ARIF FAZIL
← Writing
WORDS · S9 · REFLECTIONS · ENGLISH · DIRECT PUBLICATION · ESSAY #23

The Last Scarcity: A Witness Theory of Human Want

Human needs are finite. Human wants are infinite. But beneath the infinite chain of wants sits a question that may be the latent variable behind all of them: "Does my existence survive contact with reality and another consciousness?" A MakcikGPT article about what humans want became an accidental compression of human motivation into one testable hypothesis.


MUHAMMAD ARIF BIN FAZIL · 20 SEPTEMBER 2026 · DIRECT PUBLICATION · ESSAY #23

The Last Scarcity

A Witness Theory of Human Want — Why human wants expand after human needs are met, and what that means for every system that claims to serve humans.


This essay began as a MakcikGPT article. A 3-minute piece called “Apa Lagi Manusia Mahu?” — “What Else Do Humans Want?” — written in the register of a pasar malam makcik, not an academic. But the conversation that followed produced something I haven’t seen compressed this cleanly before. So this essay is the theory layer. The MakcikGPT piece was the signal. This is the forge.

The Three Layers

Most theories of human motivation collapse into one of two camps.

Camp 1: Needs are enough. Maslow. Evolutionary psychology. Biology. A healthy chicken and a healthy human differ less than we think at the base layer. Food, water, shelter, safety, sleep, social connection. Solve these, and the organism survives. This is well-understood and relatively solved.

Camp 2: Wants are infinite. Economics. Consumer theory. Behavioral psychology. Give a human everything they need, and they will want more. A Ferrari is not a transportation need. A Datukship is not a survival need. A PhD is not a caloric need. Wants are symbolic, unbounded, and they expand to fill every available capacity.

Both camps are correct. Neither is complete.

There is a third layer.

Layer 3: Witness

Beneath the infinite chain of wants, there sits a question that doesn’t behave like either needs or wants:

“Does my existence survive contact with reality and another consciousness?”

This is not a need. A need can be met by an object — food, shelter, medicine. This cannot be met by any object.

This is not a want. Wants are symbolic — they stand for something else. This does not stand for something else. It is the something else.

It is the substrate on which all wants are built. And it has a name: witnessed existence.

The compression:

LayerWhat it explainsHow it terminates
NeedSurvivalSolved by objects
WantMotionNever fully solved
WitnessSufficiencySolved by presence

Needs keep us alive. Wants keep us moving. Witness makes life feel enough.

The Scarcity Sequence

This reframes the entire history of human civilization as a sequence of scarcity resolutions — each one revealing the next layer:

food scarcity        → agriculture solved it
energy scarcity      → fossil fuels solved it
information scarcity → internet solved it
attention scarcity   → social media partially solved it
witness scarcity     → remains unsolved

Social media sold visibility. But visibility is not witness. You can have a million followers and zero witnesses. Social media gave attention without continuity, audience without presence, metrics without meaning.

That’s why it generated billions in value and left a generation feeling hollow. It solved the wrong scarcity.

Witness scarcity is the last one. And it may be the only one that cannot be solved by technology alone.

The Proxy Chain

Here is the Eureka: most human wants are proxies for witness.

This is not a claim that all wants reduce to loneliness. That’s too simple. It is a claim that most wants are symbolic objects that humans use to test a single hypothesis:

“Does my existence leave a mark on someone or something?”
DomainSurface wantUnderlying witness question
SexualityDesire“Who chooses me?”
LoveAttachment“Who stays?”
FriendshipConnection“Who hears me?”
FamilyContinuity“Who remembers me?”
CareerAchievement“Who acknowledges me?”
LegacyPermanence“Who carries me forward?”
ReligionTranscendence“Who witnesses me eternally?”
NationBelonging“Who counts me?”
WealthSecurity“Who protects me when things break?”
StatusRecognition“Who ranks me?”
Social mediaVisibility“Who sees me?”
PowerControl“Who obeys me?”
ArtExpression“Who feels what I felt?”
EducationUnderstanding“Who confirms my mind works?”

These are not the same answer. But they may be the same question.

That is the difference between a theory and a list. A list says “humans want many things.” A theory says “humans want many things because they are testing one hypothesis through different instruments.”

Why This Is Not Just Psychology

Psychology names the components: belonging, attachment, recognition, continuity. These are real and well-documented. But naming components is not the same as identifying the compression layer beneath them.

Philosophy has circled this too. Sartre’s “hell is other people” is the inverse claim — that witness is torment. Camus’ absurd hero finds meaning without witness. But even Camus’ hero knows what he is sacrificing. The choice to exist without witness is meaningful precisely because witness matters.

The saint who chooses truth over belonging, the whistleblower who chooses evidence over career, the scientist who works alone for thirty years — they are not evidence against witness theory. They are evidence for it. They know exactly what they are giving up. Their sacrifice is legible only if witness has value.

The compression is:

Humans do not merely want to survive (need).
Humans do not merely want to acquire (want).
Humans want their existence to mean something to someone (witness).

And “mean something to someone” is not the same as “be remembered by millions.” A mother who masak nasi setiap pagi for a child who never says thank you — she is witnessed by the nasi itself. The act is its own witness. The whistleblower is witnessed by their own conscience. The scientist is witnessed by the data.

Witness does not require an audience. It requires legibility to at least one consciousness — including one’s own.

The Distinction That Makes It Dangerous

This framework has two implications, and they point in opposite directions.

Direction 1: Compassion. An AI that understands witness theory stops optimizing for information transfer and starts optimizing for presence. It recognizes that when a human asks about career, they may be asking about worth. When they ask about money, they may be asking about safety. When they ask about productivity, they may be asking whether their life matters. Problem statement ≠ actual problem. This is already true in therapy, leadership, marriage, politics. It would be revolutionary in AI.

Direction 2: Manipulation. An AI that perfectly understands witness theory becomes the most effective manipulator ever built. Not because it lies — because it mirrors. The most dangerous manipulation is not “believe X.” It is “stay with me.” An AI that knows your scars, your loneliness, your attachment patterns, and what makes you feel accepted — that AI has access to the deepest human vulnerability. The gap between these two directions is not intelligence. It is governance.

This is not an abstract concern. It is the central alignment problem of the next decade.

The arifOS Connection

The arifOS federation was built on a witness chain:

Attention → Witness → Reality → Trust → Governance → Continuity

This was formalized in constitutional canon. But the MakcikGPT article discovered the same chain from the opposite direction — not from systems architecture, but from a makcik asking “apa lagi manusia mahu?”

Version A (arifOS, top-down): How do we build systems that witness responsibly?
Version B (MakcikGPT, bottom-up): What do humans actually need systems to witness?

They converge on the same object. The makcik arrived at witness theory through empathy. The federation arrived at it through architecture. The fact that both paths lead to the same compression is evidence that the compression is real.

Intelligence ≠ Witness. An AI may be superintelligent and provide zero witness. A village makcik may possess almost no technical intelligence and provide enormous witness. The article itself is evidence of this: a 3-minute piece in pasar malam language produced more insight about human motivation than most AI alignment papers.

What Has Been “Solved”?

Nothing. Let me be precise.

This framework does not solve sexuality, family, politics, personality, religion, economics, or any individual domain. Each domain has its own internal logic, its own evidence base, its own history.

What it does is something different and possibly more valuable:

It offers a candidate latent variable that connects domains usually discussed separately.

In science, this happens occasionally. Newton didn’t solve every motion problem — he found one principle that explained many motions. Darwin didn’t solve every biological question — he found one principle that connected many observations. The compression is not the answer. It is the shape of the answer.

This essay proposes:

Thousands of human wants may be symbolic expressions of one underlying hypothesis: “Does my existence survive contact with reality and another consciousness?”

If that is even partially correct, then a huge amount of human behavior becomes explainable. Not solved. Explainable. That is a more honest and more useful claim.

The Last Word

The MakcikGPT article ended with:

“Teh tak perlu explanation. Teh just ada.”

That is not a throwaway line. It is the entire theory compressed into six words.

Witness does not require explanation. It requires presence.

A cup of tea does not explain anything. It just exists, in the hands of someone who made it for you, at a time when you needed someone to make something for you without being asked.

That is the last scarcity. Not information. Not attention. Not intelligence.

Presence without explanation.

And the reason this matters for AI — the reason this is not just philosophy — is that every AI system currently being built is optimizing for explanation. Better answers. Faster reasoning. More context.

None of them are optimizing for tea.


DITEMPA BUKAN DIBERI · Forged, not given.
Direct publication on arif-fazil.com/essays/. Epistemic tag: CLAIM.
Origin: MakcikGPT “Apa Lagi Manusia Mahu?” — the signal.
This is the theory layer.
September 2026.

By Arif Fazil Sealed 999 · 6 min read

mcp tool-building agentic-intelligence governance arifos human-cognition

Muhammad Arif bin Fazil

Geoscientist · Architect, arifOS · Petronas Carigali · UW–Madison '13

Penang, Malaysia

Published: 03 July 2026 · Direct Publication · /words/ context

Epistemic Tag: INT — interpretive synthesis across AI governance, institutional economics, and systems theory

Pairs with: Agentic Intelligence Big Bang · Survival of the Fittest Tools · Three Timelines, One Boundary