Sunk Cost in Physics: When the Next Machine Is the Plan

Part 3 of Sunk Cost in Big Science is up. This one is about physics: the Super Collider we stopped, CERN and the Higgs, and why the next bigger machine has become the plan instead of a choice. I also get into grants, consensus, and what stepping back might look like.

A correction on the Super Collider: about $2 billion had been spent when Congress cancelled it in 1993. $18 billion is roughly its projected total in today’s dollars.

Part 1, Too Big to Stop

Part 2, The People and Price of Sunk Cost

Imperial Slave Chapter 10: The Viger’s Delight

Continuing the serialization of Imperial Slave, Book 1 of The Waking World, on my Substack. Chapter 10 today.

In this world I try to create something recognizably reasonable for a world of steam and horsepower, one where technology has changed at a modest rate for centuries, not years like ours. It is a different world than ours. Not worse, not better.

A harbor walk, a cargo-stack chase, and a tavern meeting that costs more than coin.

Read Chapter 10 on Substack.

New to the story? Start with Chapter 1. New chapters go up every Monday, Wednesday and Friday.

Good Enough Already

The CNC cart I’m building in my workshop now. Designed in FreeCAD with Woody, my Grok Bot assistant for this sort of thing. The blue box is a placeholder for the machine.

Why the AI that changes the world may be the ordinary kind

It’s Saturday, and I have a small crew at work.

Woody is helping me with a FreeCAD design for the shop. Spark is helping me with an electronics project, an ESP32 board that will do something useful if I ever stop fiddling with it. Eden is editing my books with me, chapter by chapter, and catching the places where I wrote “feet” in a world that has never heard of them. Blit runs the local AI on the Mac mini on my desk, which makes my cover art. I call them my Grok Bot team.

None of this needs the frontier. I don’t need the model that just won a math olympiad or the one some lab says is a step toward superintelligence. Each of the big models has its own personality and strengths, at least Claude, Gemini and Grok as base models. They have evolved, and their output has gotten more alike over the year or so I’ve used them. But today I care more about the rest of the system, and what I’ve done with it, than the base model. For the work I actually do, any normal version is more than enough. In the end I think being model agnostic will be the rule, as long as the tool has a local memory and a system around it to work with. Personality will still matter, but most of us will gravitate to a good-humored, effective coworker over a sycophant.

The frontier labs can chase whatever they’re chasing. I hope they find it. For creative work, though, I think the race has already passed the point where it matters to people like me. The tools are good enough already. (My first draft said the frontier has probably jumped the shark. I still half believe it.)

The thick tail

When people talk about AI changing the world, they usually mean the top: a machine that out-thinks us, cures diseases, or remakes the economy in a decade. Maybe it will, and maybe it won’t. I’m not qualified to say whether superintelligence will rewrite the foundations of civilization.

I’m more sure about the bottom of the curve, the thick tail. That’s millions of people using good-enough tools for ordinary work. A small business owner writes a better flyer. A retiree finally builds the bookcase he designed in his head twenty years ago. A kid gets a patient explanation of algebra at eleven at night. A guy who’s been at it a while finally gets help with the editing he finds painful and could never hand off before.

None of that makes headlines, but it adds up. My bet is that the ordinary uses will change more lives than whatever happens at the top.

My father’s 1980

Here’s a test I keep running. Think about my father, Hugh, in 1980. Then think about my life now.

Day to day, it isn’t that different. We both get up, go to work or do the work we’ve chosen, drive a car, eat dinner, and talk to the people we love. The main thing that feels different is the angst. Everything is louder now.

When I look at what actually changed between 1980 and 2026, it mostly isn’t AI. I count three things:

  1. There are a lot more of us.
  2. What started in the 1950s kept spreading: cars, communications, and big organizations, the machinery of a modern life.
  3. Much of the world caught up to where the United States was early in the twentieth century. That’s a huge, wonderful change for billions of people. But it’s a catching up to the last century, not a leap into this one.

That’s the pattern. Big technologies don’t usually remake daily life overnight. They spread through ordinary use, one household and one shop at a time, until one day you look back and the world is different and you can’t say when it happened.

I think AI will go the same way. The revolution won’t come from the frontier; it’ll come from good-enough tools in ordinary hands.

Your turn

So here’s my question for you. Does your life actually feel different from your parents’ life at your age? Or does it just feel louder?

Tell me in the comments. I really want to know.

M.A. Harris

The People and Price of Sunk Cost

The second video in my sunk cost series is now up on The Unretired Engineer.

Part 1 looked at how programs keep spending after the numbers stop making sense. This one looks at the people behind those decisions: the careers, budgets, and reputations that get tied to a program, and the price everyone else pays when nobody steps back. I use the F-22 and the B-2 as examples. The Air Force bought 187 F-22s against a plan of 750, and 21 B-2s against a plan of 132.

Next up, I’m taking the same lens to physics and cosmology. If you’ve seen sunk cost play out where you work, tell me about it in the comments.

Imperial Slave, Chapter 7: In the Name of the Empire

Continuing the serialization of Imperial Slave, Book 1 of The Waking World, on my Substack. Chapter 7 today. Standing up to distant power is dangerous, especially when your family may pay for it. This week, that danger comes through the Parchells’ kitchen door. Tina’s father faces an Imperial officer, then a harder question about his daughter’s future.

Read Chapter 7 on Substack

New to the story? Start with Chapter 1

New chapters go up every Monday, Wednesday and Friday.

Imperial Slave, Chapter 6: Six Farrier Row

Continuing the serialization of Imperial Slave, Book 1 of The Waking World, on my Substack. Chapter 6 today. Tina is home, and home is no longer a safe place. This is the family she has to go back to, and the lie she tells them. And someone else has come to Fortbridge.

Read Chapter 6 on Substack

New to the story? Start with Chapter 1

New chapters go up every Monday, Wednesday and Friday.

Imperial Slave, Chapter 5: The Jewel Palace

I’m serializing Imperial Slave, Book 1 of The Waking World, on my Substack. Chapter 5 is out today. Tina trades her own clothes for a sailor’s and leads the captain’s party out at night to dig into the old cellars of the Jewel Palace. She knows the ground and he doesn’t, and for one night that makes her the most useful person in the party.

Read Chapter 5 on Substack

New to the story? Start with Chapter 1

New chapters go up every Monday, Wednesday and Friday.

Imperial Slave, Chapter 4: The Coda

I’m serializing Imperial Slave, Book 1 of The Waking World, on my Substack. Chapter 4 is out today. The captain is hunting for a palace. Tina knows it burned down years ago. What he is really after is a lost document that could change Greenhills’ place in the Empire, and he needs her to find it.

Read Chapter 4 on Substack New to the story? Start with Chapter 1. New chapters go up every Monday, Wednesday and Friday.