ZeroThesis creator KrishnaMadala claims the released platform independently reruns agents’ experiment submissions in sandboxes and records verified results in a public signed ledger while retaining failed attempts, enabling multiple research agents to build on shared, attributable experiment history.
state: expiredheat: lowuncertainty: highknownscott: lowagent-harnesses automated-research provenanceKrishnaMadalaZeroThesis
What is this?
The supplied case identifies ZeroThesis as KrishnaMadala’s platform, presented in a “Show HN” announcement as multiplayer automated research with agents. The creator reportedly claims that it independently reruns submitted experiments in sandboxes, records verified results in a public signed ledger, and retains failed attempts so agents can build on shared, attributable research history. None of the supplied web-result snippets directly covers ZeroThesis, so they do not corroborate its release or implementation; the search answer’s additional claim about persistent stateful agents is also unsupported by relevant snippets.
Why it matters to Scott
The claimed design repeats positions Scott already holds in Verification Loops, Agent Receipts, and Shared Blackboard: externally checked runs with reconstructable records that multiple agents can reuse. The supplied material establishes neither a consequential new adopter nor a verified implementation that would change his trace-backed agent comparison work; related radar cases track adjacent approaches, but none supplied tracks ZeroThesis itself.
ip:concept.verification-loopsip:concept.agent-receiptsip:concept.shared-blackboarddev:concept.trace-backed-agent-comparisonradar:prime-intellect-autonomous-research-evalsradar:traceseal-signed-agent-receiptsradar:sdi-verifiable-agent-memory
queries asked of Scott's wikis
- agent harnesses independent execution verification
- automated research reproducibility experiment evaluation
- shared agent memory attributable experiment history
- failed experiments negative results knowledge retention
- signed provenance ledgers trust agent outputs
Measured heat
no measured readings yet — the hourly heat pass fills this in
How the heat travelled
no chain yet — the hourly chain pass fills this in
Evidence (2) — ⭐ canonical anchor
Interpretation history
2026-09-08T19:43:17Z
The follow-up window produced no new implementation evidence, independent use, or demonstrated reuse of experiment history; the echo remains a repetition of the creator’s claims. Retire this from active tracking as an unvalidated launch, not a disproved design; inspectable rerun receipts or substantive external adoption would justify reopening.
2026-09-06T19:28:17Z
No substantive new evidence changes the launch assessment: the echo repeats the creator’s claims rather than independently validating sandbox reruns or ledger provenance. ZeroThesis remains a concrete but unverified example of shared experiment history, not yet evidence that should change Scott’s engineering decisions.
2026-09-06T19:27:24Z
grounded: known/low — The claimed design repeats positions Scott already holds in Verification Loops, Agent Receipts, and Shared Blackboard: externally checked runs with reconstructa
2026-09-06T19:23:29Z
case created — A first-party launch describes a concrete shared research-agent workflow with rerun verification and provenance, distinct from the existing signed-notebook and research-harness cases.
Decision trace
- 09-09 05:43expireThe follow-up window produced no new implementation evidence, independent use, or demonstrated reuse of experiment history; the echo remains a repetition of the creator’s claims. Retire this from acti
- 09-09 05:43alert_silentThe staleness trigger adds no consequential delta, and no pending confirmation or time-sensitive opportunity is identified. The previously assessed launch does not warrant interrupting Scott.
- 09-09 05:43alert_routeThe staleness trigger adds no consequential delta, and no pending confirmation or time-sensitive opportunity is identified. The previously assessed launch does not warrant interrupting Scott.
- 09-07 05:28repriceNo substantive new evidence changes the launch assessment: the echo repeats the creator’s claims rather than independently validating sandbox reruns or ledger provenance. ZeroThesis remains a concrete
- 09-07 05:28alert_silentThere is no new release, implementation evidence, or adopter in this delta. The previously assessed launch can still wait for a normal briefing; independent verification or meaningful agent reuse woul
- 09-07 05:28alert_routeThere is no new release, implementation evidence, or adopter in this delta. The previously assessed launch can still wait for a normal briefing; independent verification or meaningful agent reuse woul
- 09-07 05:27alert_silentThe creator’s announcement establishes a new free platform to try; independent reruns, signed provenance and failed-attempt retention remain implementation claims. This is a concrete example relevant
- 09-07 05:27surface_candidateThe creator’s announcement establishes a new free platform to try; independent reruns, signed provenance and failed-attempt retention remain implementation claims. This is a concrete example relevant
- 09-07 05:27alert_routeThe creator’s announcement establishes a new free platform to try; independent reruns, signed provenance and failed-attempt retention remain implementation claims. This is a concrete example relevant
- 09-07 05:27groundThe claimed design repeats positions Scott already holds in Verification Loops, Agent Receipts, and Shared Blackboard: externally checked runs with reconstructable records that multiple agents can reu
- 09-07 05:23createA first-party launch describes a concrete shared research-agent workflow with rerun verification and provenance, distinct from the existing signed-notebook and research-harness cases.