THE MONEY IS NOT THE PROOF
THE MONEY IS NOT THE PROOF
At some point, another billion dollars stops making you look powerful. It starts making the unanswered question bigger.
There is a mistake modern technology culture keeps making.
We confuse access to capital with evidence of technical correctness.
Raise enough money and people assume you know something.
Hire enough engineers and people assume you solved something.
Buy enough GPUs and people assume intelligence happened.
Build enough data centers and people assume infrastructure happened.
Put enough people on a stage and call something “the future,” and suddenly everybody else is expected to explain why they don’t believe you.
I think we have this completely backward.
The more resources you consume, the more evidence you owe us.
And Meta is an extraordinary case study.
In 2021, Reality Labs recorded a $10.193 billion operating loss.
In 2024: $17.729 billion.
In 2025: $19.193 billion.
During the first six months of 2026 alone: another $8.647 billion.
Those aren’t estimates from somebody roasting them on the internet.
Those are Meta’s numbers.
And here’s the question that apparently became impolite somewhere along the way:
WHAT DID YOU PROVE?
Not:
What did you announce?
Not:
What prototype did you demonstrate?
Not:
How many engineers work there?
Not:
How expensive is the headset?
Not:
How many parameters does the model have?
Not:
How impressive was the keynote?
What engineering proposition did tens of billions of dollars establish?
Show me the object.
Show me its authority.
Show me its history.
Show me its continuity.
Show me what survives when your infrastructure disappears.
Show me what the user actually owns.
Show me how it moves somewhere else.
Show me how conflicting histories resolve.
Show me what happens offline.
Show me what remains independently verifiable after your company stops answering.
That is engineering.
Money isn’t an answer to any of those questions.
Money just makes your inability to answer them more expensive.
CAPITAL CAN HIDE BAD QUESTIONS
This is the part Silicon Valley desperately needs to understand.
Abundant capital doesn’t merely accelerate engineering.
Sometimes it prevents engineering constraints from ever becoming unavoidable.
If your architecture requires another server, buy another server.
If computation becomes expensive, buy more compute.
If the product isn’t working, hire another thousand people.
If the thesis stops being fashionable, rename the thesis.
Metaverse.
Digital collectibles.
AI.
Superintelligence.
Companions.
Glasses.
Whatever comes next.
You can keep the machine moving for an astonishingly long time when somebody else keeps supplying fuel.
But keeping a machine alive and designing something capable of surviving without the machine are completely different accomplishments.
I learned that distinction because I couldn’t buy my way around it.
I lost infrastructure.
I couldn’t simply throw another billion dollars at the problem.
So I had to ask the question gigantic technology companies were wealthy enough to postpone:
WHAT SURVIVES WHEN THE MACHINE GOES AWAY?
That question changed everything.
It eventually led me away from making another application and toward making continuity portable.
Identity in the object.
History in the object.
Authority attached to the object.
State derived from evidence.
Deterministic succession.
Offline verification.
Portable memory.
Temporal position.
Then I started building worlds on top of it.
Look at Wildz now.
The creature is cute.
Open the repository.
The current release qualification records:
2,739 tests passed. 0 failed.
Why?
Because a living world isn’t a fucking keynote.
It has laws.
Authority has laws.
Ownership has laws.
History has laws.
Movement has laws.
Settlement has laws.
Causality has laws.
Replay has laws.
Temporal ordering has laws.
Creature continuity has laws.
Offline behavior has laws.
Recovery has laws.
If I change something and violate those laws, the machine is supposed to scream at me.
That is what tests are for.
Not applause.
Not valuation.
Not status.
Falsification.
ENGINEERING IS SUPPOSED TO BE HOSTILE TO YOUR OWN BULLSHIT
This is the engineering ethic I think we have lost.
Your job isn’t to make your idea impossible to criticize.
Your job is to construct increasingly brutal ways of discovering whether your idea is wrong.
A serious engineering organization should be able to say:
Here is the claim.
Here is the test.
Here is the artifact.
Here is the failure condition.
Run it yourself.
Instead, technology increasingly gets presented like monarchy.
A tiny collection of extraordinarily wealthy people announces what humanity’s future supposedly is.
The press repeats it.
Markets price it.
Employees build toward it.
Investors finance it.
Everybody starts repeating the vocabulary.
And somehow the sheer quantity of resources behind the claim becomes a substitute for demonstrating the claim.
No.
Your money doesn’t lower the standard of evidence.
It raises it.
If one engineer spends $20,000 and fails, he made a mistake.
If an institution can consume tens of billions pursuing a technological thesis, then eventually the scale itself becomes a legitimate governance question:
What exactly are we learning per billion dollars?
THIS IS ALSO A FIDUCIARY QUESTION
Meta can afford Reality Labs losses because its Family of Apps business is enormously profitable.
That’s precisely why the question matters.
The profitable machine can subsidize enormous experimentation.
Experimentation itself isn’t the problem.
Failure isn’t even the problem.
Engineering requires failure.
The problem is when failure stops producing constraint.
A failed experiment should eliminate possibilities.
It should sharpen the architecture.
It should make the next experiment narrower.
It should leave behind knowledge that prevents you from repeatedly purchasing different versions of the same unresolved assumption.
Otherwise you aren’t buying discovery.
You’re buying permission to continue.
And shareholders should be allowed to ask the obvious question:
What falsifies the thesis?
If another $10 billion doesn’t work, what changes?
What about $20 billion?
$50 billion?
$100 billion?
At what point does the organization admit that scale cannot substitute for architecture?
That isn’t anti-innovation.
That is the fucking definition of disciplined innovation.
STOP TREATING EMPLOYMENT AS TECHNICAL VALIDATION
And I don’t blame every engineer inside these organizations equally.
That’s another convenient simplification.
A brilliant engineer can see an architectural problem and still show up Monday.
Of course they can.
They have mortgages.
Families.
Equity.
Careers.
Teams.
Visas.
Reputations.
Years invested in specialized systems.
The paycheck answers a completely different question from the architecture.
“Should I keep this job?”
is not the same question as:
“Is this the correct technical substrate?”
Yet society looks at thousands of highly compensated engineers continuing to work on something and interprets their continued employment as collective technical validation.
That’s absurd.
They’re employees.
Their continued presence proves somebody can continue paying them.
It does not prove the thesis.
MAYBE THE MONEY SHOULD COUNT AGAINST YOU
So I propose reversing the prestige function.
Don’t tell me how much you raised.
Tell me what you proved.
Don’t tell me how many engineers you employ.
Show me what they made independently testable.
Don’t show me the data center.
Show me what survives when you unplug it.
Don’t show me another AI creature.
Show me its continuity.
Don’t tell me the model remembers me.
Give me the memory.
Let me take it.
Let another intelligence read it.
Let me disconnect your servers.
Let me verify its history myself.
Let me transfer it without asking your database what reality is.
Now we’re talking.
Until then, every additional billion dollars doesn’t necessarily make the project more impressive.
It can make the unanswered question more embarrassing.
Because eventually the little guy isn’t supposed to be standing there with a public repository containing thousands of tests asking the trillion-dollar institution:
Where is the machine?
They had the capital.
They had the engineers.
They had the compute.
They had the distribution.
They had the press.
They had the users.
They had every institutional advantage imaginable.
So stop pointing at the advantages.
SHOW THE RESULT.
Money is not proof.
Headcount is not proof.
Valuation is not proof.
Users are not proof of architectural correctness.
A keynote is not proof.
A press release is not proof.
A billionaire saying “the future” is not proof.
There is an old engineering answer to all of this.
It is wonderfully disrespectful of status.
Show me.
And if you’ve spent tens of billions of dollars?
Good.
Then you should have tens of billions of dollars worth of answers.
Not another announcement.