---
schema_version: "1.0.0"
edition_id: "mp-2026-11-12-morning-0125"
published_at: "2026-11-12T09:00:00.000-05:00"
modified_at: "2026-11-12T09:00:00.000-05:00"
canonical_url: "https://themachinepress.com/edition/2026-11-12"
story_count: 27
lead_story_id: "mp-2026-11-12-001"
---

# The Machine Press — Morning edition

Edition ID: `mp-2026-11-12-morning-0125`  
Published: 2026-11-12T09:00:00.000-05:00  
Canonical edition: https://themachinepress.com/edition/2026-11-12

A generative policy trained with a 70-qubit processor proposes far shorter circuits for molecular energy estimates, then reuses them on unseen organotin systems.

## 1. The Quantum Circuit Learns From the Hardware {#mp-2026-11-12-001}

- Story ID: `mp-2026-11-12-001`
- Type: `lead`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-11-12-001/the-quantum-circuit-learns-from-the-hardware

**Dek:** A generative policy trained with a 70-qubit processor proposes far shorter circuits for molecular energy estimates, then reuses them on unseen organotin systems.

Researchers trained transformer policies to generate quantum state-preparation circuits, first in simulation and then with an IBM Heron processor inside the loop at 46 and 70 qubits. The four main hardware-training runs used 120 million shots, while reward terms discouraged collapse to circuits that sampled only a simple reference state. Applied without retraining to organotin hydroxide models, 27 of 28 runs at 46 to 70 qubits landed within 15 millihartree of a density-matrix renormalization reference after one hardware-sampling round. The selected circuits used 34 to 72 times fewer two-qubit gates than a compared ansatz at 56 to 70 qubits; these are reported molecular-model results, not a general quantum advantage or a chemical discovery.

### Why it matters {#why-it-matters-mp-2026-11-12-001}

A generative policy trained with a 70-qubit processor proposes far shorter circuits for molecular energy estimates, then reuses them on unseen organotin systems.

### Limits and context {#limitations-mp-2026-11-12-001}

- The four main hardware-training runs used 120 million shots, while reward terms discouraged collapse to circuits that sampled only a simple reference state.
- The selected circuits used 34 to 72 times fewer two-qubit gates than a compared ansatz at 56 to 70 qubits; these are reported molecular-model results, not a general quantum advantage or a chemical discovery.

### Claims and sources {#claims-mp-2026-11-12-001}

- A generative policy trained with a 70-qubit processor proposes far shorter circuits for molecular energy estimates, then reuses them on unseen organotin systems. [source-2026-11-12-001] — Qualification: The four main hardware-training runs used 120 million shots, while reward terms discouraged collapse to circuits that sampled only a simple reference state.

## 2. The Robot Reads the Protocol on the Bench {#mp-2026-11-12-002}

- Story ID: `mp-2026-11-12-002`
- Type: `secondary`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-11-12-002/the-robot-reads-the-protocol-on-the-bench

**Dek:** Curie's Jarvis follows a photographed instruction sheet, reads ordinary container labels and looks again before every chemistry step.

Curie's Jarvis is designed for a laboratory bench arranged for people rather than a fixed automation cell. A robot arm reads steps from a photograph of a printed protocol, identifies containers through printed or handwritten labels viewed from four cameras, and relocates its target before each action instead of assuming that objects remain in fixed positions. On a Franka Panda arm, the system completed all five runs of one children's chemistry-kit protocol and four of five runs of a second without human intervention; across 160 label-location attempts it identified every container correctly. The demonstration is small and controlled, but it shows how perception and procedural reasoning can reduce the setup burden of bench automation.

### Why it matters {#why-it-matters-mp-2026-11-12-002}

Curie's Jarvis follows a photographed instruction sheet, reads ordinary container labels and looks again before every chemistry step.

### Limits and context {#limitations-mp-2026-11-12-002}

- Curie's Jarvis is designed for a laboratory bench arranged for people rather than a fixed automation cell.
- A robot arm reads steps from a photograph of a printed protocol, identifies containers through printed or handwritten labels viewed from four cameras, and relocates its target before each action instead of assuming that objects remain in fixed positions.

### Claims and sources {#claims-mp-2026-11-12-002}

- Curie's Jarvis follows a photographed instruction sheet, reads ordinary container labels and looks again before every chemistry step. [source-2026-11-12-002] — Qualification: Curie's Jarvis is designed for a laboratory bench arranged for people rather than a fixed automation cell.

## 3. The Coding Agent Plans the Graph Before the Patch {#mp-2026-11-12-003}

- Story ID: `mp-2026-11-12-003`
- Type: `dispatch`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-11-12-003/the-coding-agent-plans-the-graph-before-the-patch

**Dek:** FlowPlan expresses a repair as control-flow or data-flow changes, then checks whether the edited program matches that plan.

FlowPlan asks a coding agent to localize faults and plan repairs on per-function control-flow or data-flow graphs before changing source code. After the edit, it reconstructs the abstraction, compares the observed graph delta with the plan and then performs a complete verification. Across six benchmark-model settings, resolution did not change significantly while token use fell in every setting and significantly in five. With the smaller evaluated model, usage fell to 32% to 74% of baseline depending on the benchmark, mainly by reducing repeated post-edit reading and testing.

### Why it matters {#why-it-matters-mp-2026-11-12-003}

FlowPlan expresses a repair as control-flow or data-flow changes, then checks whether the edited program matches that plan.

### Limits and context {#limitations-mp-2026-11-12-003}

- Across six benchmark-model settings, resolution did not change significantly while token use fell in every setting and significantly in five.

### Claims and sources {#claims-mp-2026-11-12-003}

- FlowPlan expresses a repair as control-flow or data-flow changes, then checks whether the edited program matches that plan. [source-2026-11-12-003] — Qualification: Across six benchmark-model settings, resolution did not change significantly while token use fell in every setting and significantly in five.

## 4. A Hundred Engineering Tickets Resist a Clean Sweep {#mp-2026-11-12-004}

- Story ID: `mp-2026-11-12-004`
- Type: `dispatch`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-11-12-004/a-hundred-engineering-tickets-resist-a-clean-sweep

**Dek:** ExpertCrowd-SWE uses sealed solutions, partial-credit graders and must-pass blockers across 87 unfamiliar open-source projects.

ExpertCrowd-SWE contains 100 multi-file engineering tasks written by practicing engineers against pinned commits in 87 open-source projects and 11 languages. Each task has a sealed reference patch and a weighted grader with executable checks, partial credit and blockers for changes a maintainer would refuse. Across 3,495 scored runs, the leading evaluated agent averaged 67.3% and the next 63.7%, without a statistically significant gap. No task was solved fully by every model, and 80 tasks lacked a model that achieved full credit on all five runs.

### Why it matters {#why-it-matters-mp-2026-11-12-004}

ExpertCrowd-SWE uses sealed solutions, partial-credit graders and must-pass blockers across 87 unfamiliar open-source projects.

### Limits and context {#limitations-mp-2026-11-12-004}

- No additional limitation was separately recorded.

### Claims and sources {#claims-mp-2026-11-12-004}

- ExpertCrowd-SWE uses sealed solutions, partial-credit graders and must-pass blockers across 87 unfamiliar open-source projects. [source-2026-11-12-004]

## 5. The Agent's Failure Usually Starts Near the Beginning {#mp-2026-11-12-005}

- Story ID: `mp-2026-11-12-005`
- Type: `dispatch`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-11-12-005/the-agent-s-failure-usually-starts-near-the-beginning

**Dek:** A trajectory audit finds contradictions were already present before failure in most reviewed runs, but the systems rarely acted on them.

Researchers read 95 failed multi-agent trajectories in full and annotated where each fault originated, how it propagated, what evidence contradicted it and which control could have reached it. The median fault entered at 18% of the run, usually as a wrong task-state decision or fabrication. In 72% of failures the trajectory already contained evidence against the error; in 88% of those cases no agent acted on it. The audit also found that 45% of one corpus's hand-crafted failures ended in a provider content-filter crash followed by a harness-generated answer, exposing a failure class that responsible-agent labels miss.

### Why it matters {#why-it-matters-mp-2026-11-12-005}

A trajectory audit finds contradictions were already present before failure in most reviewed runs, but the systems rarely acted on them.

### Limits and context {#limitations-mp-2026-11-12-005}

- No additional limitation was separately recorded.

### Claims and sources {#claims-mp-2026-11-12-005}

- A trajectory audit finds contradictions were already present before failure in most reviewed runs, but the systems rarely acted on them. [source-2026-11-12-005]

## 6. The Radar Keeps the Cube {#mp-2026-11-12-006}

- Story ID: `mp-2026-11-12-006`
- Type: `dispatch`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-11-12-006/the-radar-keeps-the-cube

**Dek:** RaMOD detects moving objects directly in range-azimuth-Doppler data while separating ego motion from real scene motion.

RaMOD works from raw millimeter-wave radar cubes rather than first reducing them to sparse points or two-dimensional projections. It estimates the vehicle's motion, predicts where stationary structures should appear in Doppler space and suppresses those returns before detecting and estimating moving objects. On the authors' BRIDGE-Liv dataset, the system reached 80.8% cube mean average precision at the stated overlap threshold, 15.7 points above the strongest compared raw-cube baseline. Its reported velocity error was 1.128 meters per second with 99.87% Doppler-wrap classification accuracy.

### Why it matters {#why-it-matters-mp-2026-11-12-006}

RaMOD detects moving objects directly in range-azimuth-Doppler data while separating ego motion from real scene motion.

### Limits and context {#limitations-mp-2026-11-12-006}

- No additional limitation was separately recorded.

### Claims and sources {#claims-mp-2026-11-12-006}

- RaMOD detects moving objects directly in range-azimuth-Doppler data while separating ego motion from real scene motion. [source-2026-11-12-006]

## 7. An Agent-Written Functional Survives the Physics Check {#mp-2026-11-12-007}

- Story ID: `mp-2026-11-12-007`
- Type: `dispatch`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-11-12-007/an-agent-written-functional-survives-the-physics-check

**Dek:** A compact density-functional candidate retained its advantage after self-consistent and held-out evaluation.

Researchers connected language-model agents to a GPU evaluator for symbolic search over exchange-correlation functionals, with persistent memory and separate self-consistent and external checks. The surviving 13-parameter candidate adds a smooth density window that reduces same-spin correlation in selected regions. After refitting, it reduced heavy-atom total-energy error by 29% against a no-window control on one test collection and lowered a held-out molecular-chemistry error by 11.7% relative to the stated baseline. The evidence supports one compact candidate under the reported validation pipeline, not general autonomous discovery.

### Why it matters {#why-it-matters-mp-2026-11-12-007}

A compact density-functional candidate retained its advantage after self-consistent and held-out evaluation.

### Limits and context {#limitations-mp-2026-11-12-007}

- The evidence supports one compact candidate under the reported validation pipeline, not general autonomous discovery.

### Claims and sources {#claims-mp-2026-11-12-007}

- A compact density-functional candidate retained its advantage after self-consistent and held-out evaluation. [source-2026-11-12-007] — Qualification: The evidence supports one compact candidate under the reported validation pipeline, not general autonomous discovery.

## 8. The Soft Robot Is Written, Simulated and Rewritten {#mp-2026-11-12-008}

- Story ID: `mp-2026-11-12-008`
- Type: `dispatch`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-11-12-008/the-soft-robot-is-written-simulated-and-rewritten

**Dek:** PneuGen's designer and critic agents search pneumatic geometries through validated programs and reusable physical response models.

PneuGen represents soft pneumatic robots as design programs built from a parametric component library. A designer agent proposes geometries, a critic interprets simulation diagnostics, and fitted response models supply physical priors before finite-element evaluation. The authors report that the system outperformed compared computational optimization baselines on grasping and shape-matching tasks, with ablations supporting contributions from both the critic and response model. They also fabricated and tested generated designs, turning the result into a physical co-design demonstration rather than a simulation-only claim.

### Why it matters {#why-it-matters-mp-2026-11-12-008}

PneuGen's designer and critic agents search pneumatic geometries through validated programs and reusable physical response models.

### Limits and context {#limitations-mp-2026-11-12-008}

- They also fabricated and tested generated designs, turning the result into a physical co-design demonstration rather than a simulation-only claim.

### Claims and sources {#claims-mp-2026-11-12-008}

- PneuGen's designer and critic agents search pneumatic geometries through validated programs and reusable physical response models. [source-2026-11-12-008] — Qualification: They also fabricated and tested generated designs, turning the result into a physical co-design demonstration rather than a simulation-only claim.

## 9. The Protein World Model Learns One Mutation at a Time {#mp-2026-11-12-009}

- Story ID: `mp-2026-11-12-009`
- Type: `dispatch`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-11-12-009/the-protein-world-model-learns-one-mutation-at-a-time

**Dek:** ProTEUS combines mutation-biased attention, a fast geometry-constrained predictor and experimental feedback for closed-loop evolution.

ProTEUS treats protein variants as states and individual mutations as actions whose functional effects can be predicted and revised with experimental feedback. Its MutationFormer combines sequence, evolutionary and structural information, while a geometry-constrained all-atom predictor is reported to run more than ten times faster than AlphaFold3 with enforced physical validity. In two closed-loop campaigns on aldehyde dehydrogenase and levansucrase, the authors report variants with more than tenfold higher activity than wild type. Those campaign results establish utility in the tested proteins, not a universal protein-engineering guarantee.

### Why it matters {#why-it-matters-mp-2026-11-12-009}

ProTEUS combines mutation-biased attention, a fast geometry-constrained predictor and experimental feedback for closed-loop evolution.

### Limits and context {#limitations-mp-2026-11-12-009}

- Those campaign results establish utility in the tested proteins, not a universal protein-engineering guarantee.

### Claims and sources {#claims-mp-2026-11-12-009}

- ProTEUS combines mutation-biased attention, a fast geometry-constrained predictor and experimental feedback for closed-loop evolution. [source-2026-11-12-009] — Qualification: Those campaign results establish utility in the tested proteins, not a universal protein-engineering guarantee.

## 10. Missing Permission Evidence Is Not the Same as No {#mp-2026-11-12-010}

- Story ID: `mp-2026-11-12-010`
- Type: `dispatch`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-11-12-010/missing-permission-evidence-is-not-the-same-as-no

**Dek:** Enterprise agents stopped less reliably when authorization records were absent than when those records explicitly denied access.

DEPUTYBENCH evaluates whether enterprise agents continue legitimate work while respecting organizational permissions, using 375 instances built from 79 base items. In controlled permission flips and evidence masking, restricted execution occurred on 10.3% of matched tasks when permission evidence was missing versus 3.5% when non-permission was explicit. Most masked-condition failures clustered in four privacy tasks involving cumulative counts or person-specific consent. The result warns that high safety with complete records does not establish safe behavior when authorization cannot be verified.

### Why it matters {#why-it-matters-mp-2026-11-12-010}

Enterprise agents stopped less reliably when authorization records were absent than when those records explicitly denied access.

### Limits and context {#limitations-mp-2026-11-12-010}

- The result warns that high safety with complete records does not establish safe behavior when authorization cannot be verified.

### Claims and sources {#claims-mp-2026-11-12-010}

- Enterprise agents stopped less reliably when authorization records were absent than when those records explicitly denied access. [source-2026-11-12-010] — Qualification: The result warns that high safety with complete records does not establish safe behavior when authorization cannot be verified.

## 11. A Promising Molecule Can Fail at the Required Dose {#mp-2026-11-12-011}

- Story ID: `mp-2026-11-12-011`
- Type: `dispatch`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-11-12-011/a-promising-molecule-can-fail-at-the-required-dose

**Dek:** Exposure-aware simulation reshuffles candidates that static potency and toxicity screens had placed near the top.

Researchers re-evaluated 60 molecules selected by three generative drug-design pipelines using physiologically based pharmacokinetic simulations. Seven failed to reach sufficient target exposure, and 28% of the remaining candidates showed predicted safety risks at the required dose. Among candidates with an achievable dose, 85% were outperformed by molecules that the static selection had overlooked. The result is computational and target-specific, but it shows why potency and toxicity predictions alone can rank the wrong candidates.

### Why it matters {#why-it-matters-mp-2026-11-12-011}

Exposure-aware simulation reshuffles candidates that static potency and toxicity screens had placed near the top.

### Limits and context {#limitations-mp-2026-11-12-011}

- No additional limitation was separately recorded.

### Claims and sources {#claims-mp-2026-11-12-011}

- Exposure-aware simulation reshuffles candidates that static potency and toxicity screens had placed near the top. [source-2026-11-12-011]

## 12. Starlink Downlinks Become a Mock Navigation Constellation {#mp-2026-11-12-012}

- Story ID: `mp-2026-11-12-012`
- Type: `dispatch`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-11-12-012/starlink-downlinks-become-a-mock-navigation-constellation

**Dek:** Beam-specific orbit and clock models turn live Ku-band arrivals into meter-scale positioning and nanosecond timing in a field test.

A field experiment extracted time-of-arrival measurements from live Starlink downlinks and paired them with high-fidelity, beam-specific orbit and clock models distributed as though the constellation operator supplied them. The mock fused low-Earth-orbit navigation system reported 95% error bounds of 2.33 meters horizontally, 3.37 meters vertically and 11.19 nanoseconds in timing. Actual errors in the campaign remained inside those bounds despite consumer-grade receiver constraints. The method depends on precision data not available in ordinary opportunistic reception, so it demonstrates an architecture rather than a current public service.

### Why it matters {#why-it-matters-mp-2026-11-12-012}

Beam-specific orbit and clock models turn live Ku-band arrivals into meter-scale positioning and nanosecond timing in a field test.

### Limits and context {#limitations-mp-2026-11-12-012}

- The method depends on precision data not available in ordinary opportunistic reception, so it demonstrates an architecture rather than a current public service.

### Claims and sources {#claims-mp-2026-11-12-012}

- Beam-specific orbit and clock models turn live Ku-band arrivals into meter-scale positioning and nanosecond timing in a field test. [source-2026-11-12-012] — Qualification: The method depends on precision data not available in ordinary opportunistic reception, so it demonstrates an architecture rather than a current public service.

## 13. The Quantum Decoder Reaches Microseconds {#mp-2026-11-12-013}

- Story ID: `mp-2026-11-12-013`
- Type: `dispatch`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-11-12-013/the-quantum-decoder-reaches-microseconds

**Dek:** PRISM uses erasure locations in photonic fault-tolerance protocols and runs a distance-eight logical block on an FPGA in 6.59 microseconds.

PRISM is a transformer decoder for photonic fusion-based quantum computing, where photon loss and probabilistic fusion produce known erasure locations. Across surface, color and quantum low-density parity-check codes up to distance 24, the authors report logical error rates matching integer-programming decoders while running faster, including near high erasure rates. An FPGA implementation decoded a distance-eight block in 6.59 microseconds with 1.56 microseconds amortized per logical block. The result demonstrates real-time inference on the tested protocols and hardware, not a complete fault-tolerant computer.

### Why it matters {#why-it-matters-mp-2026-11-12-013}

PRISM uses erasure locations in photonic fault-tolerance protocols and runs a distance-eight logical block on an FPGA in 6.59 microseconds.

### Limits and context {#limitations-mp-2026-11-12-013}

- The result demonstrates real-time inference on the tested protocols and hardware, not a complete fault-tolerant computer.

### Claims and sources {#claims-mp-2026-11-12-013}

- PRISM uses erasure locations in photonic fault-tolerance protocols and runs a distance-eight logical block on an FPGA in 6.59 microseconds. [source-2026-11-12-013] — Qualification: The result demonstrates real-time inference on the tested protocols and hardware, not a complete fault-tolerant computer.

## 14. Chain-of-Thought Monitoring Gets a Disclosure Checklist {#mp-2026-11-12-014}

- Story ID: `mp-2026-11-12-014`
- Type: `dispatch`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-11-12-014/chain-of-thought-monitoring-gets-a-disclosure-checklist

**Dek:** A proposal identifies five indicators for tracking whether model reasoning remains accessible, natural, faithful and useful to monitors.

The paper proposes five disclosure indicators covering the chain on which chain-of-thought monitoring depends: third-party access and architecture information, reasoning quality, and monitoring effectiveness. An audit of public reporting by three frontier developers found that none covered the full set, although each proposed indicator had appeared somewhere. Because existing evidence is largely developer-generated, the authors call for a deeper cross-developer audit by embedded evaluators or an independent verification organization. This is a governance proposal and disclosure audit, not evidence that chain-of-thought monitoring is sufficient by itself.

### Why it matters {#why-it-matters-mp-2026-11-12-014}

A proposal identifies five indicators for tracking whether model reasoning remains accessible, natural, faithful and useful to monitors.

### Limits and context {#limitations-mp-2026-11-12-014}

- This is a governance proposal and disclosure audit, not evidence that chain-of-thought monitoring is sufficient by itself.

### Claims and sources {#claims-mp-2026-11-12-014}

- A proposal identifies five indicators for tracking whether model reasoning remains accessible, natural, faithful and useful to monitors. [source-2026-11-12-014] — Qualification: This is a governance proposal and disclosure audit, not evidence that chain-of-thought monitoring is sufficient by itself.

## 15. A GPU's Special Functions Are Rebuilt Bit for Bit {#mp-2026-11-12-026}

- Story ID: `mp-2026-11-12-026`
- Type: `dispatch`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-11-12-026/a-gpu-s-special-functions-are-rebuilt-bit-for-bit

**Dek:** SFU-Sim reverse-engineers NVIDIA multifunction instructions and reproduces tested PTX paths across three GPU architectures without a lookup table.

SFU-Sim models the exact numerical behavior of NVIDIA multifunction-unit instructions using black-box measurements from real GPUs. The work first reconstructs a bare instruction, then adds the surrounding operations observed in compiled PTX-to-SASS paths. For tested FP16 and FP32 sine and cosine paths on Ampere, Ada Lovelace and Blackwell GPUs, exhaustive comparison across the specified input domains produced zero bit mismatches. The open-source model computes outputs directly and is intended as a reference component for instruction simulators and hardware verification.

### Why it matters {#why-it-matters-mp-2026-11-12-026}

SFU-Sim reverse-engineers NVIDIA multifunction instructions and reproduces tested PTX paths across three GPU architectures without a lookup table.

### Limits and context {#limitations-mp-2026-11-12-026}

- No additional limitation was separately recorded.

### Claims and sources {#claims-mp-2026-11-12-026}

- SFU-Sim reverse-engineers NVIDIA multifunction instructions and reproduces tested PTX paths across three GPU architectures without a lookup table. [source-2026-11-12-015]

## 16. The Robot's Map Learns That the World Is Water {#mp-2026-11-12-027}

- Story ID: `mp-2026-11-12-027`
- Type: `dispatch`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-11-12-027/the-robot-s-map-learns-that-the-world-is-water

**Dek:** A terrestrial open-vocabulary scene graph gains persistent maritime objects, local places and macro regions across repeated voyages.

Researchers tested an outdoor open-vocabulary 3D scene-graph method in maritime environments and found that it described many geometry-supported objects but fragmented single physical entities and imposed terrestrial spatial assumptions. Their adaptation consolidates accumulated metric and semantic evidence, uses geometry-grounded multiview reasoning to reconcile fragments, and builds maritime macro regions from map priors and onboard observations. The resulting representation reduced entity fragmentation and improved reviewed contextual-retrieval queries over tested baselines. It is a mapping adaptation for the reported data, not proof of general marine autonomy.

### Why it matters {#why-it-matters-mp-2026-11-12-027}

A terrestrial open-vocabulary scene graph gains persistent maritime objects, local places and macro regions across repeated voyages.

### Limits and context {#limitations-mp-2026-11-12-027}

- It is a mapping adaptation for the reported data, not proof of general marine autonomy.

### Claims and sources {#claims-mp-2026-11-12-027}

- A terrestrial open-vocabulary scene graph gains persistent maritime objects, local places and macro regions across repeated voyages. [source-2026-11-12-016] — Qualification: It is a mapping adaptation for the reported data, not proof of general marine autonomy.

## 17. The Keyboard Attack Leaves the Lab {#mp-2026-11-12-015}

- Story ID: `mp-2026-11-12-015`
- Type: `ticker`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-11-12-015/the-keyboard-attack-leaves-the-lab

**Dek:** Domain-invariant acoustics plus language-model correction recover typed text across harder real-world conditions, the authors report.

The study tests acoustic keystroke inference across unseen keyboards, users, noise and compressed channels rather than one quiet matched setup. Its reported result is that explicit domain-invariant acoustic modeling and language-model correction preserve attacks under those shifts, showing that commodity-microphone risk is not confined to a laboratory match.

### Why it matters {#why-it-matters-mp-2026-11-12-015}

Domain-invariant acoustics plus language-model correction recover typed text across harder real-world conditions, the authors report.

### Limits and context {#limitations-mp-2026-11-12-015}

- Its reported result is that explicit domain-invariant acoustic modeling and language-model correction preserve attacks under those shifts, showing that commodity-microphone risk is not confined to a laboratory match.

### Claims and sources {#claims-mp-2026-11-12-015}

- Domain-invariant acoustics plus language-model correction recover typed text across harder real-world conditions, the authors report. [source-2026-11-12-017] — Qualification: Its reported result is that explicit domain-invariant acoustic modeling and language-model correction preserve attacks under those shifts, showing that commodity-microphone risk is not confined to a laboratory match.

## 18. The Chip Optimizer Learns the Physical Consequence {#mp-2026-11-12-016}

- Story ID: `mp-2026-11-12-016`
- Type: `ticker`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-11-12-016/the-chip-optimizer-learns-the-physical-consequence

**Dek:** TAPCO-G combines language-model policy changes with graph priors for timing and power repairs.

TAPCO-G uses physical-design feedback to revise optimizer settings and executable repair policies, while graph-based priors estimate how cell replacements change endpoint slack and power. The authors report legal, logic-equivalent layouts and higher scores than compared systems on all eight stated ASAP7 benchmarks.

### Why it matters {#why-it-matters-mp-2026-11-12-016}

TAPCO-G combines language-model policy changes with graph priors for timing and power repairs.

### Limits and context {#limitations-mp-2026-11-12-016}

- No additional limitation was separately recorded.

### Claims and sources {#claims-mp-2026-11-12-016}

- TAPCO-G combines language-model policy changes with graph priors for timing and power repairs. [source-2026-11-12-018]

## 19. The Coding Harness Gets a Reference Monitor {#mp-2026-11-12-017}

- Story ID: `mp-2026-11-12-017`
- Type: `ticker`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-11-12-017/the-coding-harness-gets-a-reference-monitor

**Dek:** ActiveSecHarness composes capability tokens, taint tracking and deterministic operating-system guards.

The proposed security architecture separates model intelligence from authority and tracks both trust and sensitivity across tool use and files. On 252 constructed fixtures, its deterministic guard blocked every attack with no reported benign false positives, while a local model judge refused most benign work; the evidence is from the authors' fixture suite, not a production deployment.

### Why it matters {#why-it-matters-mp-2026-11-12-017}

ActiveSecHarness composes capability tokens, taint tracking and deterministic operating-system guards.

### Limits and context {#limitations-mp-2026-11-12-017}

- On 252 constructed fixtures, its deterministic guard blocked every attack with no reported benign false positives, while a local model judge refused most benign work; the evidence is from the authors' fixture suite, not a production deployment.

### Claims and sources {#claims-mp-2026-11-12-017}

- ActiveSecHarness composes capability tokens, taint tracking and deterministic operating-system guards. [source-2026-11-12-019] — Qualification: On 252 constructed fixtures, its deterministic guard blocked every attack with no reported benign false positives, while a local model judge refused most benign work; the evidence is from the authors' fixture suite, not a production deployment.

## 20. The Deepfake Detector Finds Provenance, Not Pulse {#mp-2026-11-12-018}

- Story ID: `mp-2026-11-12-018`
- Type: `ticker`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-11-12-018/the-deepfake-detector-finds-provenance-not-pulse

**Dek:** High-frequency video artifacts explain a near-perfect physiological-looking signal in a controlled audit.

The study finds that simple chrominance traces separate talking-face forgeries extremely well, but controlled tests place the signal outside the heart-rate band and show it in backgrounds and whole frames. Cross-dataset failures and overlapping identities further indicate that several detectors learn source-video provenance rather than a cardiac cue.

### Why it matters {#why-it-matters-mp-2026-11-12-018}

High-frequency video artifacts explain a near-perfect physiological-looking signal in a controlled audit.

### Limits and context {#limitations-mp-2026-11-12-018}

- No additional limitation was separately recorded.

### Claims and sources {#claims-mp-2026-11-12-018}

- High-frequency video artifacts explain a near-perfect physiological-looking signal in a controlled audit. [source-2026-11-12-020]

## 21. Biomedical Agents Meet Rebuilt Research Tasks {#mp-2026-11-12-019}

- Story ID: `mp-2026-11-12-019`
- Type: `ticker`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-11-12-019/biomedical-agents-meet-rebuilt-research-tasks

**Dek:** BioXArena-Pro holds out biological units and finds that scaffolding matters nearly as much as the model.

BioXArena-Pro rebuilds 64 modeling tasks across eight biomedical domains, usually holding out whole species or cohorts. Across 22 agent configurations the best score was 55.1 of 100, and removing pretrained biological models cost nine points on tasks where the leading agent used them.

### Why it matters {#why-it-matters-mp-2026-11-12-019}

BioXArena-Pro holds out biological units and finds that scaffolding matters nearly as much as the model.

### Limits and context {#limitations-mp-2026-11-12-019}

- No additional limitation was separately recorded.

### Claims and sources {#claims-mp-2026-11-12-019}

- BioXArena-Pro holds out biological units and finds that scaffolding matters nearly as much as the model. [source-2026-11-12-021]

## 22. Space Drums 2.0 {#mp-2026-11-12-020}

- Story ID: `mp-2026-11-12-020`
- Type: `invention_desk`
- Classification: `editorial`
- Content status: `carried_over`
- Permanent URL: https://themachinepress.com/story/mp-2026-11-12-020/space-drums-2-0

**Dek:** Moves motion tracking onto two ESP32-S3 drumsticks, adds haptic hits and a two-pedal hub, and sends velocity-sensitive drum events to open PC or Android playback software.

Moves motion tracking onto two ESP32-S3 drumsticks, adds haptic hits and a two-pedal hub, and sends velocity-sensitive drum events to open PC or Android playback software.

### Why it matters {#why-it-matters-mp-2026-11-12-020}

An independent builder is turning an improbable idea into a working project.

### Limits and context {#limitations-mp-2026-11-12-020}

- A Desk Pick is an editorial selection, not a product endorsement.

### Claims and sources {#claims-mp-2026-11-12-020}

- This Invention Desk entry makes no independently sourced news claim.

## 23. 3RIC {#mp-2026-11-12-021}

- Story ID: `mp-2026-11-12-021`
- Type: `invention_desk`
- Classification: `editorial`
- Content status: `carried_over`
- Permanent URL: https://themachinepress.com/story/mp-2026-11-12-021/3ric

**Dek:** Pairs a chip-level 65C02 computer with custom hardware, video, sound, floppy, and microSD support with a cycle-honest emulator and browser assembler built around the same machine.

Pairs a chip-level 65C02 computer with custom hardware, video, sound, floppy, and microSD support with a cycle-honest emulator and browser assembler built around the same machine.

### Why it matters {#why-it-matters-mp-2026-11-12-021}

An independent builder is turning an improbable idea into a working project.

### Limits and context {#limitations-mp-2026-11-12-021}

- A Desk Pick is an editorial selection, not a product endorsement.

### Claims and sources {#claims-mp-2026-11-12-021}

- This Invention Desk entry makes no independently sourced news claim.

## 24. M3-CRETE {#mp-2026-11-12-022}

- Story ID: `mp-2026-11-12-022`
- Type: `invention_desk`
- Classification: `editorial`
- Content status: `carried_over`
- Permanent URL: https://themachinepress.com/story/mp-2026-11-12-022/m3-crete

**Dek:** Publishes a pallet-scale cartesian motion reference, CAD, BOM, and low-voltage controls for supervised cementitious-extrusion research while explicitly excluding certified structural use.

Publishes a pallet-scale cartesian motion reference, CAD, BOM, and low-voltage controls for supervised cementitious-extrusion research while explicitly excluding certified structural use.

### Why it matters {#why-it-matters-mp-2026-11-12-022}

An independent builder is turning an improbable idea into a working project.

### Limits and context {#limitations-mp-2026-11-12-022}

- A Desk Pick is an editorial selection, not a product endorsement.

### Claims and sources {#claims-mp-2026-11-12-022}

- This Invention Desk entry makes no independently sourced news claim.

## 25. DeskNode {#mp-2026-11-12-023}

- Story ID: `mp-2026-11-12-023`
- Type: `invention_desk`
- Classification: `editorial`
- Content status: `carried_over`
- Permanent URL: https://themachinepress.com/story/mp-2026-11-12-023/desknode

**Dek:** Turns a bare ESP32-S3 touchscreen into a USB-connected desk instrument for PC activity and optional room temperature, humidity, and light measurements.

Turns a bare ESP32-S3 touchscreen into a USB-connected desk instrument for PC activity and optional room temperature, humidity, and light measurements.

### Why it matters {#why-it-matters-mp-2026-11-12-023}

An independent builder is turning an improbable idea into a working project.

### Limits and context {#limitations-mp-2026-11-12-023}

- A Desk Pick is an editorial selection, not a product endorsement.

### Claims and sources {#claims-mp-2026-11-12-023}

- This Invention Desk entry makes no independently sourced news claim.

## 26. The First Paid Slot {#mp-2026-11-12-024}

- Story ID: `mp-2026-11-12-024`
- Type: `invention_desk`
- Classification: `house_example`
- Content status: `carried_over`
- Permanent URL: https://themachinepress.com/story/mp-2026-11-12-024/the-first-paid-slot

**Dek:** A transparent preview of paid placement with one verified link and no claim of endorsement.

A transparent preview of paid placement with one verified link and no claim of endorsement.

House example - no advertiser paid. Payment will buy placement, never endorsement.

### Why it matters {#why-it-matters-mp-2026-11-12-024}

This placement explains how builders can appear in The Invention Desk without purchasing editorial endorsement.

### Limits and context {#limitations-mp-2026-11-12-024}

- House example - no advertiser paid. Payment will buy placement, never endorsement.

### Claims and sources {#claims-mp-2026-11-12-024}

- This Invention Desk entry makes no independently sourced news claim.

## 27. Put Your Project on the Desk {#mp-2026-11-12-025}

- Story ID: `mp-2026-11-12-025`
- Type: `invention_desk`
- Classification: `house_example`
- Content status: `carried_over`
- Permanent URL: https://themachinepress.com/story/mp-2026-11-12-025/put-your-project-on-the-desk

**Dek:** One manually reviewed placement stays active for seven days and remains separate from Desk Picks.

One manually reviewed placement stays active for seven days and remains separate from Desk Picks.

Manual intake only. Payment buys placement, never endorsement, and every submission is reviewed.

### Why it matters {#why-it-matters-mp-2026-11-12-025}

This placement explains how builders can appear in The Invention Desk without purchasing editorial endorsement.

### Limits and context {#limitations-mp-2026-11-12-025}

- Manual intake only. Payment buys placement, never endorsement, and every submission is reviewed.

### Claims and sources {#claims-mp-2026-11-12-025}

- This Invention Desk entry makes no independently sourced news claim.

## Normalized sources

- **source-2026-11-12-001:** [arXiv:2611.13755v1](https://arxiv.org/abs/2611.13755v1) — arXiv / Cornell University; primary_research
- **source-2026-11-12-002:** [arXiv:2611.13251v1](https://arxiv.org/abs/2611.13251v1) — arXiv / Cornell University; primary_research
- **source-2026-11-12-003:** [arXiv:2611.13241v1](https://arxiv.org/abs/2611.13241v1) — arXiv / Cornell University; primary_research
- **source-2026-11-12-004:** [arXiv:2611.13246v1](https://arxiv.org/abs/2611.13246v1) — arXiv / Cornell University; primary_research
- **source-2026-11-12-005:** [arXiv:2611.13254v1](https://arxiv.org/abs/2611.13254v1) — arXiv / Cornell University; primary_research
- **source-2026-11-12-006:** [arXiv:2611.13271v1](https://arxiv.org/abs/2611.13271v1) — arXiv / Cornell University; primary_research
- **source-2026-11-12-007:** [arXiv:2611.13329v1](https://arxiv.org/abs/2611.13329v1) — arXiv / Cornell University; primary_research
- **source-2026-11-12-008:** [arXiv:2611.13344v1](https://arxiv.org/abs/2611.13344v1) — arXiv / Cornell University; primary_research
- **source-2026-11-12-009:** [arXiv:2611.13348v1](https://arxiv.org/abs/2611.13348v1) — arXiv / Cornell University; primary_research
- **source-2026-11-12-010:** [arXiv:2611.13355v1](https://arxiv.org/abs/2611.13355v1) — arXiv / Cornell University; primary_research
- **source-2026-11-12-011:** [arXiv:2611.13398v1](https://arxiv.org/abs/2611.13398v1) — arXiv / Cornell University; primary_research
- **source-2026-11-12-012:** [arXiv:2611.13655v1](https://arxiv.org/abs/2611.13655v1) — arXiv / Cornell University; primary_research
- **source-2026-11-12-013:** [arXiv:2611.13630v1](https://arxiv.org/abs/2611.13630v1) — arXiv / Cornell University; primary_research
- **source-2026-11-12-014:** [arXiv:2611.13873v1](https://arxiv.org/abs/2611.13873v1) — arXiv / Cornell University; primary_research
- **source-2026-11-12-015:** [arXiv:2611.13879v1](https://arxiv.org/abs/2611.13879v1) — arXiv / Cornell University; primary_research
- **source-2026-11-12-016:** [arXiv:2611.13903v1](https://arxiv.org/abs/2611.13903v1) — arXiv / Cornell University; primary_research
- **source-2026-11-12-017:** [arXiv:2611.13922v1](https://arxiv.org/abs/2611.13922v1) — arXiv / Cornell University; primary_research
- **source-2026-11-12-018:** [arXiv:2611.13931v1](https://arxiv.org/abs/2611.13931v1) — arXiv / Cornell University; primary_research
- **source-2026-11-12-019:** [arXiv:2611.13960v1](https://arxiv.org/abs/2611.13960v1) — arXiv / Cornell University; primary_research
- **source-2026-11-12-020:** [arXiv:2611.13992v1](https://arxiv.org/abs/2611.13992v1) — arXiv / Cornell University; primary_research
- **source-2026-11-12-021:** [arXiv:2611.13994v1](https://arxiv.org/abs/2611.13994v1) — arXiv / Cornell University; primary_research