[δ] Most Engineering Leaders Read Wrong (AI Makes It Worse)


The δeltas newsletter

November 17th, 2026

Most Engineering Leaders Read Wrong

↓

Most engineering leaders read wrong, and AI is just making it worse.

You're flooded with docs from your team, stakeholders, and management. It's really tempting to have AI summarize all of this text. And frankly, AI probably wrote most of the text in the first place.

The best leaders read differently. And it all starts with a book from 1940: How to Read a Book, by Mortimer Adler.

Reading is passive. Questions make it active.

Most people who read, even when they read with intent, are still very, very passive. Reading is an inherently passive activity, and passive activities help our brains shut down.

Have you ever read a paragraph, gotten to the end, and realized you have no idea what it actually said, so you have to go back? That's because you're not focused. You're not reading actively.

The key to switching from passive to active reading is to ask questions. Whenever you have a piece of text you want to read, whether it's a book, a paper in a professional journal, an email, a document, or even a Slack message, you need to ask:

  1. What is it about as a whole?
  2. What is being said in detail, and how?
  3. Is it true?
  4. And what of it?

A quick skim answers the first question. The second is where you get into the details, and that's where AI summarization really hurts you, because you won't be able to get into the detail.

These questions focus you. They don't just shift you mentally from a passive activity to an active, creative one. They also keep you laser focused on a question, and on a specific answer you're trying to build for yourself.

Earn the right to disagree

How do you know if it's true? This is another common mistake.

You have to earn the right to disagree. Before you can say "I agree" or "I disagree," you have to earn the right, and the way you earn it is by being able to say "I understand."

You can't say it's true or not true, or that you agree or disagree, before you get yourself up to the level of the person who wrote the document. If you don't understand something, technically or organizationally, you first have to get to a position where you do understand.

Follow up. Be active. Ask questions, read some more, talk to people. Make sure you can explain in your own words what the author of the email, the document, or the Slack message actually meant. Then you can make a decision: I agree, or I disagree.

What you really want to avoid is just passively accepting the information as is. Your job as a leader is to stress-test assumptions and play devil's advocate.

What of it?

The most important question is "what of it?" The way I liked to think about it at Amazon was: So what? Who cares? Does it really matter?

As we're flooded with all of this information, it's our job as leaders to help focus everyone we interact with on what really, really matters.

Say the decision is between two technical options. What's the cost of A versus B? What's the benefit of making a decision? Should we invest more time in gathering more data and running experiments, so we can make a better-informed decision? Or can we just move fast?

Maybe all of this is a lot of really technical work that won't change the top line or the bottom line. It will have no business impact, and it won't make life better for customers, partners, or stakeholders in any way.

That's part of your job. What of it? Does it really matter? Or is it just a waste of time?

There's just no depth

AI is producing so much slop everywhere. People have AI write something just so that someone else's AI can summarize it for them. There's just no depth.

But the best readers demand the rigor. They ask questions. They ask whether the statements that land on their desk are actually true. They don't just take them for granted.

And most importantly, they ask: what of it? Does it matter? With so much text being produced everywhere, that's the most important question we have to ask.

Let's do real work. Work that truly matters.

video preview​

When you're ready, here's how I can help:

Manager coaching: Six months of coaching for the job nobody trained you for.

​

4030 Wake Forest Road STE 349, Raleigh, NC 27609
​Unsubscribe · Preferences​

The δeltas newsletter by Gilad Naor

Weekly tech leadership advice, plus behavioral interview coaching for EMs and Staff+. Learn to manage with AI, become a leader people choose to follow, and land your next job.

Read more from The δeltas newsletter by Gilad Naor

The δeltas newsletter November 10th, 2026 A short story in five commits ↓ A couple of weeks ago I organized a creative writing session with John Crickett. The challenge was to write a short story in the format of a git log. This is my second attempt at this story. $ git log commit 8b04e7d Author: Dan <druby@tidesheet.co> Date: Fri Dec 3 01:17:44 2027 -0800 The AI dropped a settlement. WTF. Fixed manually. I'll look at it tomorrow. commit c5e8190 Author: Priya <psingh@tidesheet.co> Date: Mon...

The δeltas newsletter November 2026 Three interview mistakes managers make ↓ "Tell me about a time someone did not meet your expectations": the biggest mistakes you can make in your answer "Tell me about a time you managed a low performer." If you have a manager interview coming up, you can bet you're going to hear this question. I conducted manager interviews at Meta and Amazon, and I've coached hundreds of managers on how to ace manager behavioral interviews. In this video, I walk through...

The δeltas newsletter October 19th, 2026 The Story of Hard Work & Reward ↓ “I understand you’re not feeling well. I still think it would be best if we can talk in person. Can you make it to the office by 11?” That’s how Dan’s morning started, grabbing a new box of tissues and heading to the car in his sweatpants. Dan’s grogginess disappeared in a flash when Claire started going over his performance review. The Plan would need to be adjusted. The end of 2027 was the goal Dan had set with his...