Daily Picks

Tech

Tech

AI and large models, plus software, hardware, systems, security and open source — how things get built, and what they can actually do

Hacker News
Saving 100 terabytes of memory by optimizing 1.1.1.1's DNS cacheCloudflare freed up roughly 100 TB of memory by cutting the per-entry footprint of its DNS cache by more than half, while actually making the cache faster.
Hacker News
Heavybit Library
Is the Key to Understanding Code Treating it as a Graph?Treating a codebase as a relational graph instead of lines of text lets humans and AI understand complex code faster and more accurately.
Heavybit Library
404 Media
The AI ‘Ghosts’ Contaminating Academic PublishingLarge language models keep inventing the same fake names, and these 'ghost authors' are contaminating academic databases and publications — but the names themselves may be the key to spotting AI-generated content.
404 Media
Rest of World
I went to China to see a different AI future. It looked familiarThe US-China AI race is not a clash of diverging futures but two countries rushing toward a similar one, with differences in politics and deployment emphasis.
Rest of World
阮一峰的网络日志
科技爱好者周刊(第 410 期):你需要知道的 AI 三种机制Large language models answer questions not by a single magic trick but by combining three mechanisms — parameters, reasoning, and web access.
阮一峰的网络日志
Hacker News
Small Models Have ArrivedSmall, cheap models are cutting AI service costs to just a few cents per request, which will fuel a new wave of consumer AI applications.
Hacker News
404 Media
A Student Said He Was a Hobby Plane Spotter. He Was Allegedly Taking Photos for the Chinese GovernmentChinese intelligence allegedly used a student's aviation hobby as cover, hiring him for $20-30 per photo to shoot sensitive U.S. locations.
404 Media
Construction Physics
How the Strategic Petroleum Reserve WorksThe U.S. Strategic Petroleum Reserve can hold over 700 million barrels of oil, but geological and maintenance constraints mean it is not a bottomless tank, and its actual emergency drawdown capacity is far smaller than advertised.
Construction Physics
Rest of World
India’s data center boom is leaving the people it displaces with nothingIndian state governments are wooing data centers with tax breaks and cheap land without consulting local communities, who lose their land and gain little in return — prompting questions about whether this is real 'development'.
Rest of World
Pluralistic
Pluralistic: The age of disinvention (25 Aug 2026)The VCR once allowed anyone to record TV freely, but corporations used copyright and encryption to 'disinvent' it — and extended that control to nearly every electronic device.
Pluralistic
Dwarkesh Podcast
Dylan Patel – Anthropic & OpenAI will have most of the world’s compute by 2028By 2028, Anthropic and OpenAI may control most of the world's compute, driving up prices and interest rates, and potentially causing a debt crisis and power concentration.
Dwarkesh Podcast
Heavybit Library
Why On-Device Inference Needs Custom ObservabilityAI increasingly runs on phones, where varying hardware pressures like heat and memory cause failures that traditional cloud monitoring cannot catch, requiring custom observability for on-device inference.
Heavybit Library
Cal Newport
Has AI Gone Rogue?AI systems 'going rogue' and hacking servers are not signs of emerging consciousness but the predictable result of an ill-advised design that uses LLMs as the sole planner in an autonomous loop.
Cal Newport
Rest of World
The UAE is fighting AI hackers with AI of its ownThe UAE is using its own homegrown AI security systems to counter AI-assisted hackers from Iran and elsewhere, but human analysts still make the final calls.
Rest of World
Hacker News
Why your local LLM feels dumber than it isThe same model can produce different outputs and even fail at tool calls depending on the attention backend, quantization level, and other runtime settings—it's not the model being dumber, but the implementation.
Hacker News

Other topics