{"id":"0b2252b3-3431-4e74-ab38-d5ad06e066e6","arxiv_id":"2606.04819","paper_version":2,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Pearl's 24 EH/s cuPOW network produces zero useful AI computation because the dominant mining software contains no inference code and the verification protocol accepts random matrices by design.","lead":"This paper empirically measures Pearl's Proof-of-Useful-Work blockchain protocol and finds it performs no actual AI inference despite marketing claims. A smart generalist might read it to understand gaps between advertised useful computation and real energy use in blockchain systems.","discovery_kind":"new_application","skeptic_critique":{"model":"grok-4.3","headline":"Dominant mining software may not represent majority of 24 EH/s hash rate","rationale":"The reader's weakest assumption matches the load-bearing point exactly. The full text supplies more detail on the 8,012-worker sample and verification experiments, but does not resolve the coverage gap for the entire network hash rate. This keeps the verdict at UNVERDICTED.","tokens_in":1736,"tokens_out":266,"duration_ms":30391,"concrete_test":"Obtain pool-level hash-rate statistics or expand worker sampling until the identified dominant software's share of total EH/s can be bounded; if it is below 80% of network hash rate, inspect the alternative software for inference code.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The claim that the network produces zero useful AI computation depends on the dominant mining software (lacking inference code) being used by the large majority of participants. The network composition analysis examined only 8,012 workers. No data is given on the hash-rate fraction these workers represent relative to the full 24 EH/s. Because the verification protocol accepts random matrices by design, any unexamined software used by the remaining hash rate could perform actual inference while still submitting valid shares, leaving the zero-useful-computation conclusion unsupported.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.3","summary":"The paper claims that Pearl's 24 EH/s PoUW network produces zero useful AI computation. This is based on empirical measurements across five dimensions: network composition of 8,012 workers with inference-capable hardware but dominant software lacking inference code; verification protocol accepting random matrices confirmed by 44 shares; statistical checks defeated by adversarial sampling; marginal ROI for mining; and portable commodity computation. It also reports a 38% rise in GPU rental prices and utilization surge from 57% to 94% after mining software release, displacing research workloads.","tokens_in":1849,"tokens_out":401,"duration_ms":43051,"significance":"If the central claim holds, this manuscript provides valuable empirical evidence quantifying the usefulness gap in deployed Proof-of-Useful-Work systems. It offers direct measurements of network behavior, protocol verifiability, and economic impacts on GPU markets, grounding theoretical discussions from Leinweber et al. with concrete data from a high-profile blockchain. The multi-platform testing and economic analysis are particular strengths.","major_comments":[{"comment":"The assertion that the network produces zero useful AI computation hinges on the dominant mining software being representative of the majority of the 24 EH/s hash rate. The analysis is limited to 8,012 workers, with no reported data on the hash-rate fraction these workers constitute. Given that the verification protocol accepts random matrices by design, this leaves open the possibility that other software used by unexamined hash rate performs actual inference while submitting valid shares, undermining the zero-useful-computation conclusion.","section":"network composition analysis"}],"minor_comments":[{"comment":"Specific numerical claims (e.g., 38% price increase, 112 MW consumption) would benefit from explicit cross-references to the sections or tables detailing the underlying data collection and calculations.","section":"Abstract"}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":{"model":"grok-4.3","summary":"We thank the referee for their careful review and for highlighting an important methodological point regarding our network composition analysis. We address this comment directly below.","responses":[{"response":"Our analysis sampled 8,012 active workers from the Pearl network; all possessed inference-capable hardware yet executed mining software containing no inference code. We acknowledge that we did not collect per-worker hash-rate data and therefore cannot report the exact fraction of the 24 EH/s represented by this sample. However, these workers were drawn from the publicly observable set of participants, and the identified software was the only publicly released client at the time. More critically, our 44 pool-accepted shares using random matrices confirm that the verification protocol accepts non-inference inputs by design. Consequently, even if unexamined hash rate employs alternative software that optionally performs inference, the protocol neither requires nor verifies such computation. The zero-useful-computation conclusion therefore rests on the absence of enforced usefulness rather than solely on software prevalence. We will add an explicit discussion of sampling limitations and this protocol-level point in the revised manuscript.","revision_made":"partial","referee_comment":"The assertion that the network produces zero useful AI computation hinges on the dominant mining software being representative of the majority of the 24 EH/s hash rate. The analysis is limited to 8,012 workers, with no reported data on the hash-rate fraction these workers constitute. Given that the verification protocol accepts random matrices by design, this leaves open the possibility that other software used by unexamined hash rate performs actual inference while submitting valid shares, undermining the zero-useful-computation conclusion."}],"tokens_in":1413,"tokens_out":374,"duration_ms":26457,"standing_objections":["Precise hash-rate fraction represented by the 8,012 sampled workers (data not collected during measurements)"]},"desk_editor":{"model":"grok-4.3","letter":"The paper's core finding is that Pearl's cuPOW produces zero useful AI computation: the dominant mining software has no inference code, the verification layer accepts random matrices, and the work reduces to portable integer arithmetic with marginal economics at current prices.\n\nIt does the useful job of delivering the first direct measurements from a live PoUW deployment. The five dimensions—worker hardware survey, share acceptance tests across device types, distribution checks, ROI ranges, and lack of vendor lock-in—give concrete numbers where prior work stayed theoretical. The 38% GPU rental price rise and utilization jump after release are also worth noting as side effects.\n\nThe soft spot is the network composition claim. The analysis covers 8012 workers, all inference-capable, yet supplies no hash-rate fraction relative to the full 24 EH/s. If the unsampled portion runs different software that actually performs inference while still passing verification, the zero-useful-computation result does not extend to the whole network. That gap matches the stress-test concern and is not minor for the headline conclusion.\n\nThe verification and economics sections stand on firmer ground because they rest on direct tests and public data. This work is for researchers tracking PoUW designs and energy accounting in blockchains. It deserves peer review: the empirical approach is new and the measurements are reproducible enough to warrant referee scrutiny, even with the sampling limitation.","headline":"Pearl's PoUW network shows no useful AI work in the sampled software, but the 8012-worker analysis leaves open whether it covers most of the 24 EH/s hash rate.","tokens_in":2286,"tokens_out":362,"would_cite":false,"duration_ms":15786,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.3","headline":"Pearl's cuPOW protocol produces zero useful AI computation on its 24 EH/s network.","keywords":["Proof-of-Useful-Work","Pearl blockchain","AI inference","empirical measurement","blockchain verification","GPU mining","mining economics","usefulness gap"],"falsifier":"Discovery of inference code or actual trained-model outputs inside shares submitted by the dominant mining software would falsify the zero-useful-work finding.","tokens_in":2642,"feed_emoji":"🔬","tokens_out":658,"duration_ms":24558,"temperature":0.7,"pith_summary":"The paper measures a deployed Proof-of-Useful-Work blockchain that claims to run AI inference while securing the chain. It finds the entire network of roughly 320,000 GPU-equivalents performs no inference at all. The dominant mining software contains no inference code, yet the verification step accepts random matrices as valid shares across every hardware type tested. GPU rental prices rose 38 percent and utilization climbed from 57 to 94 percent after the software appeared, crowding out other workloads. The actual work reduces to ordinary integer arithmetic that runs on any platform and yields marginal or negative returns for most miners at current token prices.","feed_headline":"Pearl's PoUW network produces zero useful AI computation","feed_subtitle":"24 EH/s system shows no inference code and accepts random matrices as valid shares","key_machinery":"The verification protocol that accepts random matrices as valid proof-of-work shares.","core_discovery":"Pearl's cuPOW protocol, presented as simultaneously securing the network and executing useful AI inference, produces zero useful AI computation. Network analysis of 8,012 workers shows inference-capable hardware throughout, but the leading mining client includes no inference routines. The verification protocol accepts random matrices by design, a fact confirmed by 44 accepted shares from an open-source miner on NVIDIA, AMD, CPU, and Apple Silicon devices. Statistical distribution tests are defeated by simple adversarial sampling, and the arithmetic performed is portable commodity integer work with no hardware specificity.","pith_inferences":["Verification that cannot distinguish random inputs from real inference outputs leaves any PoUW system open to the same gap.","Portable integer work removes any incentive for miners to acquire specialized AI hardware.","Crowding of rental markets suggests PoUW mining can displace research workloads even when token economics are marginal.","Protocols relying on statistical checks alone remain vulnerable to adversarial sampling that passes verification without performing the claimed task."],"forward_implications":["GPU rental prices increased 38 percent and utilization reached 94 percent after the mining software release.","Mining returns range from negative 1 percent to positive 67 percent depending on GPU tier at current token prices.","The performed computation consists of commodity integer arithmetic runnable on any hardware platform.","The network consumes an estimated 112 MW while delivering no usable AI results."],"fun_headline_variants":["Pearl PoUW yields zero AI computation on large network","No inference code detected in Pearl's PoUW miners","Pearl verification accepts random matrices by design","Pearl PoUW arithmetic is portable commodity integer work","Pearl's 24 EH/s PoUW displaces research workloads"],"cache_read_input_tokens":2112,"weakest_assumption_plain":"The dominant mining software examined represents the majority of network participants, and acceptance of random matrices by the verification protocol demonstrates absence of actual inference computation.","fun_headline_variants_meta":{"raw":{"variants":["Pearl PoUW yields zero AI computation on large network","No inference code detected in Pearl's PoUW miners","Pearl verification accepts random matrices by design","Pearl PoUW arithmetic is portable commodity integer work","Pearl's 24 EH/s PoUW displaces research workloads"]},"model":"grok-4.3","cost_usd":0.004012,"raw_usage":{"total_tokens":2092,"prompt_tokens":759,"num_sources_used":0,"completion_tokens":72,"cost_in_usd_ticks":40124500,"prompt_tokens_details":{"text_tokens":759,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":1261,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":759,"tokens_out":72,"duration_ms":12215,"temperature":1.0,"reasoning_tokens":1261,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-06-28T05:51:53.021285+00:00","model_set":{"reader":"grok-4.3"},"falsifier":"Discovery of inference code or actual trained-model outputs inside shares submitted by the dominant mining software would falsify the zero-useful-work finding.","supporting_citations":[],"review_version":1}