Updated look back: this article revisits a 2018 story with what has happened since.
In February 2018, MIT Technology Review put zero-knowledge proofs on its list of 10 Breakthrough Technologies under the title "Perfect Online Privacy", and singled out zk-SNARKs as the practical version. zk-STARKs, published a few weeks earlier, did not replace SNARKs, but they fixed the two weaknesses MIT pointed at and now secure a major Ethereum scaling network.
Industry observers note that the short version: STARKs took a large share of the stage, SNARKs adapted, and Bulletproofs found a narrower job. Here is how the three compare and how the story played out.
In a recent development, A zero-knowledge proof lets you prove that a statement is true without showing the data behind it. MIT's examples were proving you are over 18 without giving your birth date, or proving you can afford a payment without showing your balance.
Recent findings suggest that the idea is old. According to ethereum.org, it first appeared in a 1985 paper. What changed was that it became usable. MIT's entry, written by Mike Orcutt, credited much of that progress to Zcash, a currency debuted in late 2016, and listed Zcash, JPMorgan Chase and ING as key players.
The same entry named two drawbacks. zk-SNARKs were slow and heavy to compute, and they needed a "trusted setup". It added that researchers were already looking for alternatives without those problems, but did not name any.
According to recent reports, zk-SNARK stands for zero-knowledge succinct non-interactive argument of knowledge. "Succinct" is the selling point. Zcash's documentation says a proof is only a few hundred bytes and can be checked in a few milliseconds, even for a very large computation.
Industry observers note that that is what makes shielded Zcash transactions possible. The transaction is encrypted on the blockchain, yet the network can still confirm that it follows the rules.
Recent findings suggest that the cost is the setup. Early SNARKs needed a one-time ceremony that produced public parameters. The ceremony also produced secret randomness, and Zcash's own documentation notes anyone who kept it could forge proofs and create counterfeit coins. Zcash ran two multi-party ceremonies, one for its Sprout version and one for Sapling, so that no single person held the secret. Users still had to trust that at least one participant destroyed their share.
The STARK paper, "Scalable, transparent, and post-quantum secure computational integrity", was submitted to the IACR ePrint archive on January 10, 2018 by Eli Ben-Sasson, Iddo Bentov, Yinon Horesh and Michael Riabzev. Its title states the three goals.
According to recent reports, transparent means no trusted setup. STARKs use public randomness, so there is no secret to leak.
It has been revealed that post-quantum refers to the underlying math. Ethereum.org explains that SNARKs depend on elliptic curve cryptography, which a future quantum computer may break, while STARKs rely only on collision-resistant hash functions.
Scalable means the work of checking a proof grows far more slowly than the computation being proved.
Industry observers note that the trade-off is size. STARK proofs are larger, and ethereum.org notes that this makes them more expensive to verify on Ethereum. Figures quoted in the Ethereum Stack Exchange thread that cites this article give a sense of scale: about 288 bytes for a SNARK against about 45,000 bytes for a STARK in one comparison, and "hundreds of kilobytes" in one developer's real project. Those numbers depend heavily on the computation, so treat them as orders of magnitude.
Recent findings suggest that bulletproofs came out of Stanford in 2017, from Benedikt Bünz, Jonathan Bootle, Dan Boneh, Andrew Poelstra, Pieter Wuille and Greg Maxwell. Like STARKs, they need no trusted setup.
They were designed for one job: proving that a hidden amount in a confidential transaction falls within a valid range. The Stanford project page notes they cut that proof from over 10 kB to under 1 kB. It also says a Bulletproof takes longer to verify than a SNARK, which is why they are not used for proving large computations.
Sources indicate that monero added Bulletproofs in its October 2018 network upgrade. Its documentation reports that average transaction size fell by at least 80 percent.
According to recent reports, partly. Three things happened after 2018.
In a recent development, STARKs reached production. Eli Ben-Sasson, the paper's first author, co-founded StarkWare with Alessandro Chiesa and Uri Kolodny. The organization used STARKs first in StarkEx, a scaling service for individual clients, and then in Starknet, which launched its Alpha on Ethereum mainnet in November 2021. Starknet's documentation describes it as a permissionless validity rollup that settles on Ethereum using STARK proofs.
It has been revealed that zero-knowledge moved from privacy to scaling. MIT framed the technology as a privacy tool. Its largest use on Ethereum became zk-rollups, which process transactions off-chain and post one proof that they were all valid. Ethereum.org lists Starknet alongside ZKsync, Scroll, Linea, Polygon zkEVM and others. Ethereum.org also puts the cost of verifying a single SNARK on Ethereum at about 500,000 gas and says STARKs cost more, which keeps small proofs attractive.
SNARKs fixed their own weakness. PLONK, published in August 2019 by Ariel Gabizon, Zachary Williamson and Oana Ciobotaru, is a universal SNARK: one setup can serve many different programs instead of one ceremony per program. Halo went further. Zcash says Halo removes the trusted setup entirely, and its Orchard shielded protocol, activated with the NU5 upgrade in May 2022, runs on the Halo 2 proving system.
That last point matters for the old "SNARKs need a trusted setup" comparison. It was accurate in 2018. It no longer holds for every SNARK.
In a recent development, the fair verdict: STARKs did not push SNARKs aside, but they did prove that transparency and hash-based security work at scale. The choice today is less about which family wins and more about what you need: the smallest proof, no setup, or protection against quantum attacks.
Recent findings suggest that enjin holders vote through November 13 on moving Matrixchain to OpenGov on November 16. Here is how the nine tracks, the 1,000 ENJ deposit and conviction voting work, and what the Relaychain vote indicates.Enjin's fresh proposal deposit sounds steep at 1,000 ENJ. It's actually the cheap part.The Vote Closes Today and Matrixchain Governance Hangs on ItEnjin holders….
SlowMist says five forged receipts with valid signatures let an attacker drain about $304,500 from Ankr's Ethereum bridge using one old Optimism deposit. Here is how the check failed.Every one of the five receipts was signed by the bridge's own trusted signer. Every one pointed to a deposit that had already been paid out.Five Withdrawals Against One Settled….
The SEC's crypto broker-dealer proposal, RIN 3235-AN48, is under White House review with no public text. Here is what the four named rules require today and what review means.Four rules written decades before Bitcoin existed are now sitting at the White House for a crypto rewrite, and nobody outside the SEC has seen a word of the draft. What is known is the….
Industry observers note that base's Denim upgrade replaced Flashblocks with five canonical blocks per second. What that means for sequencers, RPC providers and smart contract developers is where the real story sits.For about sixteen months, Base's fastest blocks were only promises. Since November 5, they are the ledger itself.Base Swaps Previews for Real BlocksDenim went live on Base ma….
One email a week with our fresh explainers and the news behind them.
More explainers and analysis from NullTX.
