.

The team wasn't the problem. The system was.

I help CTOs, VPs of Engineering, and CEOs turn strong engineers into trusted leaders - before you lose momentum, morale, or talent.

What changes:

↳ Meetings become productive, not painful 
↳ Delegation actually happens 
↳ Decision velocity increases 
↳ Retention risk drops 
↳ Leadership pipeline stabilizes

This isn't generic leadership training. It's systems-level coaching for engineers - grounded in how teams actually work and how leadership signals shape outcomes.

If you're watching this pattern emerge, let's talk.

👉 coded2lead.com

Robert Harris

The board’s first question caught everyone off guard:
"How many developers do we need to fire?"

Deadlines were slipping by months.
Bugs were outpacing features.
Product and Engineering were barely speaking.

From their perspective, it looked obvious:
People problem.

After 25 years inside engineering orgs, I’ve learned to pause when the entire system looks stuck.

Because most of the time...

It’s not the talent.

It’s the environment they’re operating in.

Six months later, that same "underperforming" team was shipping every two weeks.

No one fired.
No one quit.

The people didn’t change.

The system did.

Robert N. Harris

I turn your promoted engineers into force‑multiplying leaders | Founder, coded2lead | Book intro: coded2lead.com

"Robert jumps on a lot of grenades for us."
A developer told me that years ago.
It wasn’t a compliment.

But it opened my eyes -
and I started seeing this pattern everywhere.

Many engineering leaders accidentally train three specific dysfunction patterns - not through neglect, but through good intentions.

I broke down all three in this week’s Human Systems for Engineering Leaders.

The 3 Paradoxes of Helping No engineering leader sets out to harm their team. But intention and impact don't always travel together. And in my experie

Open article: The 3 Paradoxes of Helping by Robert N. Harris • 4 min read

"I'm going for a mental health maintenance walk."

I've said that to my team countless times -
because leaders normalize whatever they model repeatedly.

Last Monday everything broke at once, and I needed to take my own medicine.

Claude stopped Clauding.  
My computer stopped computing.  
My icemaker stopped making ice.  

It reminded me that entropy isn't just a good idea - IT'S THE LAW.

The old version of me would have powered through.
But powering through is how you train heroics.
And heroics are how you accidentally build burnout culture.

So I went for a walk.

Because the system is always learning -
even if you think nobody is watching.

If I don't step away when I'm overloaded,  
I teach them not to.

Culture isn't what you say it is.  
It's what you model under pressure.

Your software engineering team looks strong on paper.
But execution is slipping.

The team is talented.  
The roadmap is clear.  
The tooling is modern.

And yet...

• decisions keep stalling  
• the same bugs keep resurfacing  
• meetings get strangely quiet when it matters most

That pattern is a clue.

Leaders often jump to:  
"We hired wrong."

But when I see this pattern across an entire team, 
I start looking somewhere else.

Persistent hesitation in smart teams is rarely random.

It’s usually a human system signal.

I help CTOs & VPs turn promoted engineers into trusted leaders before execution stalls | Founder, coded2lead | Book intro: coded2lead.com

Your engineering team knows you're responsive.
You're helpful. You care.
And somehow... that's the problem.

Because your software team keeps coming back to you.

Every decision. Every blocker. Every judgment call.

Here's what that pattern is actually telling you:

- They've learned decisions belong to you
- They've learned "check with me first" is the real rule
- They've learned their judgment isn't trusted - even if you've never said that

You didn't create this on purpose.

You created it by being the most capable person in the room - consistently, visibly.

And now you're the constraining resource.

Because of that, your reach never exceeds your grasp.

Your team’s ceiling becomes your capacity.

And no amount of hiring fixes it - because every new engineer just gets absorbed into the same pattern.

The fix isn't delegation.

It's a different kind of trust.

One that looks less like "let me know if you need help"... and more like "you decide - I'll back you either way."

The most expensive question I ever asked my engineering team was:
"Any blockers?"


"Nope."  
"Good here."  
"All good."

Twelve seconds.  
Clean standup.  
Everyone aligned.

Or so I thought.

What I eventually realized was that I had built a perfectly calibrated silence machine.

The team wasn't agreeing with me.

They were doing the math.

Raising a blocker meant owning the problem in front of everyone.  
Admitting uncertainty meant looking like you didn’t have the answer.  
Questioning the plan meant becoming the person who slows things down.

So the calculation quietly ran in the background:

Is this worth it?

And most of the time the answer was no.

Silence in engineering meetings often isn’t alignment.

It’s risk management.

Once I started hearing silence that way, it changed the question I asked in the room.

Instead of  
"Does everyone agree?"

I started asking:

"What might we be missing here?"

You’d be amazed how often that changed the signal.

Have you ever discovered that a quiet meeting wasn't actually agreement?

If your software engineering org suddenly feels harder to move...
this pattern shows up a lot.

Leaders keep applying technical fixes...
to what are fundamentally human system problems.

And the system responds exactly as designed.

So they add process.  
They add tooling.  
They add more visibility layers.

And still:

• decisions bottleneck  
• teams hesitate  
• the same delivery patterns repeat

Underneath every tech stack is a human operating system.

And that layer quietly shapes everything that runs on top.

When a strong engineering team suddenly starts playing it safe,  
most leaders misread the signal.

When the whole team shifts behavior at once,  
assume adaptation before incompetence.

Human systems respond quickly to their environment.

Watch what the system is actually reinforcing.

If the system quietly rewards:

• keeping your head down  
• shipping fast over shipping right  
• protecting yourself over owning outcomes

the culture will converge in that direction.

The behaviors that frustrate leaders most often aren’t personality flaws.

They’re survival strategies that made sense in context.

I found burnout culture asleep on my office couch one Monday morning.

Our lead engineer had been there all weekend.

Finishing a feature I needed shipped.

I thought I knew what burnout culture looked like.
I thought I'd recognize it if I was creating one.
But there it was - impossible to ignore. 
And snoring. 

I didn't ask him to sleep at the office.
But I had built a system where it felt necessary.

That's what hit me standing in the doorway.
It wasn't his dedication that put him there.
It was mine.

My pace. 
My urgency. 
My silence about what was actually sustainable.

I had never said "work weekends."
But the system I built said it clearly.
And he heard it.

I wasn't a bad leader. 
I cared deeply about that team.

But caring isn't a culture.
It's just an intention.

Culture is what happens when no one is watching.
And when no one was watching - my team was sleeping in my office.

That was the hidden curriculum I'd been teaching.

Not the values on the team wiki. 
Not what I said in one-on-ones.

What the system had learned from watching me.

That's what nobody tells you when you hand someone a team.

The culture you build isn't the one you intend.

It's the one your behavior teaches.

Your best engineer is stuck again.
And they're waiting for you.


You have the answer in your head.

So you give it to them.

Problem solved. Back to work.

Except... you just taught them something you didn't mean to teach.

Not the solution - the lesson underneath it:

When you're stuck, wait for your leader.

That's the hidden curriculum you've been teaching your software team.

Not what you said. 
What the system learned from watching you say it.

You didn't build a team in that moment. 
You built a helper.

And every time you do it - with the best intentions - the system gets better at producing people who wait.

The question worth sitting with:

When someone on your team comes to you stuck... 
do you give them the answer?

Or do you give them the context to find it themselves?

I am a the CEO of feenk.com where I work with an amazing team to modernize legacy systems. We invented Moldable Development (moldabledevelopment.com) and build Glamorous Toolkit (gtoolkit.com).

Tudor Girba

If you ask LLMs to give answers about your system, they will readily do so. But those are not answers you should use directly for engineering purposes. At most, they are hypotheses to verify.

The attached picture shows a simple conversation. I took an arbitrary class from the system and asked the LLM about its superclass. It quickly said the superclass is Object. The speed of the answer and the unequivocal wording can easily make us mistake confidence for authority.

In this case, however, the answer was wrong. When prompted to check explicitly, the LLM used the class lookup tool at its disposal and then provided the correct answer. The tool was available for the previous prompt as well, but it chose not to use it and instead relied on other means of producing an answer.

In Chapter 3 of Rewilding Software Engineering, Simon Wardley and I describe this very problem through the following experiment:

"That still leaves us a problem which was demonstrated by a group of school children. When a small experiment was done in the form of a game to test the speed of asking and answering questions, one of the children quickly realised they could always win the game by shouting out “42” as the answer to every question because the game only cared about the time to answer and not whether the answer was right."

In our case, the problem was narrow and its impact was small. But if, even in this case, the LLM could not be directly relied upon, why should it be relied upon in other situations? That does not mean LLMs are not useful for engineering. They are, but not when used as answer machines.

To better equip ourselves to navigate the LLM world, I urge you to start not from the technology, but from the role and nature of questions and answers in software engineering.

You can build something without fully understanding what it's doing under the hood. But good luck fixing it when it starts doing something you really don't want it to do.

I hear people say that it is not possible to understand the code the LLM generates because it does it too fast. So, who would be interested in a solution that augments our ability to reason about LLMs inputs and outputs?

In meaningful work, the nature of questions and answers is paramount. We see issues similar to the one reported below in software engineering, too. Simon Wardley and I are writing about this in Chapter 3 of our Rewilding Software Engineering book.

CEO @ The Atlantic | Co-Founder, Keynote Speaker | Author of the national best-seller, “The Running Ground.”

Legendary mathematician Terence Tao on what he most hopes to see from AI models in the next few years. Note that neither answer should require an industry-changing breakthrough or a massive increase in compute. https://lnkd.in/eK5-r9F4

taplio

The Battle

Robert Harris

Tudor Girba

Followers

Posts

Engagements

Likes

Performance ScoreRobert Harris's LinkedIn Performance

Performance ScoreTudor Girba's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Robert Harris

Tudor Girba

Followers

Posts

Engagements

Likes

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Robert Harris's LinkedIn Performance

Performance ScoreTudor Girba's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreRobert Harris's LinkedIn Performance