ZD 25.35: Choosing the Tech Life
I survived 25 years in technology and I have some things to share.
Hard earned lessons from a life in technology. Jump to the Distilled Spirit for a look into the AI bubble; getting the most out of agent mode; Ray Diallo’s leadership principles. Don’t miss the classic hip-hop musical coda.
25 Lessons for 25 Years
I started my first technology job 25 years ago this week. I was a young recovering liberal arts major with few real prospects who was pretty good at computer. Since that fateful day, I have always been working in technology. Roles have changed quite a bit. It has ranged from computer and server janitor work. There was an intense decade-long period of database and application design and deployment. I was involved in what seemed like a long-term side career was also a layer of technology exhibition and event design, and construction management.
It has been a long, strange journey. Here are twenty five lessons from the years of building things worth sharing.
SOLID principles are for system design not just software.
All software is just layers of indirection. You can always add another layer of indirection when needed.
Know your code works. Unit tests are amazing. Testable software is durable software.
Scripted, automated, repeatable is the most powerful design pattern on earth.
Data structures will outlast you and possibly your application. Get them right before you multiply.
Defaults are destiny, pick them very intentionally. Conversely check the settings on everything; don’t take the defaults as designed.
Favor composition over inheritance. Small, sharp tools win battles consistently. Winning battles consistently wins wars.
All abstractions leak. There are always exceptions to any real world data type.
Plans are merely a basis for changes. No plan survives contact with reality. Keep them simple and flexible.
It takes three versions to build something right. The first is figuring out what the problem even is, the second attempt will let you figure out the problem and the third try is to perfect the solution.
Users trust you with their data, don’t break that trust.
Always make sure it works on someone else's computer too.
State is the root of all bugs. Build around composable stateless components and pure functions rather than complex monoliths. Take it one step at a time.
Don’t lose the ship. Never lose control of the control plane in the process of putting out the fire.
Make self-supporting tools with as few external dependencies as possible. Always mitigate your dependencies.
In construction, anything not square is expensive.
The amount of major businesses held together with ancient shell scripts and flat file exchanges is astounding. Perl and BASH will never die.
Rubber duck coding works for more than just coding. Explaining the problem often leads to solving the problem.
Networking is more art than science, bordering on black magic and voodoo. Always be disciplined when you troubleshoot, but don’t forget that it is almost always DNS.
When problem solving, keep Occam's razor in mind.
Hanlon’s razor is great perspective on humanity.
Conway's law explains most organizations.
Postel’s law is great advice in software and in life.
Splurge on things you touch, wear or look at regularly. How it feels matters, build quality shows over time. A great chair, giant monitors and a Keychron are worthwhile investments for your health and sanity.
Never trust comments in code nor handwritten labels on used hardware of questionable vintage.
It has been an interesting ride. I hope these lessons serve you well.
Musical Coda
The Distilled Spirit
Electric Economics
⚡ Electric Slide ()
Anything that can go electric economically will go electric. China has a huge advantage in the manufacturing side. Really huge read on the whole subject.
To be blunt: in the Electric Era, maintaining design leadership without manufacturing leadership is not a coherent strategic position, and one that gets less coherent the better you believe AI will get.
🧨 The Next Energy Crunch Has Arrived ()
Oil output is peaking and things get exciting when oil output peaks.
According to the US Bureau of Labor Statistics the average price of electricity per kilowatt-hour in US cities is now above 18 cents, after hovering around 13-14 cents/kW all throughout the 2010’s. That’s a whopping 30% increase in four years even among friends.
All Things AI
A few things happened in AI this week.
Is the Bubble Bursting?
Is the party over? What is the real cost? Who will pay for it?
💥 Is the AI Bubble Bursting? ()
There has been a perceptible vibe shift but have things really changed?
But I suspect that for many people the “A.I. bubble” is something larger and slightly harder to define than “equity valuations”--a way of articulating and describing the inflated rhetoric and excessive encroachment around A.I. that we’ve all experienced over the last few years.
🧨 Irrational Exuberance (The Rebooting)
Is AI a societal or financial danger?
The AI safety fears in the near term aren’t a rogue agentic AI toaster destroying the world trying to make paper clips, but a massive bubble taking down the stock market it dominates. The economy is currently propped up by data centers and health and social services. That’s hardly ideal.
🌿 AI and the Environment ()
There is a lot of conjecture around AI and the environment. Here is a good guide to the facts on the ground.
You should be wary of anyone saying that an article going into details on one specific environmental issue with AI is “painting an incomplete picture” of AI’s environmental impact, because painting that full picture would require crazy amounts of very detailed writing.
Is AI Taking Your Job?
A new Stanford study made everyone talk about it.
🐤A Primer on the Study ()
One author’s primer on the study, including most of the graphs you have seen across the internet.
While more educated people work in jobs that have higher average AI exposure overall, these findings suggest that less educated workers get less protection from AI with greater experience.
❓ Are They Starting to Take Our Jobs ()
The Zvi looks at the report from a few angles and raises more than a few questions.
There is one place I am very confident AI is making things harder. That is the many ways it is making it harder to find and get hired for what jobs do exist.
🤷♀️ AI and Jobs Again ()
Noah smith was not particularly impressed.
Honestly, I don’t put a lot of stock in this measure of AI exposure. We need to wait and see if it correctly predicts which types of people lose their jobs in the AI age, and who simply level up their own productiveness
What’s Next in AI
What is the next new thing around the corner?
📱 Small Language Models ()
Small language models designed to work on mobile devices are changing the compute paradigm.
Wearables integrate SLM-powered coaching agents, and glasses provide live translation and productivity functions without constant uplink. Federated learning strengthens this design by enabling local training with aggregated updates, improving models without leaking private data.
🧠 Mass Intelligence ()
Economy of scale is getting everyone is getting access to very powerful AI. \
When powerful AI is in the hands of a billion people, a lot of things are going to happen at once. A lot of things are already happening at once.
AI You Can Use
📔 Creating Your Style Bible ()
Good, complete instructions on how to create and use a style guide to get AI writing like you.
Perfection is not the bar. Perfection is a trap. It is like chasing a rainbow. You can get close, and it will look beautiful, but you will never actually hold it.
The win is producing more of your voice in less time. That is what the style bible and proper training give you.
👩💻 Using Claude for Compounding Engineering (Every)
Claude Code is amazingly powerful. Here is how Every uses it successfully and pretty strongly recommends subagents.
A subagent is a lightweight AI program you can spin up for a specific role. Think of them as separate conversation windows with specialized instructions. Each one has its own system prompt, its own memory, and access to the same tools as Claude Code in general.’
🕵️♀️ Agent Mode in the Real World ()
How much can Agent mode do for you?
Why the excitement? Because this isn't just another ChatGPT update. This is the difference between having a smart conversation partner and having a digital employee who can actually do the work while you grab coffee.
Thinking Better
🔟 Ten Leadership Principles from Ray Dailo ()
Ten takeaways from Ray Dailo’s Principles.
🔪 Why People Snitch (Knowable Magazine)
Explore the psychology of tattling, whistleblowing and snitching.
Fun Things
💪 Health Effects of Your Protein Diet ()
The science on your massive protein intake is interesting.
🚘 Ford and the Birth of The Model T ()
Making the Model T changed manufacturing.
🏁 Revolutionizing Drag Racing ()
Inventing top fuel drag racing from the driver’s seat.
The Look
More strange things in the housing market via Sherwood News:
Did you enjoy reading this post? Hit the ♥ button above or below because it helps more people discover great Substacks like this one and it helps train your algorithm to get you more posts you like. Please share here or in your networks to help us grow!