Argus Digest: test

Source contribution summary for this digest
Source	Type	Included	Scored	28d Digest Rate	28d Avg Score	28d Hotlist Hit	7d Article Age	28d Confidence
Hacker News	commentary	1	10	10%	0.23	0%	2.4d	Stable
Medium AI (keyword)	commentary	1	10	7%	0.11	0%	0.9h	Stable
Medium Artificial Intelligence (keyword)	commentary	1	10	7%	0.11	0%	0.9h	Stable
MyFT	news	1	10	6%	0.05	0%	4.3h	Stable
Reddit AI Wars	news	1	10	~10%	~0.14	~0%	15.3h	Low sample
arXiv CompSci ML	research	1	10	~7%	~0.09	~0%	4.0h	Low sample
Ars Technical All News	news	1	9	13%	0.14	0%	8.5h	Stable
OpenClaw: discovery-rank	curated	1	9	Collecting data	Collecting data	Collecting data	Unknown	Collecting
The Verge	news	1	8	12%	0.18	0%	8.4h	Stable
The Atlantic	news	1	4	11%	0.13	0%	8.3h	Stable
Wired AI News	news	1	2	~11%	~0.15	~0%	8.0h	Low sample
Atlas Obscura	commentary	1	1	~31%	~0.63	~0%	6.1h	Low sample
MIT Sci, Tech & Society	research	1	1	Collecting data	Collecting data	Collecting data	No recent data	Collecting
The Onion	news	1	1	~21%	~0.50	~0%	7.1h	Low sample
Tom’s Hardware	news	1	1	9%	0.21	0%	8.3h	Stable
Bloomberg Markets	news	0	10	2%	0.02	0%	4.8h	Stable
Bogleheads forum	news	0	10	5%	0.06	0%	1.1h	Stable
Guardian	news	0	10	9%	0.19	0%	10.9h	Stable
NYT front page	news	0	10	7%	0.08	0%	8.1h	Stable
Reddit AntiAI	news	0	10	6%	0.11	0%	8.0h	Stable
WSJ US Business	news	0	10	4%	0.06	0%	9.8h	Stable
Zero Hedge	commentary	0	10	3%	0.02	0%	9.0h	Stable
arXiv CompSci CL	research	0	10	~6%	~0.10	~0%	4.0h	Low sample
TechCrunch	news	0	8	9%	0.10	0%	5.5h	Stable
Seeking Alpha News	commentary	0	7	3%	0.01	0%	1.0h	Stable
India Times	news	0	6	0%	0.01	0%	2.9h	Stable
WSJ Tech	news	0	6	7%	0.08	0%	7.1h	Stable
Reddit R/FuturesTrading	news	0	3	~2%	~0.00	~0%	9.3h	Low sample
SEC Speeches Statements	policy_release	0	3	Collecting data	Collecting data	Collecting data	9.9h	Collecting
FT Alphaville	news	0	2	~9%	~0.11	~0%	4.5h	Low sample
Futurism	news	0	2	13%	0.15	0%	5.0h	Stable
Latent Space	commentary	0	2	Collecting data	Collecting data	Collecting data	3.6h	Collecting
NYT Economy	news	0	2	Collecting data	Collecting data	Collecting data	3.1h	Collecting
WSJ Social Economy	news	0	2	5%	0.01	0%	6.4h	Stable
Better Markets Substack	news	0	1	Collecting data	Collecting data	Collecting data	8.1h	Collecting
Cassandra Unchained by Michael J Bury	commentary	0	1	Collecting data	Collecting data	Collecting data	12.6h	Collecting
Closing Line Substack	news	0	1	Collecting data	Collecting data	Collecting data	9.4h	Collecting
Daring Fireball	commentary	0	1	~17%	~0.15	~0%	7.2h	Low sample
Grumpy Economist (Cochrane)	commentary	0	1	Collecting data	Collecting data	Collecting data	No recent data	Collecting
MIT Research General	research	0	1	Collecting data	Collecting data	Collecting data	4.7h	Collecting
Next Event Horizon Substack	news	0	1	Collecting data	Collecting data	Collecting data	9.0h	Collecting
Polymarket Substack	news	0	1	Collecting data	Collecting data	Collecting data	12.3h	Collecting
Reddit R/Ethereum	news	0	1	~5%	~0.06	~0%	7.5h	Low sample
SEC Press Releases	policy_release	0	1	Collecting data	Collecting data	Collecting data	9.5h	Collecting
Silver Bulletin	commentary	0	1	Collecting data	Collecting data	Collecting data	6.6h	Collecting

New type of dice guarantees no tie when deciding who goes first

Hacker News | positive | Published: 21:50 Sep 01, 2026 (Eastern)

An article from Hacker News titled "New type of dice guarantees no tie when deciding who goes first" was published with a link to a CBC lite story and a discussion thread.

Keywords: dice, board games, invention, whimsical

Prompted up some dinner from my beautiful AI wife; I'm such a good chef.

Reddit AI Wars | positive | Published: 16:14 Sep 03, 2026 (Eastern)

A Reddit post titled "Prompted up some dinner from my beautiful AI wife; I'm such a good chef." was submitted by user needtoventsohereiam to the Reddit AI Wars community, featuring an image and the text "Delicious! So nourishing 😋".

Keywords: AI, Reddit, humor, cooking, eccentric behavior

Durham Town Hall in Durham, England

Atlas Obscura | positive | Published: 16:00 Sep 03, 2026 (Eastern)

Following refurbishment work, Durham Town Hall in Durham, England, has opened to the public, offering audio guides, interactive displays, and short films to explore the building's history since 1350. Exhibits cover topics such as the Prince Bishops, local government evolution, the Mayor and Mayor's Bodyguard, Trade Guilds and Freemen meeting in the Guildhall, and local figures like the 18th-century performer Joseph Boruwlaski. Visitors can tour the Hall, galleries, and mayors chambers.

Keywords: Durham Town Hall, Joseph Boruwlaski, Local History, Trade Guilds, England

You can watch the coolant flow inside Ugreen’s liquid-cooled power bank

The Verge | neutral | Published: 18:00 Sep 03, 2026 (Eastern)

Ugreen has released the MagFlow Pro Magnetic Power Bank, featuring a 10,000mAh battery, CryoPulse technology, and a Cyber Window that allows users to watch the coolant flow inside. Available starting today, the power bank is priced at $149.99.

Keywords: Ugreen, power bank, gadgets, CryoPulse technology, MagFlow Pro

LLM-Guided Reinforcement Learning for Adaptive NPC Behavior in Multi-Agent Combat Games

arXiv CompSci ML | neutral | Published: 00:00 Sep 04, 2026 (Eastern)

Researchers investigate a runtime strategy-selection framework where a large language model guides a trained reinforcement learning policy to adapt non-player character (NPC) behavior in multi-agent combat games without modifying its underlying policy. Testing an LLM-augmented configuration using a locally hosted Mistral 7B model against three scripted opponent types in Unity, the study found that augmented agents more than doubled their win rate against Balanced opponents, achieved higher win rates and faster kills against Evasive opponents, but struggled against Aggressive opponents due to the LLM's nearly constant preference for encirclement regardless of opponent type.

Keywords: artificial intelligence, video games, reinforcement learning, large language models, NPC behavior

Just like a fruit fly, a new algorithm never forgets old scents

Ars Technical All News | neutral | Published: 14:22 Sep 03, 2026 (Eastern)

Researchers have developed a new insect-inspired "sparse coding" algorithm that enables fast learning and prevents catastrophic forgetting, mimicking the memory processes of a fruit fly.

Keywords: algorithm, fruit fly, sparse coding, machine learning, scent

Machines May Develop Senses We Do Not Have

Medium Artificial Intelligence (keyword) | neutral | Published: 02:56 Sep 04, 2026 (Eastern)

The article titled "Machines May Develop Senses We Do Not Have" notes that humans naturally design machines around themselves.

Keywords: machines, senses, artificial intelligence, technology

How the Early Dreams of the Internet Changed

The Atlantic | neutral | Subscription | Published: 16:41 Sep 03, 2026 (Eastern)

This edition of *The Atlantic*’s Time-Travel Thursdays explores how early perceptions of personal computers and the internet grew out of 20th-century American anxieties about conformity and the countercultural ideals of the 1960s. Highlighting Apple's 1984 Macintosh commercial and subsequent tech boosterism, the article examines how digital technology was initially framed as a tool for individual freedom and rebellion against centralized institutional control. While these promises were partially fulfilled by splintering 20th-century mass culture into numerous minicultures, the article notes that modern digital devices still serve corporate interests through tracking and algorithms, even as a transgressive ethos of subversion continues to thrive across the political spectrum.

Keywords: Internet history, Technology, Counterculture, Digital culture

How architects turned a hulking brick box into MIT’s newest academic hub

MIT Sci, Tech & Society | neutral | Published: 00:00 Sep 04, 2026 (Eastern)

Architects from Diller Scofido + Renfro (DS+R) transformed the Metropolitan Storage Warehouse—a fortress-like, 19th-century brick building in Cambridge, Massachusetts—into MIT’s new academic hub for the School of Architecture and Planning (SA+P). Working with MIT, the project team preserved the historic south facade to satisfy the Cambridge Historical Commission while carving large glass-fronted voids into the north side to bring in natural light and create floating, double-height studios. The building also features an interior circulation system extending the concept of MIT's Infinite Corridor, large central stairs, and asymmetric spaces that combine collaborative research areas with smaller offices and seminar rooms. Set for a ceremonial moving-in procession on Sept. 8, the renovated building serves as a living laboratory for adaptive reuse and interdisciplinary education.

Keywords: MIT, architecture, Met Warehouse, design, academic hub

America must learn AI lessons from Astro Boy

MyFT | neutral | Subscription | Published: 00:00 Sep 04, 2026 (Eastern)

Asian cultures harbor a very different attitude towards technologies that are regarded as threatening in the west.

Keywords: Artificial Intelligence, Astro Boy, Technology, Culture

DuplexSpeechBench-IFEval: Evaluating Implicit Instruction Following in Full-Duplex Voice Agents

OpenClaw: discovery-rank | neutral | Published: Unknown

The article introduces DuplexSpeechBench-IFEval (DSB-IFEval), a benchmark designed to evaluate implicit instruction-following in real-time spoken interactions for full-duplex voice agents. Comprising 1,038 test cases across eight assistant roles and five conditioning protocols, DSB-IFEval measures real-time floor management using an Instruction Adherence Score (IAS) and persona-consistent content using an LLM-judged Persona Adherence Score (PAS). Testing across six real-time speech systems reveals architecture-dependent trade-offs, showing that models like F-Actor and PersonaPlex experience drops in adherence under persona-only conditioning, while systems such as GPT-Realtime, MiniCPM-o, and Fun-Audio-Chat maintain persona-consistent content but fail to adapt their floor behavior between explicit and persona-only instructions. Additionally, the findings indicate that systems struggle to override directives when faced with safety conflicts.

Keywords: DuplexSpeechBench-IFEval, voice agents, AI benchmark, conversational AI

MapQuest Jumps To Top Of App Store After Refusing To Rename ‘Lake Ontario’

The Onion | neutral | Published: 15:57 Sep 03, 2026 (Eastern)

MapQuest became the top navigation app in Apple’s App Store after reportedly refusing to comply with a presidential executive order to rename Lake Ontario to Lake America, surpassing competitors such as Google Maps, Waze, and Apple Maps.

Keywords: MapQuest, The Onion, Lake Ontario, App Store, Satire

Lexar announces ‘world’s thinnest portable SSD’ in celebration of the brand’s 30th anniversary – Lexar Muse drive achieves 0.15-inch thickness with proprietary pogo-pin cable, magnetic sleeve for phone mounting

Tom’s Hardware | neutral | Published: 03:00 Sep 04, 2026 (Eastern)

To celebrate its 30th anniversary, Lexar has announced the Muse drive, described as the world's thinnest portable SSD with a thickness of 0.15 inches (3.8 mm) and dimensions of 80x48 mm. The slim drive is designed to magnetically attach to an iPhone via a magnetic sleeve for direct recording, utilizing a proprietary pogo-pin cable to achieve its compact form factor.

Keywords: Lexar, SSD, portable storage, tech anniversary, gadgets

Nobody Is Saying Why OpenAI and Anthropic Had Outages Today

Wired AI News | neutral | Published: 17:56 Sep 03, 2026 (Eastern)

On Thursday morning, frontier AI models from Anthropic, OpenAI, and xAI experienced rare outages affecting their respective chatbots, alongside scattered reports of potential issues with Google Gemini. While simultaneous outages often suggest a shared third-party vendor, OpenAI and Anthropic did not cite an external source, and major internet infrastructure providers reported no outages. OpenAI attributed its downtime to a routing error affecting ChatGPT and Codex, while Anthropic deployed a fix for a partial outage impacting various Claude models. SpaceX stated that xAI's Grok issues stemmed from an outage at its Memphis compute center. Major infrastructure providers like Cloudflare, Amazon Web Services, and Microsoft Azure reported no outages.

Keywords: OpenAI, Anthropic, ChatGPT, Claude, Grok, outages

AI-Written Tests Are Testing the AI, Not Your Code

Medium AI (keyword) | neutral | Published: 02:56 Sep 04, 2026 (Eastern)

Engineering teams adopting AI pair-programming are exhibiting a concerning pattern regarding code testing.

Keywords: AI, software engineering, coding, technology