Tech employment is losing relevance

Tech employment has been bleeding for long enough that it no longer feels like a crisis. It feels like background conditions.

Layoffs recur. Reorgs stack. Compensation bands flatten. Expectations rise while guarantees quietly disappear. After a while, the useful question stops being why is this happening? and becomes something else:

What assumptions does this system require me to accept in order to function?

That question is where this work began.

Dependency creates compliance

The strongest reason people tolerate deteriorating conditions is not loyalty, optimism, or inertia. It’s dependency.

More specifically: fixed, high-overhead dependency.

Housing is the most obvious example. Once your shelter requires continuous income at a certain level, a wide range of downstream behavior becomes predictable. You accept conditions you would otherwise reject. You optimize for continuity instead of correctness.

Layered on top of housing are other couplings:

Energy costs Taxes Time scarcity

Each one narrows the margin. Together, they remove room to maneuver. At that point, most decisions are no longer strategic. They’re reactive.

This isn’t a moral failure. It’s a systems outcome.

Once dependency is established, compliance starts to resemble professionalism.

Oversized systems and lost feedback

There’s a concept in Japanese engineering culture called 人馬一体 (jinba ittai) — often translated as “person and horse as one body.”

One of the clearest modern expressions of this idea isn’t philosophical. It’s mechanical.

While much of the automotive industry chased larger vehicles, heavier platforms, bigger engines, and compensatory complexity, Mazda refused to scale the Miata.

Small car. Rear-wheel drive. Two seats. Minimal excess.

That constraint created a virtuous cycle. Lower weight reduced the need for oversized brakes. Modest power simplified cooling. Fewer compensations meant tighter feedback. The result wasn’t brute force. It was responsiveness.

That’s what jinba ittai actually produces: not power, but control.

Most modern work environments do the opposite.

They add layers to compensate for misalignment:

More process to cover unclear ownership More tooling to compensate for poor feedback More management to correct incentives that don’t line up

Each addition solves a local problem while making the whole system heavier. Eventually nothing feels responsive anymore — not because people are incapable, but because the system is engineered to resist them.

This isn’t a temporary cycle

There’s a persistent belief that tech employment is merely in a downturn — that a policy shift, rate cut, or market rebound will restore prior conditions.

I don’t see evidence for that.

Not because recovery is impossible, but because the incentives that would need to change haven’t.

Capital is still protected. Costs are still externalized. Productivity gains are still captured without restoring stability to labor. Even modest efficiency improvements shrink team size requirements before they replace individuals outright.

This isn’t about AI being brilliant. It’s about it being cheaper.

Smaller teams reduce coordination overhead. Tighter control lowers risk exposure. Combined with global labor arbitrage, interest rates, and R&D amortization rules that punish large experimental groups, the outcome isn’t a dip — it’s a structural squeeze.

That’s why the contraction doesn’t resemble past cycles. It isn’t driven by a single shock. It’s driven by incentives that continue to reward scale without responsibility.

Chosen constraints

There are always options. But meaningful ones usually require a lifestyle change, not a title change.

Some people pursue geographic arbitrage. Others optimize for early retirement math. Those paths work for some.

What I’ve been interested in is different.

I still care about building things. I still value correctness. I want shorter feedback loops and fewer intermediaries between effort and outcome.

That requires chosen constraints.

The kind where tradeoffs are visible. Where feedback is immediate. Where effort maps directly to results.

This isn’t about escaping work or avoiding responsibility. It’s about reducing impedance between what you’re capable of building and what the system allows you to attempt.

What this work is

Arpeggio is an applied engineering lab focused on energy, shelter, and autonomy.

Not as ideology. As practice.

The work asks concrete questions:

How much energy do you actually need? Where do systems fail first? What looks cheap but isn’t? What looks difficult but turns out stable? Which dependencies are worth keeping — and which aren’t?

Some of the outputs are public. Some aren’t. All of them are grounded in real constraints and real failures.

This isn’t for everyone.

You can climb. You can optimize within existing systems. Those are valid paths.

But if you’re noticing that your work extracts more agency than it returns — if the waterline around knowledge work feels like it’s rising — then it may be worth examining which constraints you’ve accepted by default.

Not to drop out.

But to restore a bit of jinba ittai between you and the systems you rely on.