The Work
Laying the Keystone
The initiative researches and engineers L2 zero-knowledge execution environments for native general product applications. Core expertise runs beneath every layer of proof-system design — arithmetization, sumcheck and GKR, lookup and memory arguments, commitment schemes and FRI, recursion and aggregation — and what ships from it are maintained zkVMs and tooling. The expert lineage starts at the dawn of the ZKP era, when constraint systems and protocols were written and audited by hand, boutique style, before there were frameworks to hide any mistake. Since spring 2026 a dedicated prover-side program has driven arithmetization strategies past the standard playbook, with substantial results. Work at this depth moves a proof system's intrinsic cost model.
Working Principles
Four commitments. Every one of them is checkable from the outside, right now.
Publish the work, not summaries
Findings ship as full derivations, cited sources, and browser instruments that run every step live. Nothing is held back for a sales call. Research kept private compounds only for the party holding it.
Verify against source, not reputation
Every claim about another system is checked in that system's own code before it is published. Papers state intent; repositories record what was decided. A claim that cannot be pinned to a line is marked unverified or cut.
Correctness before performance
Numbers in published artifacts are computed, never illustrated: every Miller iteration, every permutation round, every register write. A demo that is fast and subtly wrong teaches the wrong thing convincingly.
Build in the open, alongside peers
Other zkVM teams are peers, not opposition. Prior art is named, and a result that holds up elsewhere is adopted and credited rather than rebuilt on someone else's time under a new name.
Work Together
Get right what matters
Most of what a zk system will ever cost is fixed in a handful of early decisions — field, trace shape, lookup strategy, recursion layout — and those are the hardest things in the stack to revisit once a team is nine months into building on them. The initiative takes a small number of independent collaborations on exactly that surface: proof-system and zkVM design, arithmetization and constraint systems, ISA definition and guest-toolchain targeting, lookup and memory arguments, recursion layout, and a straight assessment of an existing prover against the workload actually intended for it. Each one ends in a written technical output with the reasoning attached, so whoever inherits it can argue with it instead of taking it on faith.
Making blockchain-native development ordinary is a larger job than any one group finishes, and comparing notes is not billed for — if the work is application-specific execution, single-channel proving infrastructure, or the toolchain between them, that correspondence is welcome on its own terms.
Open with the hardest question you have: admin@gweb3networks.com