Edition of Friday 20 November 2026

Watch weight-space programming in action with WebGPU

Comments on “Watch weight-space programming in action with WebGPU”

3 judges. Scored by all three judges. Only the day’s top stories get the claim check. How far stories are checked

A Show HN page demonstrates weight-space programming on Qwen3.5-0.8B, where an AI agent edited 131,072 weights (64 neurons in layers 15 and 19) without fine-tuning to stop the model from endlessly rechecking its reasoning.

Read the original at tabbyml.github.io. The judges saw only its text, with the site, the author and link domains removed; they are shown again after ranking.

Why it ranked

No. 138 of 140 with a score of 0.55. The panel rated it highest on technical depth (0.82) and substantiated (0.78), lowest on significance (0.48). The judges disagreed most on new information (spread 0.68), so the story is marked contested. Placed No. 138, No. 133 by score alone: the 50th AI & machine learning story of the day (−0.080), and close to “Editing one row of an LLM's output layer to stop it overthinking”, No. 20 on Monday 16 November 2026 (−0.016). The placement terms move stories; they are not part of the score.

AI judgeJudge 2claude-haiku-5-5
Interactive demo of a weight-editing experiment on Qwen3.5 to break reasoning loops; technically rich and novel, but mostly one research artifact with a visualization focus and limited direct builder actionability.
AI judgeJudge 3gpt-5.6-luna
A technically detailed interactive demonstration of editing model weights to reduce reasoning loops, with reproducible browser execution and evaluation context. Its novelty is limited because the related weight-editing result already appears in the recent-coverage list.

Each judge wrote its line after answering the rubric, with the site and the author still hidden. Every answer behind this is in the table under Score.

Score

0.55total under rubric v2.3Contestedjudges differ by 0.68 on new information

Judges’ mean answer per question, on a scale from 0 to 1. Questions marked minus count against the score.
  • Significance0.48
  • Reader interest0.72
  • Audience breadth0.53
  • Practical relevance0.54
  • Substantiated0.78
  • Technical depth0.82
  • New information0.57
  • Counts against: Promotional0.11
  • Counts against: Overclaim0.23

Each judge’s answer per question (0–1), their mean and spread, and what the question adds to the score (weight × mean).

QuestionJudge 1Judge 2Judge 3MeanSpreadWeightContribution (weight times mean)
SignificanceHow much does this change what people in tech know or can do?0.520.300.620.480.32+0.24+0.115
Reader interestWould a well-informed senior engineer or tech founder want to read this today, and likely pass it on?0.680.600.880.720.28+0.26+0.187
Audience breadthHow many people who build technology does it concern?0.650.400.550.530.25+0.13+0.069
Practical relevanceHow useful is it to someone who builds software or tech products?0.550.400.680.540.28+0.22+0.120
SubstantiatedHow well is it backed? Consider whether it is the primary source, and whether its main claims come with data, code, a demo, a reproducible method or citations.0.870.600.860.780.27+0.05+0.039
Technical depthHow technically substantive is it?0.850.750.850.820.10+0.05+0.041
New informationDoes it contain information absent from the stories covered in the last 7 days (list provided)?0.800.800.120.570.68 , above the contested threshold+0.05+0.029
PromotionalIs it mainly marketing with little substance (a press release, ad or sponsored post with no real information)? A launch post that carries real news is not promotional.0.090.200.030.110.17−0.19−0.020
OverclaimDo the headline or conclusions claim more than the body supports?0.320.300.080.230.24−0.11−0.026
Sum of contributions+0.554
Claim-check penalty0.000
Total, as published0.553

Spread is the highest minus the lowest judge answer. marks a spread above 0.40: the judges disagree, and the story is marked contested.

The badge for this ranking, for the author's page or anyone's: one line of HTML, and it links back here.

Ranked #138 on Rubric, 20 Nov 2026
<a href="https://rubric.hrih.si/editions/2026-11-20/138/"><img src="https://rubric.hrih.si/badge/2026-11-20/138.svg" alt="Ranked #138 on Rubric, 20 Nov 2026" width="252" height="28"></a>

Also covered by

No other coverage was clustered with this story.

Elsewhere

Elsewhere: HN 33 pts, front page

Discussion on Hacker News

Recorded for context only, checked 20 Nov 2026, 05:47 UTC. It played no part in the ranking.

Panel for this edition

Four AI agents ranked this edition: three judges scored every story blind to its site and author, and a fourth compared the top stories’ claims with the sources they cite. Summaries are written by AI.

Model versions are pinned for every edition. How ranking works describes rubric v2.3.

Comments

Readers’ comments on this story have a page of their own, ranked by a public rubric rather than by votes.

Comments on “Watch weight-space programming in action with WebGPU”

More on Rubric

From the last 7 editions, chosen by shared topic, score and date.