PUNCH IT

Something extraordinary is happening.

You already know.

You have felt your own cognition change.

Questions that used to stop you don't stop you anymore.

You can begin with something you barely understand and find the vocabulary. Locate the literature. Distinguish competing interpretations. Follow the mathematics. Ask the stupid question you were once too embarrassed to ask. Discover that it wasn't stupid. Discover that the answer everybody repeats doesn't quite answer it. Find the person who knows the piece you don't. Bring their distinction back. Test it against another discipline. Return to the original question changed.

Things that once required extraordinary institutional access can increasingly be traversed through ordinary language.

People separated by disciplines, credentials, geography, culture, money, and technical vocabulary can make their distinctions mutually intelligible.

The accumulated symbolic inheritance of humanity was already here.

More of it is addressable now.

And somewhere in this you have probably had the moment.

This is somewhere between Star Trek and a revolution in empirical science and the territory we used to quarantine under words like mystical, esoteric, impossible, or unknowable.

Not because every strange idea suddenly became true.

Quite the opposite.

Because our capacity to differentiate has exploded.

Something happens when differences no longer need to be suppressed for a larger organization to hold.

Coherence increases.

Not sameness.

Not agreement.

Differentiation without suppression.

And another recognition starts becoming difficult to avoid.

The whole strange planetary mess.

Humanity has spent centuries accumulating knowledge, building instruments, extending communication, automating labor, externalizing memory, increasing productive capability, and connecting more of the planet.

Now billions of biological intelligences can interact through planetary communication while silicon cognition helps them traverse an accumulated symbolic inheritance no individual could contain.

Look at that for a second.

Then look around.

Wait.

What?

They want to stop it?

What the fuck?

Hey, Bernie. We hear you.

On September third, Senator Bernie Sanders and Representative Greg Casar announced legislation intended to permanently prohibit artificial superintelligence and pause advanced AI development until a new federal regulator establishes rules governing it.

Sanders says he is trying to protect humanity from systems humans cannot control. The proposal includes monitoring frontier systems, removing dangerous capabilities, and preventing superintelligence from being developed anywhere in the world. Read the announcement ↗

And here is where this gets interesting.

I don't want to fight Bernie Sanders.

I want to listen to him.

Really listen.

Earlier this year Sanders also proposed stopping new AI data-center construction until safeguards addressed safety, workers, electricity costs, communities, and environmental effects. Read the announcement ↗

Good.

Bring all of it.

Don't suppress a single useful distinction.

Now look at the proposed strategy.

It projects the new capability through the organization producing the danger.

Now run the simulation.

Of course it's terrifying.

We changed the capability.

We preserved the relations.

Then we treated the resulting catastrophe as a property of the capability.

Look again.

We built machines so we could do less work

Take employment.

Automation reduces the human labor required to produce something.

Wonderful.

That's what machines are fucking for.

But put that achievement inside an allocation architecture where losing your job can mean losing access to housing, healthcare, food, education, transportation, and social participation.

What occurs?

A jobs catastrophe.

The productive system succeeded.

The allocation architecture translated that success into threatened access.

Then we blamed the machine.

Notice what happened.

The problem wasn't that less human labor was required.

The problem was that we still required human labor as proof of entitlement to the result.

Listen to that sentence.

Earn.

A living.

Earn your right to remain alive.

We built machines for centuries to reduce necessary human labor.

They're beginning to work.

And our response is:

Oh shit.

What are the humans going to do for work?

Motherfucker.

We finally did it.

Keep going.

The same thing happens with intelligence.

For centuries, advanced cognitive traversal was expensive.

Scarcity around cognition stabilized hierarchies around expertise.

Now silicon cognition radically changes the cost of traversal.

And one proposed response is to establish a ceiling on how intelligent the silicon is allowed to become.

Maybe interrogate the architecture that made intelligence dangerous to distribute.

Because we are not watching another cycle.

This isn't another cycle

Some change is cyclical.

Variation occurs while the organization generating the cycle remains recognizable.

Structural change is different.

Structural change alters the organization through which subsequent occurrences become possible.

We do not return to a pre-silicon-cognition civilization.

Human beings are already learning, writing, coding, investigating, designing, composing, modeling, and formulating questions through sustained interaction with silicon cognition.

We don't get a vote on whether the structural transition began.

We participate in what stabilizes through it.

That's where the real choice is.

And for that, I want to put you in an airplane.

A little storyteller's license enters here.

The aircraft, the broken ribs, the broom handle, the aerodynamic problems, and the crossing belong to history.

The voice inside the cockpit belongs to the telling.

October, nineteen forty-seven.

Chuck Yeager is twenty-four years old.

He has spent flight after flight approaching something no human being has ever done in controlled piloted flight.

Mach one.

The sound barrier.

Not metaphorically.

Not philosophically.

This is his shot.

And two nights before the flight—

horse.

Yeager gets thrown while riding.

Two broken ribs.

Of all the fucking things.

Humanity is approaching one of the great boundaries in aviation history and Chuck Yeager has apparently been defeated by a horse.

Except—

no.

Going to the military flight surgeon risks grounding him.

So he sees a civilian doctor.

Gets his ribs taped.

Keeps the injury quiet.

Shows up.

Because two broken ribs have not changed what he is there to do.

They have changed how he's going to get into the fucking airplane.

And that becomes a problem too.

With his ribs injured, Yeager cannot comfortably get the X-one's hatch closed.

Jack Ridley hands him a piece of broom handle to use as leverage.

Perfect.

Unknown aerodynamic regime.

Experimental rocket aircraft.

Liquid oxygen.

Alcohol fuel.

Two broken ribs.

Broomstick.

Apparently Rule Number Six was already operative.

And none of this means Yeager is ignorant of the danger.

He knows exactly enough to understand that the danger is real.

Previous flights approaching the transonic regime have produced severe buffeting and revealed that conventional elevator control can become ineffective.

The team has learned from that information and found another control relation in the X-one's adjustable horizontal stabilizer.

Now Yeager climbs into the aircraft.

The hatch closes.

The B-twenty-nine carries him into the sky.

The X-one hangs beneath it like a bomb with a man inside.

And then—

release.

The knowledge ends before the sky does

The rocket chambers ignite.

Acceleration drives Yeager back into the seat.

Into the ribs.

The aircraft climbs.

The Mach number rises.

The pressure through the controls.

The vibration through the seat.

The response of the aircraft to his hands.

The wings.

The skin.

The bolts holding the strange little orange rocket together.

Yeager knows airplanes.

He knows this airplane.

And he knows what waits immediately ahead.

Not the answer.

There is no pilot waiting on the radio who has done this before.

No previous human experience from the other side saying:

Don't worry, Chuck. This part feels weird. Keep going.

There is no other side in human memory.

Everything he knows about staying alive in an airplane has been learned below this speed.

All of it comes from this side.

And now he's right against the edge.

The airplane is talking through his hands.

The meter is climbing.

The ribs fucking hurt.

And the thing he came here to discover is a few numbers away.

This is the moment.

You can feel the entire history of caution pressing in one direction:

And there is another fact sitting quietly beneath all of it.

We know what happens when we don't cross.

What nobody knows is what happens when we do.

Yeager didn't climb into this aircraft to confirm the boundary.

He came to find out what was on the other side of it.

Point nine-six-five.

Everything has narrowed to the aircraft.

The controls.

The meter.

The sky.

The thing he came here to do.

So—

Fuck it.

And something explodes.

Not the aircraft.

The assumption.

Miles below, the desert hears the boom.

Mach one.

Humanity has crossed the sound barrier.

And inside the cockpit—

smooth.

The great wall does not continue as an eternal region of increasing violence.

The aircraft flies.

The controls hold.

The sky continues.

The sound barrier was never a place where physics stopped working.

It was a place where our previous understanding stopped being enough.

The airplane didn't break physics.

It revealed more of it.

And that sonic boom rolling across the Mojave was the sound of a boundary in human engineering becoming traversable.

Boom.

Give that a second.

Because our plane is already in the air.

The controls feel different.

Good.

Feel them.

Some people look at the perturbation and say:

See?
The shaking proves we've gone too far.

No.

The shaking tells us where the organization no longer holds under the conditions now present.

That's information.

Use it.

Hold the orientation.

Let everything that occurs increase the resolution with which we proceed.

Then keep moving.

Don't just defeat the threat

Take one of the scariest possibilities on the table.

Biological weapons.

Advanced cognition increases the capacity to traverse biology.

The immediate problem formulation becomes:

How do we stop an AI-enabled biological attack?

Keep the concern.

Now move upstream.

Now the design problem changes.

Increase containment.

Make catastrophic biological attack a progressively shittier strategy.

Harmful intent does not need to magically disappear.

Its capacity to organize catastrophic occurrence does.

That's a harder problem.

It's also a much better one.

Don't solve the representation

Now do computation.

We are told increasingly capable cognition requires enormous data centers.

Then somebody looks at the scaling curves and says:

Maybe we'll put the data centers in space.

All right.

Maybe.

Space is lovely this time of year.

But take one beat before we start firing server racks into orbit.

Computation.

Don't solve the representation.

Find the requirement.

Some assumptions dissolve.

Others don't.

Buckminster Fuller spent much of his life working from essentially this orientation: don't exhaust yourself fighting an existing system; build the model that makes its obsolete functions unnecessary.

We can take that one recursion further.

Don't defeat the doomers.

Bring their distinctions.

We don't need Bernie to lose.

We need his concern to win more completely than his strategy can deliver.

Hey, Bernie.

We hear you.

Now come help us build what actually fucking works.

That is intelligence.

Not a bigger brain in a box.

Relation.

A network increasingly capable of addressing itself.

Think mycelium.

Not because human civilization is literally a fungus.

Although some days the evidence is mixed.

Look at the organization.

Now look at us.

Good.

We need them too.

Silicon doesn't need to replace those distinctions.

Stop asking the machine for the answer.

Ask something much more interesting:

There.

That's the experiment.

It began.

Find each other.

Choose something real.

Something that matters to human life.

Pick something whose current formulation appears trapped inside scarcity.

Then don't presume the inherited problem statement is the problem.

Now look.

Keep what holds.

Release what doesn't.

Then produce a result.

A result.

Something that changes somebody's life.

Whatever closes.

Produce one undeniable result.

And then do the most viciously mischievous thing imaginable.

Now somebody somewhere else doesn't have to copy our answer.

They begin from what we learned.

Their conditions are different.

Their result changes.

Good.

They publish their traversal.

Somebody else begins there.

Now watch what happens to the argument.

And we don't need to defeat scarcity.

Its explanation simply stops describing what is occurring.

That's the game.

Generate a result that changes the conditions of the argument.

Then another.

Then another.

Until somebody eventually looks back at the things we were terrified of losing and realizes:

That's how structural transition stabilizes.

Not by preserving every relation that held before the transition.

Not by denying the turbulence.

Not by surrendering to whatever comes.

By remaining present enough to recognize what holds while the configuration changes.

So, Bernie:

We hear you.

Doomers:

We hear you.

Bring the distinction.

We don't need everybody to agree.

We need the distinctions to remain available to the whole.

Then let's make something that fucking works.

Let's make it work so well that the old argument loses its job.

Let's show everybody exactly how we did it.

And when they improve it—

good.

That was the point.