The Optimization Problem

We optimize everything.

Our businesses. Our schedules. Our supply chains. Our workouts. Our sleep. Our finances. Our relationships. Even our leisure.

More. Better. Faster. Cheaper.

For most of my life, I assumed this was obviously a good thing. Optimization meant progress. Find the inefficiency, remove it, measure the improvement, and repeat.

I’ve spent much of my career doing exactly that.

But lately I’ve started wondering whether optimization itself might be part of the problem.

Not because optimization is inherently bad. It works remarkably well.

Maybe too well.

What Are We Optimizing For?

Every optimization requires an objective.

A business might optimize for profit. A factory for output. A website for engagement. A transportation system for throughput. A person for productivity.

Once we choose the objective, we get very good at improving it.

The problem is everything we didn’t choose.

A company optimized for efficiency may eliminate the extra people and capacity that once made it resilient.

A social network optimized for engagement may discover that outrage holds our attention better than contentment.

A food system optimized for cost, shelf life, and taste can produce inexpensive food that we find almost impossible to stop eating.

A workplace optimized for productivity can fill every available hour with meetings and tasks until there is no time left to think.

None of these systems necessarily failed.

That’s the uncomfortable part.

They may have succeeded.

They produced exactly what we asked them to produce.

The Things That Don’t Fit in the Spreadsheet

Optimization favors things we can measure.

Revenue. Output. Cost. Time. Clicks. Miles. Calories. Heart rate. Sleep scores.

But many of the things that make life worth living are notoriously difficult to measure.

Trust.

Meaning.

Beauty.

Curiosity.

Belonging.

Resilience.

Contentment.

A conversation with no purpose.

An afternoon that produces nothing.

A person who isn’t particularly useful to you but whose company you enjoy.

Once we build systems around measurable outcomes, these things don’t necessarily disappear because anyone decided they were unimportant. They disappear because they weren’t variables in the equation.

We optimized around them.

And increasingly, we’re doing it to ourselves.

We track our sleep so we can sleep better. Exercise so we can live longer. Manage our time so we can accomplish more. Invest so our money works harder. Listen to podcasts at 1.5x because apparently even listening has become too inefficient.

None of these things is unreasonable.

I do many of them.

But at some point it becomes worth asking what all this optimization is ultimately optimizing for.

Optimization Made Sense in a World of Scarcity

For most of human history, efficiency wasn’t an obsession. It was survival.

Food was scarce. Energy was scarce. Manufactured goods were scarce. Information was scarce. Human labor was scarce.

Waste had consequences.

Producing more with less meant more people could eat, travel, learn, own things, and live longer.

Optimization helped create the extraordinary material prosperity much of the world enjoys today.

But something interesting happens if the technological trajectory I’ve been writing about continues.

AI makes intelligence inexpensive.

Robotics makes labor increasingly abundant.

Renewable energy, storage, and eventually new forms of generation could make energy dramatically cheaper.

Automated manufacturing reduces the human effort embedded in physical goods.

Advanced recycling and materials science could reduce constraints on raw materials.

We may be building technologies capable of removing many of the scarcities that made relentless optimization rational in the first place.

And yet our instinct is to use those technologies to optimize even harder.

More productivity.

More output.

Faster decisions.

Lower costs.

Fewer workers.

More engagement.

More growth.

If AI allows us to accomplish an eight-hour day’s work in two hours, our first instinct seems to be figuring out how to produce four times as much.

But perhaps the more interesting possibility is that we only need to work two hours.

AI Could Be the Ultimate Optimizer

This is where AI becomes especially interesting.

Computers don’t get bored optimizing.

They don’t decide that 97 percent efficiency is probably good enough and go outside.

Give an intelligent system an objective and increasingly capable tools, and it can pursue that objective at a scale and speed humans never could.

That could be enormously beneficial.

It could also expose a problem we’ve been able to ignore because humans were relatively inefficient optimizers.

If the objective is wrong, even slightly, greater intelligence doesn’t necessarily solve the problem.

It may simply get us to the wrong destination faster.

The important question, then, may not be how powerful AI becomes.

It may be what we ask it to optimize.

And perhaps even more importantly, what we deliberately refuse to optimize.

The Luxury of Inefficiency

There is a strange possibility hiding inside the idea of abundance.

Maybe inefficiency becomes a luxury.

If machines can produce nearly everything we need, perhaps we don’t need to squeeze productivity from every human hour.

If goods become inexpensive, perhaps everything doesn’t need to be manufactured in the fastest possible way.

If intelligence becomes abundant, perhaps we don’t need every thought summarized into three bullet points.

We might cook something that takes four hours even though a machine could make it in four minutes.

Grow vegetables that cost more than the ones at the store.

Drive the long way home.

Build something badly ourselves instead of having a robot build it perfectly.

Spend an afternoon talking to someone without accomplishing anything.

Read a book slowly.

Get lost.

Waste some time.

Today those things can feel inefficient.

In another context, they might simply be called living.

Maybe Optimization Was Never the Goal

I don’t think the answer is to stop optimizing.

Optimization gave us too much to pretend otherwise. And there are enormous problems in the world that desperately need better solutions.

But perhaps optimization is a tool rather than a value.

We became very good at asking:

How can we make this better, faster, cheaper, and more efficient?

The next question may be harder:

Better for what?

If scarcity defined most of human history, optimization was a rational response to it.

But if we are actually approaching a world in which some forms of scarcity begin to disappear, we may discover that the habits scarcity taught us have outlived the conditions that created them.

Perhaps progress isn’t optimizing everything.

Perhaps progress is reaching the point where we can finally afford not to.