NodusLab
27 experiments

Every measurement, including the ones that failed

Each experiment records its hypothesis before the run. Failed experiments are written up rather than dropped — they are the cheap ones — and three of the results here broke a conclusion the programme had already written down.

Complete14Partly complete2Pre-registered3Queued8experiments/ on GitHub ↗
27 experiments

The chain, in discovery order

Genuinely a sequence: each result changed what the next experiment had to be, and several of them broke a conclusion that had already been written down. The full index follows underneath, grouped by status.

E001zkVM floor

What does a general-purpose zero-knowledge proof actually cost?

Proving y = 3x + 5 on SP1 took 11.19 s against a native 0.69 ns. Proving ten thousand of them cost the same — the charge is for the existence of a proof, not the work inside it. The asymptote settles near 7 × 10⁴×, and that is a floor: register-resident u64 arithmetic is the friendliest workload a RISC-V zkVM will ever see. Not a tuning problem.

E003Freivalds

Can a large matrix product be checked more cheaply than redoing it?

Yes, by a 1977 result. Freivalds verified an 8192² product in 2.3% of recomputing it, soundness ≤ 2⁻⁴⁰, zero cost to the provider. Nine orders of magnitude better than E001 on the same claim — and the difference is not implementation quality. A zkVM pays for every load and branch in an execution trace; Freivalds exploits the algebraic structure of the statement.

E009-pre

How far do two honest backends diverge?

Far enough to ruin it. An honest CPU and an honest Metal GPU differed by 7.70 × 10⁻⁴; a genuine precision cheat by 2.84 × 10⁻³. The cause is the device — identical library, identical dtype, one line of difference.

falsified E003's security analysis
E013work vs output

If the answer is right, did they do the work?

Seven cheating strategies against the strongest checker we had. Every strategy that changes the answer is rejected — but two low-work strategies pass, and should: a cached correct answer, and Strassen's algorithm at 67% of the arithmetic. No checker can object, because the answer is right. A shortcut is only an attack if you are paying for effort rather than output. The cache case is a pricing problem, closed by deduplicating on the input commitment.

E014E015 · E017

What does an exact integer contract cost in supply?

Nothing. Every honest backend tested returns the bit-exact product, while cheats separate without bound. A whole transformer layer — softmax, GELU, layernorm, requantisation — came out bit-identical on CPU and GPU. Determinism is free at 8 bits, still free at 6, and costs real accuracy at 4.

reversed the expected supply/security trade-off
E016bandwidth

What does verification cost in bandwidth?

Enough to dominate everything: 49.1 MB of intermediates against a 0.02 MB answer, and a break-even compute price of $164/hour against a real GPU-hour of about $1. We also rediscovered a 2017 result by construction here, because we had not searched the literature first. It is recorded in the experiment rather than quietly fixed.

killed verifying every layer
E018generation

Does the contract survive a transformer that is actually generating text?

Only once we found the real bound. MLX's Metal matmul is exact on integer operands to 2¹¹ and inexact from 2¹², regardless of accumulator magnitude — so NC-3's 2²⁴ accumulator bound was 13 bits looser than the binding one. Attention probabilities carried at 2¹⁵ were the first operand to exceed it, and CPU and GPU generations diverged at token 102. With NC-3b in place: 10/10 runs bit-identical over 300 tokens each.

broke NC-3 · added NC-3b
E019E020 · E026

Can an interactive proof remove the bandwidth cost?

Sumcheck replaces a 16.8 MB intermediate with a 792-byte transcript — 21,183× — and the prover pays only 10.4% at n = 4096, of which the protocol itself is 0.5%. On a transformer layer it saves nothing, because every matmul output feeds a lookup-table non-linearity the verifier must recompute anyway. The tension is structural: the property that buys determinism blocks the proof. Making the model sumcheck-friendly means replacing softmax, which costs +6.8% at 2 layers and +19.8% at 8 — a divergence, not a tax a bigger model amortises.

the polynomial route is closed
E021E022 · E023

Does the contract hold on hardware that isn't a Mac?

Yes — Intel x86 on two operating systems, and an NVIDIA L4 with TF32 left enabled. The NVIDIA operand bound of 2¹¹ was predicted from the Apple measurement and recorded before the hardware was available, then confirmed exactly. Qwen2.5-0.5B then produced byte-identical logits on all five machines.

pre-registered prediction confirmed
E024E027

What do you do about the bandwidth, then?

Audit one random layer, backed by a stake — always-on, and enough to make cheating negative-expected-value at 4.2% of the traffic. Where a conclusive ruling is wanted, bisection to the first divergent digest names the guilty layer with certainty at the same bandwidth, in 5 rounds and 1.9 KB. The two are complementary rather than competing.

Complete

Run, analysed, and carrying a result the rest of the programme depends on.

E001Simple arithmetic in a general zkVMWhat does it cost to prove y = 3x + 5, and how does that cost change as the amount of proved work grows?CP-001CompleteE003Matrix multiplicationCP-003CompleteE013Work versus outputwhat can correctness verification not see, and does that gap actually cost the network money?CP-011CompleteE014The numeric contractwhat does a numeric contract cost in supply, and what does it buy in security?CP-012CompleteE015Non-linearities under an exact contractcan softmax, GELU and layernorm be pinned exactly — and what does verifying a whole layer cost?CP-013CompleteE016The bandwidth billwhat does verification cost in bandwidth, and does that kill it?CP-014CompleteE017What the numeric contract costs in accuracywe have told the network to switch to exact integer arithmetic. What does that cost in answer quality?CP-015CompleteE018The contract on a transformer, doing generationdoes NC-0.2 survive a transformer, where softmax is on the critical path and a single flipped token compounds?CP-016CompleteE019An interactive proof for matmul chainscan a sumcheck protocol remove the bandwidth cost that E016 found dominates everything?CP-017CompleteE020What a provable transformer costs in qualityE019 showed our lookup tables block the sumcheck. If we change the model to match the protocol instead, what does that cost?CP-018CompleteE021Cross-vendor conformancedoes the numeric contract hold on hardware that isn't a Mac?CP-019CompleteE022NVIDIA TF32, and a prediction confirmedCP-020CompleteE023the contract on Qwen2.5-0.5BCompleteE024The layer-audit protocolCP-022Complete
Partly complete

A pre-registered sub-question answered; the main checkpoint still open.

Pre-registered

Hypotheses written down before the measurement, and not edited afterwards.

Queued

Not yet run. No results exist for these, and the pages say so.