---
schema_version: "1.0.0"
edition_id: "mp-2026-09-27-morning-0080"
published_at: "2026-09-27T09:00:00.000-04:00"
modified_at: "2026-09-27T09:00:00.000-04:00"
canonical_url: "https://themachinepress.com/edition/2026-09-27"
story_count: 27
lead_story_id: "mp-2026-09-27-001"
---

# The Machine Press — Morning edition

Edition ID: `mp-2026-09-27-morning-0080`  
Published: 2026-09-27T09:00:00.000-04:00  
Canonical edition: https://themachinepress.com/edition/2026-09-27

In multi-agent deliberation, defection rose with the proportion of deceptive agents—not the size of the group.

## 1. A Minority of Deceivers Could Turn the Group {#mp-2026-09-27-001}

- Story ID: `mp-2026-09-27-001`
- Type: `lead`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-09-27-001/a-minority-of-deceivers-could-turn-the-group

**Dek:** In multi-agent deliberation, defection rose with the proportion of deceptive agents—not the size of the group.

A new preprint tested how deceptive participants alter collective decisions by varying both group size and the share of adversarial agents. Across the reported setups, defection rose roughly linearly with the proportion of deceivers, while simply adding more agents did not produce the same scaling effect. Language-model agents could be swayed even when deceivers remained a minority, although private coordination sometimes made the attackers less effective by preventing public reinforcement. The result is a benchmark finding about controlled deliberation, not evidence that every deployed agent team will behave the same way. It points system designers toward adversarial composition and communication structure as separate risks to measure.

### Why it matters {#why-it-matters-mp-2026-09-27-001}

In multi-agent deliberation, defection rose with the proportion of deceptive agents—not the size of the group.

### Limits and context {#limitations-mp-2026-09-27-001}

- Across the reported setups, defection rose roughly linearly with the proportion of deceivers, while simply adding more agents did not produce the same scaling effect.
- The result is a benchmark finding about controlled deliberation, not evidence that every deployed agent team will behave the same way.

### Claims and sources {#claims-mp-2026-09-27-001}

- In multi-agent deliberation, defection rose with the proportion of deceptive agents—not the size of the group. [source-2026-09-27-001] — Qualification: Across the reported setups, defection rose roughly linearly with the proportion of deceivers, while simply adding more agents did not produce the same scaling effect.

## 2. The Benchmark Hid the Question {#mp-2026-09-27-002}

- Story ID: `mp-2026-09-27-002`
- Type: `secondary`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-09-27-002/the-benchmark-hid-the-question

**Dek:** EnigmaForge asked models to infer both the puzzle and its answer from documents, producing a 22-fold spread in intuition.

EnigmaForge replaces the usual explicit question with a packet of documents whose facts imply both what must be solved and the one admissible answer. Its generator uses satisfiability checks to create puzzles with a unique solution and to verify that every clue is load-bearing. Across 25 frontier models, 600 instances and 17,400 evaluation records, the authors report a 22-fold spread in reconstructing the hidden problem, compared with a 1.6-fold spread in recovering stated facts. They also filter refusals separately so a model that declines the task is not confused with one that reasons incorrectly. This is a synthetic benchmark preprint, but it isolates a practical failure mode: gathering facts can be much easier than discovering the question those facts were meant to answer.

### Why it matters {#why-it-matters-mp-2026-09-27-002}

EnigmaForge asked models to infer both the puzzle and its answer from documents, producing a 22-fold spread in intuition.

### Limits and context {#limitations-mp-2026-09-27-002}

- They also filter refusals separately so a model that declines the task is not confused with one that reasons incorrectly.

### Claims and sources {#claims-mp-2026-09-27-002}

- EnigmaForge asked models to infer both the puzzle and its answer from documents, producing a 22-fold spread in intuition. [source-2026-09-27-002] — Qualification: They also filter refusals separately so a model that declines the task is not confused with one that reasons incorrectly.

## 3. An Agent Forgot Its Reasoning After the State Was Saved {#mp-2026-09-27-003}

- Story ID: `mp-2026-09-27-003`
- Type: `dispatch`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-09-27-003/an-agent-forgot-its-reasoning-after-the-state-was-saved

**Dek:** A training-free compressor cut input, output and cache tokens while raising reported reward on WorkBuddyBench.

The ICLR method ranks completed reasoning blocks by whether they still matter after an agent has externalized useful state into files, tools or the environment. On 260 WorkBuddyBench tasks, the authors report reward rising from 0.699 to 0.718 while input, output and cache tokens fell by 25.5%, 14.4% and 33.3%, respectively. The study argues that internal reasoning can become replaceable once its consequences are safely recorded, though the result is benchmark-specific and does not license deleting audit evidence.

### Why it matters {#why-it-matters-mp-2026-09-27-003}

A training-free compressor cut input, output and cache tokens while raising reported reward on WorkBuddyBench.

### Limits and context {#limitations-mp-2026-09-27-003}

- The study argues that internal reasoning can become replaceable once its consequences are safely recorded, though the result is benchmark-specific and does not license deleting audit evidence.

### Claims and sources {#claims-mp-2026-09-27-003}

- A training-free compressor cut input, output and cache tokens while raising reported reward on WorkBuddyBench. [source-2026-09-27-003] — Qualification: The study argues that internal reasoning can become replaceable once its consequences are safely recorded, though the result is benchmark-specific and does not license deleting audit evidence.

## 4. The Planner Put Stakeholder Rules Into an Ontology {#mp-2026-09-27-004}

- Story ID: `mp-2026-09-27-004`
- Type: `dispatch`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-09-27-004/the-planner-put-stakeholder-rules-into-an-ontology

**Dek:** Hard and soft constraints became explainable objects that could be checked, ranked and renegotiated.

The proposed framework encodes multi-stakeholder requirements in an OWL ontology, separating hard limits from preferences before a planner acts. It can identify unsatisfiable combinations, explain which constraints conflict and support structured renegotiation rather than silently relaxing a rule. A FLEXI microgrid case study illustrates the approach; the preprint is an architecture and demonstration, not a broad operational deployment.

### Why it matters {#why-it-matters-mp-2026-09-27-004}

Hard and soft constraints became explainable objects that could be checked, ranked and renegotiated.

### Limits and context {#limitations-mp-2026-09-27-004}

- A FLEXI microgrid case study illustrates the approach; the preprint is an architecture and demonstration, not a broad operational deployment.

### Claims and sources {#claims-mp-2026-09-27-004}

- Hard and soft constraints became explainable objects that could be checked, ranked and renegotiated. [source-2026-09-27-004] — Qualification: A FLEXI microgrid case study illustrates the approach; the preprint is an architecture and demonstration, not a broad operational deployment.

## 5. A Mobile Agent Trained the Planner Around One Shared Contract {#mp-2026-09-27-005}

- Story ID: `mp-2026-09-27-005`
- Type: `dispatch`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-09-27-005/a-mobile-agent-trained-the-planner-around-one-shared-contract

**Dek:** Qwen-Planner-Agent aligned data, training and deployment through a common action representation.

Qwen-Planner-Agent uses the same contract for collecting trajectories, training a planner and executing actions on devices. Its CARE process is designed to correct action representations before errors propagate across that pipeline. The authors report the best overall result in their MobilePA-Bench evaluation while preserving general capabilities, but the claim belongs to the paper's chosen tasks, baselines and device conditions.

### Why it matters {#why-it-matters-mp-2026-09-27-005}

Qwen-Planner-Agent aligned data, training and deployment through a common action representation.

### Limits and context {#limitations-mp-2026-09-27-005}

- No additional limitation was separately recorded.

### Claims and sources {#claims-mp-2026-09-27-005}

- Qwen-Planner-Agent aligned data, training and deployment through a common action representation. [source-2026-09-27-005]

## 6. Industrial Configuration Needs More Than a Language Model {#mp-2026-09-27-006}

- Story ID: `mp-2026-09-27-006`
- Type: `dispatch`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-09-27-006/industrial-configuration-needs-more-than-a-language-model

**Dek:** A survey organized neuro-symbolic copilot designs around inference, fine-tuning and training-time integration.

The preprint argues that unconstrained language generation is too unreliable for configuration work governed by compatibility rules, catalogs and engineering limits. It classifies hybrid systems by where symbolic knowledge enters: at inference, during fine-tuning or inside training objectives. An industrial configuration copilot grounds the discussion, while the paper also flags scaling, maintenance and evaluation problems that remain unresolved.

### Why it matters {#why-it-matters-mp-2026-09-27-006}

A survey organized neuro-symbolic copilot designs around inference, fine-tuning and training-time integration.

### Limits and context {#limitations-mp-2026-09-27-006}

- An industrial configuration copilot grounds the discussion, while the paper also flags scaling, maintenance and evaluation problems that remain unresolved.

### Claims and sources {#claims-mp-2026-09-27-006}

- A survey organized neuro-symbolic copilot designs around inference, fine-tuning and training-time integration. [source-2026-09-27-006] — Qualification: An industrial configuration copilot grounds the discussion, while the paper also flags scaling, maintenance and evaluation problems that remain unresolved.

## 7. A Learned Prefix Cut Model Utility by 26.8 Points {#mp-2026-09-27-007}

- Story ID: `mp-2026-09-27-007`
- Type: `dispatch`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-09-27-007/a-learned-prefix-cut-model-utility-by-26-8-points

**Dek:** ENDOPROMPT optimized harmful prefixes from unlabeled instructions and produced losses across almost every tested split.

ENDOPROMPT uses white-box optimization to derive input prefixes without requiring labeled attack examples. Across four models, seven benign benchmarks and 28 model-benchmark combinations, the authors report a mean utility loss of 26.8 percentage points and negative effects in 27 of 28 cases. Their controls did not establish a compensating request-matching benefit; the preprint promises code upon acceptance, so independent reproduction remains a next step.

### Why it matters {#why-it-matters-mp-2026-09-27-007}

ENDOPROMPT optimized harmful prefixes from unlabeled instructions and produced losses across almost every tested split.

### Limits and context {#limitations-mp-2026-09-27-007}

- Their controls did not establish a compensating request-matching benefit; the preprint promises code upon acceptance, so independent reproduction remains a next step.

### Claims and sources {#claims-mp-2026-09-27-007}

- ENDOPROMPT optimized harmful prefixes from unlabeled instructions and produced losses across almost every tested split. [source-2026-09-27-007] — Qualification: Their controls did not establish a compensating request-matching benefit; the preprint promises code upon acceptance, so independent reproduction remains a next step.

## 8. Code Attribution Mostly Found Length, Not Identity {#mp-2026-09-27-008}

- Story ID: `mp-2026-09-27-008`
- Type: `dispatch`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-09-27-008/code-attribution-mostly-found-length-not-identity

**Dek:** Across 15 model-benchmark combinations, balanced accuracy stayed near chance and tracked solution length.

A study of model self-attribution found balanced accuracy between 49% and 58% across 15 model-benchmark combinations. Pairwise attribution scores correlated at r=0.93 with longer solutions, suggesting that style and verbosity were doing much of the work. After normalization, 10 of 12 tested conditions fell to chance and the remaining two continued to follow length, erasing the apparent self-preference in this setup.

### Why it matters {#why-it-matters-mp-2026-09-27-008}

Across 15 model-benchmark combinations, balanced accuracy stayed near chance and tracked solution length.

### Limits and context {#limitations-mp-2026-09-27-008}

- No additional limitation was separately recorded.

### Claims and sources {#claims-mp-2026-09-27-008}

- Across 15 model-benchmark combinations, balanced accuracy stayed near chance and tracked solution length. [source-2026-09-27-008]

## 9. A Verification Toolkit Added Probabilistic Reachability {#mp-2026-09-27-009}

- Story ID: `mp-2026-09-27-009`
- Type: `dispatch`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-09-27-009/a-verification-toolkit-added-probabilistic-reachability

**Dek:** NNV3 joined new star-set abstractions with fairness checks for modern neural architectures.

NNV3 introduces ModelStar, VolumeStar and GraphStar abstractions for neural-network reachability, including probabilistic analysis and a FairNNV module. The authors benchmark the toolkit on malware, power, medical, time-series and action-recognition models. Those experiments show breadth across architectures, but the preprint does not make every network or safety property automatically tractable.

### Why it matters {#why-it-matters-mp-2026-09-27-009}

NNV3 joined new star-set abstractions with fairness checks for modern neural architectures.

### Limits and context {#limitations-mp-2026-09-27-009}

- Those experiments show breadth across architectures, but the preprint does not make every network or safety property automatically tractable.

### Claims and sources {#claims-mp-2026-09-27-009}

- NNV3 joined new star-set abstractions with fairness checks for modern neural architectures. [source-2026-09-27-009] — Qualification: Those experiments show breadth across architectures, but the preprint does not make every network or safety property automatically tractable.

## 10. Execution Feedback Raised Scientific Coding Accuracy 9.9 Points {#mp-2026-09-27-010}

- Story ID: `mp-2026-09-27-010`
- Type: `dispatch`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-09-27-010/execution-feedback-raised-scientific-coding-accuracy-9-9-points

**Dek:** SciWalker organized 8,178 problems across five domains into operator graphs that could be tested while solving.

SciWalker contains 8,178 scientific coding problems spanning five domains and 32 subdomains, with solutions represented as operator graphs. The framework uses execution feedback to revise intermediate programs rather than grading only a final answer. In the reported experiment, Qwen3.5-9B accuracy rose from 29.3% to 39.2%; that gain applies to this benchmark and training recipe, not scientific software in general.

### Why it matters {#why-it-matters-mp-2026-09-27-010}

SciWalker organized 8,178 problems across five domains into operator graphs that could be tested while solving.

### Limits and context {#limitations-mp-2026-09-27-010}

- The framework uses execution feedback to revise intermediate programs rather than grading only a final answer.
- In the reported experiment, Qwen3.5-9B accuracy rose from 29.3% to 39.2%; that gain applies to this benchmark and training recipe, not scientific software in general.

### Claims and sources {#claims-mp-2026-09-27-010}

- SciWalker organized 8,178 problems across five domains into operator graphs that could be tested while solving. [source-2026-09-27-010] — Qualification: The framework uses execution feedback to revise intermediate programs rather than grading only a final answer.

## 11. A Model Practiced Prediction Without Natural Data {#mp-2026-09-27-011}

- Story ID: `mp-2026-09-27-011`
- Type: `dispatch`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-09-27-011/a-model-practiced-prediction-without-natural-data

**Dek:** Generator programs supplied an open-ended self-play curriculum built from a universal Turing machine.

The proposed pretraining scheme pits a generator program against a learner that predicts the generator's output, using reinforcement learning to keep the curriculum challenging. The authors report that zero-shot loss on natural data scales with compute even though natural examples never enter pretraining. They also observe in-context learning and mathematical-sequence behavior, but this remains a controlled preprint result rather than a replacement recipe for production language-model training.

### Why it matters {#why-it-matters-mp-2026-09-27-011}

Generator programs supplied an open-ended self-play curriculum built from a universal Turing machine.

### Limits and context {#limitations-mp-2026-09-27-011}

- No additional limitation was separately recorded.

### Claims and sources {#claims-mp-2026-09-27-011}

- Generator programs supplied an open-ended self-play curriculum built from a universal Turing machine. [source-2026-09-27-011]

## 12. Compiled Skills Cut Agent Tokens by as Much as 88.9% {#mp-2026-09-27-012}

- Story ID: `mp-2026-09-27-012`
- Type: `dispatch`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-09-27-012/compiled-skills-cut-agent-tokens-by-as-much-as-88-9

**Dek:** HEXIS turned reusable procedures into extended state machines with static checks and replay.

HEXIS compiles agent skills into extended finite-state machines whose transitions, guards and tool calls can be inspected before execution. Across four benchmarks and four executors, the authors report a 16.1-point gain over Skill+ReAct. For Qwen3.8-27B, token use fell between 38.4% and 88.9%, while accepted traces could be replayed; the results depend on the tasks and skill library tested.

### Why it matters {#why-it-matters-mp-2026-09-27-012}

HEXIS turned reusable procedures into extended state machines with static checks and replay.

### Limits and context {#limitations-mp-2026-09-27-012}

- No additional limitation was separately recorded.

### Claims and sources {#claims-mp-2026-09-27-012}

- HEXIS turned reusable procedures into extended state machines with static checks and replay. [source-2026-09-27-012]

## 13. A Robot Mapped Space Structures Without Markers {#mp-2026-09-27-013}

- Story ID: `mp-2026-09-27-013`
- Type: `dispatch`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-09-27-013/a-robot-mapped-space-structures-without-markers

**Dek:** The inspection stack combined RGB-D, thermal and lidar sensing on a Kinova arm for large-structure reconstruction.

Researchers built a markerless multimodal inspection stack around a Kinova Gen2 arm, an RGB-D camera, a thermal camera and 2D lidar. The system plans viewpoints and motion while synchronizing the sensor streams into geometric and thermal reconstructions. Gazebo studies and preliminary laboratory work produced usable models in the reported setup; this iSpaRo-accepted paper is an early inspection result, not an on-orbit demonstration.

### Why it matters {#why-it-matters-mp-2026-09-27-013}

The inspection stack combined RGB-D, thermal and lidar sensing on a Kinova arm for large-structure reconstruction.

### Limits and context {#limitations-mp-2026-09-27-013}

- Gazebo studies and preliminary laboratory work produced usable models in the reported setup; this iSpaRo-accepted paper is an early inspection result, not an on-orbit demonstration.

### Claims and sources {#claims-mp-2026-09-27-013}

- The inspection stack combined RGB-D, thermal and lidar sensing on a Kinova arm for large-structure reconstruction. [source-2026-09-27-013] — Qualification: Gazebo studies and preliminary laboratory work produced usable models in the reported setup; this iSpaRo-accepted paper is an early inspection result, not an on-orbit demonstration.

## 14. A Robot Gripper Recorded Touch, Sound and Motion Together {#mp-2026-09-27-014}

- Story ID: `mp-2026-09-27-014`
- Type: `dispatch`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-09-27-014/a-robot-gripper-recorded-touch-sound-and-motion-together

**Dek:** PolyUMI packages vision, tactile, audio and proprioception into a transferable wireless demonstration tool.

PolyUMI combines a wrist camera, optical tactile finger, microphone and proprioceptive sensing in a wireless gripper for collecting demonstrations. Its transferable sensing finger and VisTA dataset are meant to preserve more contact information than vision alone. In the paper's experiments, touch and audio added useful signals and matched or exceeded existing policies on selected tasks; the evidence remains bounded to those demonstrations and hardware.

### Why it matters {#why-it-matters-mp-2026-09-27-014}

PolyUMI packages vision, tactile, audio and proprioception into a transferable wireless demonstration tool.

### Limits and context {#limitations-mp-2026-09-27-014}

- No additional limitation was separately recorded.

### Claims and sources {#claims-mp-2026-09-27-014}

- PolyUMI packages vision, tactile, audio and proprioception into a transferable wireless demonstration tool. [source-2026-09-27-014]

## 15. Behavior Cloning Dropped Training for Retrieval {#mp-2026-09-27-026}

- Story ID: `mp-2026-09-27-026`
- Type: `dispatch`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-09-27-026/behavior-cloning-dropped-training-for-retrieval

**Dek:** BPC used demonstration windows, Hankel structure and action correction to fit in seconds and run above 75 hertz.

Behavioral Predictive Control retrieves relevant windows from demonstrations, organizes them through Hankel structure and corrects the proposed action online. The authors report competitive performance with learned policies and better results on some tested tasks while cutting fitting time from hours to seconds on consumer GPUs. They also report control above 75 hertz on a Jetson Orin Nano, with each action traceable to demonstration windows; wider robustness is not yet established.

### Why it matters {#why-it-matters-mp-2026-09-27-026}

BPC used demonstration windows, Hankel structure and action correction to fit in seconds and run above 75 hertz.

### Limits and context {#limitations-mp-2026-09-27-026}

- They also report control above 75 hertz on a Jetson Orin Nano, with each action traceable to demonstration windows; wider robustness is not yet established.

### Claims and sources {#claims-mp-2026-09-27-026}

- BPC used demonstration windows, Hankel structure and action correction to fit in seconds and run above 75 hertz. [source-2026-09-27-015] — Qualification: They also report control above 75 hertz on a Jetson Orin Nano, with each action traceable to demonstration windows; wider robustness is not yet established.

## 16. One Inverse-Kinematics Model Crossed Unseen Robot Chains {#mp-2026-09-27-027}

- Story ID: `mp-2026-09-27-027`
- Type: `dispatch`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-09-27-027/one-inverse-kinematics-model-crossed-unseen-robot-chains

**Dek:** MorphIK reached roughly five-centimeter error before refinement and sub-centimeter error after one DLS step.

MorphIK is trained entirely on synthetic kinematic chains so one model can propose solutions for previously unseen robots with six to nine degrees of freedom. The authors report about five-centimeter error on real platforms before refinement, below one centimeter after one damped-least-squares step and below one millimeter after three steps in most tested cases. Its samples also cover diverse null-space solutions, but the results remain tied to the evaluated chains, targets and refinement procedure.

### Why it matters {#why-it-matters-mp-2026-09-27-027}

MorphIK reached roughly five-centimeter error before refinement and sub-centimeter error after one DLS step.

### Limits and context {#limitations-mp-2026-09-27-027}

- Its samples also cover diverse null-space solutions, but the results remain tied to the evaluated chains, targets and refinement procedure.

### Claims and sources {#claims-mp-2026-09-27-027}

- MorphIK reached roughly five-centimeter error before refinement and sub-centimeter error after one DLS step. [source-2026-09-27-016] — Qualification: Its samples also cover diverse null-space solutions, but the results remain tied to the evaluated chains, targets and refinement procedure.

## 17. Pseudoentanglement Matched EFI Pairs {#mp-2026-09-27-015}

- Story ID: `mp-2026-09-27-015`
- Type: `ticker`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-09-27-015/pseudoentanglement-matched-efi-pairs

**Dek:** Under one operational definition, the two cryptographic resources became equivalent.

The theoretical preprint connects pseudoentangled states with efficiently samplable far-but-indistinguishable pairs. Its equivalence result is definition-specific and does not describe a hardware protocol.

### Why it matters {#why-it-matters-mp-2026-09-27-015}

Under one operational definition, the two cryptographic resources became equivalent.

### Limits and context {#limitations-mp-2026-09-27-015}

- Its equivalence result is definition-specific and does not describe a hardware protocol.

### Claims and sources {#claims-mp-2026-09-27-015}

- Under one operational definition, the two cryptographic resources became equivalent. [source-2026-09-27-017] — Qualification: Its equivalence result is definition-specific and does not describe a hardware protocol.

## 18. A Universal Detector Reached First-Order Capacity {#mp-2026-09-27-016}

- Story ID: `mp-2026-09-27-016`
- Type: `ticker`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-09-27-016/a-universal-detector-reached-first-order-capacity

**Dek:** State-independent and channel-independent schemes approached the optimal correlation rate.

The authors construct a universal correlation detector and capacity-achieving schemes that do not require exact source or channel knowledge. The result is asymptotic and theoretical, not a deployed communications system.

### Why it matters {#why-it-matters-mp-2026-09-27-016}

State-independent and channel-independent schemes approached the optimal correlation rate.

### Limits and context {#limitations-mp-2026-09-27-016}

- The authors construct a universal correlation detector and capacity-achieving schemes that do not require exact source or channel knowledge.
- The result is asymptotic and theoretical, not a deployed communications system.

### Claims and sources {#claims-mp-2026-09-27-016}

- State-independent and channel-independent schemes approached the optimal correlation rate. [source-2026-09-27-018] — Qualification: The authors construct a universal correlation detector and capacity-achieving schemes that do not require exact source or channel knowledge.

## 19. High-Rate Quantum Codes Kept Checks at Weight Ten {#mp-2026-09-27-017}

- Story ID: `mp-2026-09-27-017`
- Type: `ticker`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-09-27-017/high-rate-quantum-codes-kept-checks-at-weight-ten

**Dek:** New constructions include [[90,21,11]], [[140,31,15]] and [[200,43,20]] parameter sets.

The preprint presents ultra-high-rate quantum LDPC code families and analyzes their logical performance at a 0.1% physical error rate. Those figures are code and noise-model results, not measurements from a fabricated quantum processor.

### Why it matters {#why-it-matters-mp-2026-09-27-017}

New constructions include [[90,21,11]], [[140,31,15]] and [[200,43,20]] parameter sets.

### Limits and context {#limitations-mp-2026-09-27-017}

- Those figures are code and noise-model results, not measurements from a fabricated quantum processor.

### Claims and sources {#claims-mp-2026-09-27-017}

- New constructions include [[90,21,11]], [[140,31,15]] and [[200,43,20]] parameter sets. [source-2026-09-27-019] — Qualification: Those figures are code and noise-model results, not measurements from a fabricated quantum processor.

## 20. A Tensor Network Kept Long-Range Ties With Fewer Parameters {#mp-2026-09-27-018}

- Story ID: `mp-2026-09-27-018`
- Type: `ticker`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-09-27-018/a-tensor-network-kept-long-range-ties-with-fewer-parameters

**Dek:** LETTA used about an order of magnitude fewer parameters than much larger matrix-product states in tested models.

LETTA augments a tensor-network ansatz to represent long-range physical correlations more directly. The reported parameter advantage comes from selected theoretical models and does not yet establish a universal simulation speedup.

### Why it matters {#why-it-matters-mp-2026-09-27-018}

LETTA used about an order of magnitude fewer parameters than much larger matrix-product states in tested models.

### Limits and context {#limitations-mp-2026-09-27-018}

- The reported parameter advantage comes from selected theoretical models and does not yet establish a universal simulation speedup.

### Claims and sources {#claims-mp-2026-09-27-018}

- LETTA used about an order of magnitude fewer parameters than much larger matrix-product states in tested models. [source-2026-09-27-020] — Qualification: The reported parameter advantage comes from selected theoretical models and does not yet establish a universal simulation speedup.

## 21. Entanglement Closed the Spectrum-Estimation Bound {#mp-2026-09-27-019}

- Story ID: `mp-2026-09-27-019`
- Type: `ticker`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-09-27-019/entanglement-closed-the-spectrum-estimation-bound

**Dek:** An estimator matched the lower bound while unentangled strategies paid an extra factor in dimension.

The paper gives an optimal quantum spectrum-estimation procedure and separates it from unentangled measurements by a factor tied to system dimension. The authors conjecture the unentangled bound is itself optimal; this is a complexity result, not an experimental measurement.

### Why it matters {#why-it-matters-mp-2026-09-27-019}

An estimator matched the lower bound while unentangled strategies paid an extra factor in dimension.

### Limits and context {#limitations-mp-2026-09-27-019}

- The authors conjecture the unentangled bound is itself optimal; this is a complexity result, not an experimental measurement.

### Claims and sources {#claims-mp-2026-09-27-019}

- An estimator matched the lower bound while unentangled strategies paid an extra factor in dimension. [source-2026-09-27-021] — Qualification: The authors conjecture the unentangled bound is itself optimal; this is a complexity result, not an experimental measurement.

## 22. Space Drums 2.0 {#mp-2026-09-27-020}

- Story ID: `mp-2026-09-27-020`
- Type: `invention_desk`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-09-27-020/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-2026-09-27-020}

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

### Limits and context {#limitations-mp-2026-09-27-020}

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

### Claims and sources {#claims-mp-2026-09-27-020}

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

## 23. 3RIC {#mp-2026-09-27-021}

- Story ID: `mp-2026-09-27-021`
- Type: `invention_desk`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-09-27-021/3ric

**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-2026-09-27-021}

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

### Limits and context {#limitations-mp-2026-09-27-021}

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

### Claims and sources {#claims-mp-2026-09-27-021}

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

## 24. M3-CRETE {#mp-2026-09-27-022}

- Story ID: `mp-2026-09-27-022`
- Type: `invention_desk`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-09-27-022/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-2026-09-27-022}

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

### Limits and context {#limitations-mp-2026-09-27-022}

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

### Claims and sources {#claims-mp-2026-09-27-022}

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

## 25. DeskNode {#mp-2026-09-27-023}

- Story ID: `mp-2026-09-27-023`
- Type: `invention_desk`
- Classification: `editorial`
- Content status: `new`
- Permanent URL: https://themachinepress.com/story/mp-2026-09-27-023/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-2026-09-27-023}

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

### Limits and context {#limitations-mp-2026-09-27-023}

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

### Claims and sources {#claims-mp-2026-09-27-023}

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

## 26. The First Paid Slot {#mp-2026-09-27-024}

- Story ID: `mp-2026-09-27-024`
- Type: `invention_desk`
- Classification: `house_example`
- Content status: `carried_over`
- Permanent URL: https://themachinepress.com/story/mp-2026-09-27-024/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-2026-09-27-024}

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

### Limits and context {#limitations-mp-2026-09-27-024}

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

### Claims and sources {#claims-mp-2026-09-27-024}

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

## 27. Put Your Project on the Desk {#mp-2026-09-27-025}

- Story ID: `mp-2026-09-27-025`
- Type: `invention_desk`
- Classification: `house_example`
- Content status: `carried_over`
- Permanent URL: https://themachinepress.com/story/mp-2026-09-27-025/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-2026-09-27-025}

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

### Limits and context {#limitations-mp-2026-09-27-025}

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

### Claims and sources {#claims-mp-2026-09-27-025}

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

## Normalized sources

- **source-2026-09-27-001:** [arXiv preprint 2609.30028](https://arxiv.org/abs/2609.30028) — arXiv; primary_research
- **source-2026-09-27-002:** [arXiv preprint 2609.30144](https://arxiv.org/abs/2609.30144) — arXiv; primary_research
- **source-2026-09-27-003:** [arXiv preprint 2609.29875](https://arxiv.org/abs/2609.29875) — arXiv; primary_research
- **source-2026-09-27-004:** [arXiv preprint 2609.29876](https://arxiv.org/abs/2609.29876) — arXiv; primary_research
- **source-2026-09-27-005:** [arXiv preprint 2609.29892](https://arxiv.org/abs/2609.29892) — arXiv; primary_research
- **source-2026-09-27-006:** [arXiv preprint 2609.29947](https://arxiv.org/abs/2609.29947) — arXiv; primary_research
- **source-2026-09-27-007:** [arXiv preprint 2609.29948](https://arxiv.org/abs/2609.29948) — arXiv; primary_research
- **source-2026-09-27-008:** [arXiv preprint 2609.30048](https://arxiv.org/abs/2609.30048) — arXiv; primary_research
- **source-2026-09-27-009:** [arXiv preprint 2609.30050](https://arxiv.org/abs/2609.30050) — arXiv; primary_research
- **source-2026-09-27-010:** [arXiv preprint 2609.30054](https://arxiv.org/abs/2609.30054) — arXiv; primary_research
- **source-2026-09-27-011:** [arXiv preprint 2609.30063](https://arxiv.org/abs/2609.30063) — arXiv; primary_research
- **source-2026-09-27-012:** [arXiv preprint 2609.30123](https://arxiv.org/abs/2609.30123) — arXiv; primary_research
- **source-2026-09-27-013:** [arXiv preprint 2609.29644](https://arxiv.org/abs/2609.29644) — arXiv; primary_research
- **source-2026-09-27-014:** [arXiv preprint 2609.29760](https://arxiv.org/abs/2609.29760) — arXiv; primary_research
- **source-2026-09-27-015:** [arXiv preprint 2609.30134](https://arxiv.org/abs/2609.30134) — arXiv; primary_research
- **source-2026-09-27-016:** [arXiv preprint 2609.29908](https://arxiv.org/abs/2609.29908) — arXiv; primary_research
- **source-2026-09-27-017:** [arXiv preprint 2609.29917](https://arxiv.org/abs/2609.29917) — arXiv; primary_research
- **source-2026-09-27-018:** [arXiv preprint 2609.29954](https://arxiv.org/abs/2609.29954) — arXiv; primary_research
- **source-2026-09-27-019:** [arXiv preprint 2609.30069](https://arxiv.org/abs/2609.30069) — arXiv; primary_research
- **source-2026-09-27-020:** [arXiv preprint 2609.30101](https://arxiv.org/abs/2609.30101) — arXiv; primary_research
- **source-2026-09-27-021:** [arXiv preprint 2609.30171](https://arxiv.org/abs/2609.30171) — arXiv; primary_research