Note: This is a research note supplementing the book Unscarcity, now available for purchase. These notes expand on concepts from the main text. Start here or get the book.

Zero-Knowledge Proofs: The Magic Trick That Saves Privacy

You’re at a bar. The bouncer needs to know if you’re over 21. Under the current system, you hand over your driver’s license, which reveals your name, address, exact birth date, whether you’re an organ donor, and that embarrassing photo from when you thought that haircut was a good idea. All the bouncer needed was a single bit of information: yes or no, over 21.

Now picture a magic ID that says only “YES, OVER 21” and nothing else. No name. No address. No photo. Just the verified fact, cryptographically signed, mathematically provable, utterly private.

That isn’t science fiction. It’s what zero-knowledge proofs do, and they’re about to rewire how trust works in civilization.

What Zero-Knowledge Proofs Actually Are (Without the Math Headache)

A zero-knowledge proof (ZKP) is a cryptographic protocol where one party (the “prover”) can convince another party (the “verifier”) that a statement is true without revealing any information beyond the truth of that statement.

Here’s the classic thought experiment, courtesy of cryptographers Goldwasser, Micali, and Rackoff who invented ZKPs in 1985:

The Ali Baba Cave:
Imagine a cave shaped like a ring, with a magic door in the middle that only opens if you know the secret password. I want to prove to you that I know the password, but I don’t want to tell you what it is.

You stand at the cave entrance. I walk into the cave and go either left or right (you don’t see which). Then you shout out which path you want me to emerge from. If I know the password, I can always come out the side you request, either by already being there or by going through the magic door. If I don’t know the password, I have a 50% chance of being stuck on the wrong side.

Repeat this 20 times. If I emerge correctly every time, the probability that I’m faking is 1 in a million. You’re now convinced I know the password. But you still don’t know what it is.

That’s zero-knowledge: complete verification, zero information leakage.

From Cave Puzzles to Real Infrastructure: SNARKs vs. STARKs

The cave analogy is cute, but the real magic happens in the two dominant flavors of ZKP technology fighting for the future of privacy infrastructure.

zk-SNARKs: Succinct Non-Interactive Arguments of Knowledge

Introduced in 2012, zk-SNARKs are the workhorse of privacy-preserving blockchain. “Succinct” means the proofs are tiny: a few hundred bytes regardless of how complex the computation being proved. “Non-interactive” means the prover sends one message to the verifier, with no back-and-forth required.

There is a catch. Most SNARKs require a “trusted setup,” a one-time ceremony where secret random numbers are generated and then (hopefully) destroyed. If anyone keeps those secrets, they could forge proofs. Zcash, the first major cryptocurrency to use zk-SNARKs, conducted elaborate multi-party ceremonies where participants included people who literally destroyed their computers afterward. Paranoid? Maybe. But when billions of dollars depend on the math, paranoia becomes prudence.

The other vulnerability: SNARKs rely on elliptic curve cryptography, which is theoretically breakable by quantum computers. When (not if) large-scale quantum computers arrive, today’s SNARKs become yesterday’s security.

zk-STARKs: Scalable Transparent Arguments of Knowledge

Introduced in 2018 by Eli Ben-Sasson and colleagues, STARKs solve both problems. No trusted setup is required, since the randomness is publicly verifiable. And they use hash functions instead of elliptic curves, making them quantum-resistant.

The tradeoff is size. STARK proofs are larger (tens to hundreds of kilobytes instead of hundreds of bytes) and verification takes longer. It’s the classic engineering choice: more security and transparency in exchange for more computational overhead.

Feature	zk-SNARKs	zk-STARKs
Trusted Setup	Required	Not required
Proof Size	Tiny (~200 bytes)	Larger (~50KB+)
Quantum Resistant	No	Yes
Verification Speed	Very fast	Fast
Maturity	More adoption	Newer

Both are improving rapidly. Groth16 made SNARKs highly efficient. PLONK introduced universal setups that don’t need to be repeated for each application. STARKs are getting smaller and faster. The field is advancing at the pace of a technology whose time has come.

The $48 Billion Experiment: zkRollups and Ethereum’s Scaling Revolution

This is where abstract cryptography becomes concrete infrastructure.

Ethereum, the world’s programmable blockchain, has a scaling problem. It can only process about 15-30 transactions per second, roughly the capacity of a 1980s credit card network. This creates gas fees that spike during busy periods and makes many applications economically impractical.

Enter zkRollups: Layer 2 scaling solutions that batch thousands of transactions together, execute them off the main chain, and then post a single cryptographic proof back to Ethereum that all those transactions were valid.

The numbers are staggering:

An ERC-20 token approval on Ethereum costs ~45,000 gas. On a zkRollup? Less than 300 gas.
A typical ETH transfer takes ~110 bytes on Ethereum mainnet. On a zkRollup? ~12 bytes.
As of May 2026, Ethereum’s 73 active rollups secure roughly $48 billion in Total Value Locked, with the ZK trio of zkSync Era, Linea, and Scroll holding ~$9.6 billion of that.

Major players include:

zkSync Era: Over 27 million transactions monthly since launch
StarkNet: Using STARKs for quantum-resistant scaling
Aztec Network: The “private world computer,” adding optional privacy at every level
Scroll: Native zkEVM compatibility
Linea: Built by Consensys, fully aligned with Ethereum

The March 2024 EIP-4844 upgrade (proto-danksharding) reduced rollup costs by 5-10x, and the December 2025 Fusaka upgrade added PeerDAS, the data-availability sampling that lets the blob count scale toward 48. That’s a major step toward full Danksharding, not the finish line: full Danksharding’s 2D erasure coding is still ahead. When it arrives, rollups could collectively scale to millions of transactions per second with near-zero fees.

Faster transactions are only part of it. The real shift is what becomes possible when verification is cheap and privacy is default.

Proving You’re Human (Without Proving Who You Are)

The AI age poses an existential problem. How do you prove you’re a real human to an online service without creating a surveillance infrastructure that tracks your every move?

The Worldcoin Experiment:

Sam Altman’s Worldcoin (now just “World”) is the highest-profile attempt at “proof of personhood.” Their solution: iris biometrics captured by a device called the Orb, creating a cryptographic identity called World ID.

The clever part is how they use ZKPs:

Your iris creates a unique hash (the hash, not your iris image)
That hash is checked for uniqueness against all other hashes
If unique, you’re added to a Merkle tree of verified humans
When you need to prove you’re human, you provide a zero-knowledge proof of membership in that tree
The proof reveals nothing about which specific hash is yours, only that you’re one of the verified humans

As Vitalik Buterin analyzed: “Worldcoin is significantly better at preserving privacy than some alternatives.” The concern is trusting the Orb hardware. If the devices are compromised, the whole system fails.

The Growing Proof-of-Personhood Ecosystem:

Worldcoin isn’t alone. Holonym recently acquired Gitcoin Passport to expand proof-of-personhood solutions. Humanity Protocol uses palm biometrics. BrightID uses social-graph trust. Each approach has tradeoffs, but they share a common architecture: prove you’re human, prove you’re unique, reveal nothing else.

The Zero-Knowledge KYC market alone is projected to grow from $83.6 million in 2025 to $903.5 million by 2032, a 40.5% compound annual growth rate. When banks, governments, and platforms realize they can verify identity without collecting identity, the incentives flip.

The Bouncer Scenario Becomes Law:

That opening bar analogy isn’t hypothetical anymore. By law, the EU’s eIDAS 2.0 Regulation requires all 27 member states to offer citizens an EU Digital Identity Wallet by December 31, 2026. The regulation explicitly encourages zero-knowledge range proofs, so a wallet can answer “over 18: yes or no” without disclosing a birthdate, and it’s designed for selective disclosure more broadly, proving you’re a licensed professional or a resident of a given country without handing over the underlying document. Italy began beta testing in early 2026; full production rollout across the bloc is still ahead.

The United States is moving the opposite direction, at least for now. Roughly half of US states enforce some form of age-verification law by 2026, with Texas, Utah, and Louisiana all passing App Store Accountability Acts requiring platforms to verify a user’s age before allowing account creation (Utah and Louisiana’s effective dates have since slipped to 2027 amid implementation fights). None of those state laws require zero-knowledge methods. Most rely on government ID uploads or third-party age-estimation services, the exact “hand over your whole driver’s license to prove one bit of information” pattern this article opened with. The tooling to do better already exists: Google open-sourced ZKP libraries for age assurance in 2026 specifically so platforms could adopt the cryptography instead of the ID-upload pattern. Whether lawmakers require it, or leave it optional and hope platforms choose privacy over convenience, is still an open question.

Private Smart Contracts: The Aztec Revolution

For years, blockchain’s transparency was both feature and bug. Every transaction visible to everyone, forever, which is great for auditability and terrible for privacy.

Aztec Network is changing that equation. On November 19, 2025, 500 sequencers around the world began producing blocks on Aztec’s Ignition Chain, the first decentralized L2 on Ethereum with native privacy. The launch deliberately separated consensus from execution: Ignition was a functioning chain producing blocks, but the smart contract execution layer was held back for a 2-3 month soak, much like Ethereum’s Beacon Chain in 2020. On March 31, 2026, Aztec’s Alpha Network went live, the first Ethereum L2 with full private smart contract execution, written in Noir and settling to Ethereum every 12 seconds. The team shipped honest about the risk: the alpha carried a known critical vulnerability in its proving system, and they told users to “deposit only funds you can afford to lose” until the fix landed. It landed on July 21, 2026, when Aztec’s Alpha V5 went live on mainnet, resolving the critical issue after review by 234+ researchers through a public bug bounty, while proving a fully private transaction in about 2.5 seconds on a laptop and cutting fees below $0.05. The candor mattered as much as the code, and this time the code caught up.

What does “private smart contracts” actually mean?

Private state: Your account balance is encrypted. Others can’t see how much you have.
Private transactions: Senders, receivers, and amounts can all be hidden.
Private computation: Smart contract logic can execute on encrypted data.

The technology uses a hybrid approach: two layers of zero-knowledge proofs, one for privacy and one for compression, nested like a Russian doll of cryptographic verification.

Aztec raised $61 million through a novel Continuous Clearing Auction with 16,741 participants. That funding is also a distributed bet that programmable privacy will become essential infrastructure.

Meanwhile, Zcash’s shielded pool usage exploded from 10% of circulating supply in 2024 to 30% in 2025. The narrative shifted from “privacy coins are for criminals” to “privacy is for everyone.”

ZKPs Meet AI: Verifiable Machine Learning

This is where it gets weird and wonderful.

What if you could prove that an AI model gave a specific output without revealing the model’s parameters, training data, or internal architecture?

That’s Zero-Knowledge Machine Learning (ZKML), and it’s advancing rapidly.

The Problem It Solves:

AI models represent enormous investment. OpenAI isn’t going to publish GPT-5’s weights. But how do you verify that a proprietary model actually made a specific prediction? How do you audit for bias without exposing trade secrets? How do you prove an image was generated by a specific AI (not a human) for content authentication?

Real Applications:

zkLLM: Proving that an LLM produced specific output, enabling authentication of AI-generated content
Federated Learning: Proving that training updates are valid without exposing the underlying training data
Model Integrity: Verifying that a deployed model is the audited version, not a modified copy
Academic Integrity: Proving that student work was not generated by an LLM (by proving it was produced through a verified human process)

A February 2025 survey catalogs the explosion of ZKML research from 2017 to 2024, covering verifiable training, verifiable inference, and verifiable testing.

The implications for AI governance are profound. Regulators could audit AI systems for bias without accessing proprietary models. Users could verify that their data was processed according to stated policies. And AI companies could prove compliance without revealing competitive secrets.

Private Voting: Democracy Without Surveillance

Elections present an impossible-seeming requirement: every vote must be secret (to prevent coercion), yet every election must be auditable (to prevent fraud). ZKPs thread this needle.

zkVoting, a 2024 paper from cryptographic researchers, describes a system with “end-to-end verifiability.” Voters can verify their votes are cast as intended, recorded as cast, and tallied as recorded, all while maintaining complete ballot secrecy.

The architecture combines several cryptographic techniques:

Fake credentials allow coerced voters to submit decoy ballots that are indistinguishable from real ones
Merkle membership proofs let voters prove eligibility without revealing identity
Homomorphic tallying counts encrypted votes without ever decrypting individual ballots

Suffragium goes further, combining ZKPs with Fully Homomorphic Encryption (FHE) for a “trustless and tamper-resistant voting platform.”

Real-world adoption is beginning. Estonia has explored ZKPs for its pioneering digital voting system. Platforms like Voatz and FollowMyVote are integrating ZKP-based verification.

For DAOs (Decentralized Autonomous Organizations), the ElectAnon protocol provides anonymous, self-tallying, ranked-choice voting, enabling governance structures that are both radically transparent in process and completely private in participation.

The Connection to Unscarcity: How ZKPs Enable DPIF

The Unscarcity framework proposes a Distributed Proof-of-Integrity Framework (DPIF), a transparent ledger where every governance decision, resource allocation, and contribution is auditable. The apparent contradiction is obvious: how can everything be auditable without everything being surveillable?

Zero-knowledge proofs are the answer.

Civic Standing Without Surveillance

Civic Standing is your reputation in the MOSAIC, your verified track record of contributions to civilization. But proving your contributions shouldn’t require exposing your entire life.

With ZKPs, you can prove:

“I earned 500 Impact for Community Service” without revealing which specific family you helped
“I have sufficient Civic Standing to vote on this proposal” without revealing your exact score
“I completed my Civic Service” without revealing what you did or where

As the book notes: “Zero-knowledge proofs let you prove you earned Impact for ‘Community Service’ without revealing exactly where you live or which specific family you helped. Credit without surveillance. The math vouches for you without showing your homework.”

The Diversity Guard’s Cryptographic Seal

When the Diversity Guard validates a subjective contribution - poetry that moved a community, care work that healed a family - the consensus is “cryptographically sealed and recorded permanently on the distributed ledger (DPIF).”

ZKPs make this verification trustworthy:

Each reviewer’s assessment is independently committed
The final validation proves diverse agreement without revealing individual reviewer identities
The poet knows their work was validated; the world knows the validation was legitimate; nobody knows who said what

Emergency Powers with Mathematical Expiration

The Emergency Protocol grants temporary authority through “cryptographic tokens with embedded expiration timestamps.” When the timestamp passes, the authorization becomes mathematically invalid, and no political pressure can extend it.

This is cryptographic constitutionalism: rules enforced by math, not interpretation.

The Foundation of Trust Must Be Seen

Axiom II: Truth Must Be Seen is the rule that no decision happens in the dark. Every decision affecting resources or rights must be observable, auditable, and traceable. ZKPs don’t contradict this; they enable it.

The decision-making process is transparent: what inputs were considered, which algorithms processed them, what outputs resulted. But the identity of affected individuals can remain private. You can audit the system without surveilling the citizens.

The crucial insight is that transparency and privacy stop being opposites once you have zero-knowledge proofs. The process stays glass-walled while the persons stay protected.

The Road Ahead: 2026 and Beyond

The zero-knowledge proof market was valued at $1.28 billion in 2024, projected to reach $7.59 billion by 2033 at a 22.1% compound annual growth rate.

Key developments to watch:

Infrastructure Maturation:

Aleo launched mainnet in September 2024, a layer-1 blockchain with native ZKP support
Fermah released a universal proof generation layer, a tokenized marketplace for ZKP computation
Every major zkRollup has decentralization on their 2025-2026 roadmap

Hardware Acceleration:

Specialized chips for ZKP computation are entering production
Cloud providers (including Google Cloud’s collaboration with Aleo) are adding ZKP-optimized infrastructure
Proving times that once took minutes now take seconds

Standardization:

ZKProof 8 (Rome, May 9-10, 2026) immediately preceded Eurocrypt 2026, drawing both academic cryptographers and the rollup ecosystem to the same room
International bodies are beginning to establish interoperability standards
The EU AI Act’s transparency requirements are creating demand for ZKP-based compliance

Post-Quantum Transition:

STARKs and other hash-based approaches are gaining adoption
Hybrid systems that work with both current and quantum-resistant cryptography are emerging
The race to quantum-proof the world’s cryptographic infrastructure is accelerating

The Bigger Picture: Privacy as Infrastructure

Zero-knowledge proofs are civilizationally important for one reason:

Every previous privacy technology has been defensive: encryption, anonymizers, obfuscation. You hide from those who would surveil you. The asymmetry favors the surveillers, who have resources, persistence, and time.

ZKPs flip the asymmetry. They make verification possible without surveillance, let you prove compliance without exposing data, and make privacy the default rather than the exception.

In a world where AI can analyze everything and remember forever, where every digital interaction leaves traces, where the cost of surveillance approaches zero, privacy-by-design isn’t a luxury. It’s the only way to maintain human dignity.

The Unscarcity framework imagines a civilization where AI handles the grunt work of coordination: distributing resources, matching needs with capabilities, maintaining infrastructure. But that coordination requires trust. And trust requires verification. And verification has traditionally required surveillance.

Zero-knowledge proofs break that chain. They enable a governance system that is simultaneously:

Transparent: Every process is auditable
Private: Every person is protected
Verified: Every claim is mathematically proven
Decentralized: No single party controls the truth

Beyond the technical achievement, this is the cryptographic foundation for the next stage of human cooperation.

The math works, the implementations exist, and the market is growing exponentially. The only question is whether we’ll use this technology to build systems worthy of conscious beings, or let it become another tool for the already powerful.

The answer to that question isn’t written in code. It’s written in the choices we make about what to build next.

References

Unscarcity, Chapters 2, 3, 4, 6, and 9
Zero-Knowledge Proof Market Report 2033, Grand View Research
zk-SNARKs vs zk-STARKs Explained, Chainlink
Zero-Knowledge Proofs: STARKs vs SNARKs, Consensys
Zero-Knowledge Rollups, Ethereum.org
The Future of Layer 2 Rollups, ZKP Labs
Worldcoin Proof of Personhood Protocol, World Whitepaper
What do I think about biometric proof of personhood?, Vitalik Buterin
Aztec: The Private World Computer, Aztec Network
Privacy-Focused Aztec Network’s Ignition Chain Lights Up on Ethereum, CoinDesk
Announcing the Aztec Alpha Network (March 31, 2026), Aztec
Aztec Launches Alpha V5 on Mainnet With Faster Private Proving (July 21, 2026), Crypto News
L2BEAT - Ethereum L2 TVS Dashboard, L2BEAT
ZKProof 8 - Rome, May 9-10, 2026, ZKProof
From Aztec to Zcash: The Year Pragmatic Privacy Took Root, The Block
A Survey of Zero-Knowledge Proof Based Verifiable Machine Learning, arXiv 2025
Leveraging Zero-Knowledge Proofs in Machine Learning and LLMs, Cloud Security Alliance
zkVoting: Zero-knowledge proof based coercion-resistant voting, IACR
ElectAnon: Blockchain-Based Anonymous Voting Protocol, arXiv
Zero-Knowledge Proof Wikipedia, Wikipedia

Prove You're 21 Without Showing ID: Zero-Knowledge Math