Daily Tech News - 2026-02-22

From the dizzying heights of zettahash mining to the internal drama of AI labs, today’s tech landscape is defined by a race for scale and a struggle for standardization.

The AI sector continues its rapid, if ironic, expansion. As new labs spin out of one another—each claiming to be the "responsible" alternative to the last—the practical reality of development is shifting toward automation. OpenAI’s Codex is redefining the role of the engineering agent, while industry experts predict that generative AI will finally take the sting out of legacy migrations, a task long loathed by software engineers.

This push for intelligence is having a tangible impact on hardware and high-performance computing. Raspberry Pi’s stock surged 30% this week, fueled by social media buzz that these nimble devices are becoming the go-to power source for localized AI chatbots. Simultaneously, the Bitcoin network has reached a staggering milestone, with a hash rate now hovering around one zettahash per second. This massive computational power dwarfs all other proof-of-work systems, highlighting a specialized infrastructure that continues to auto-adjust its difficulty to maintain a constant pulse of block production.

Even in the realms of pure math and web plumbing, efficiency remains the priority. Whether it’s calculating the 10 millionth Fibonacci number using "certificates of correctness" to speed up verification or managing the gigabytes of bandwidth saved through aggressive feed rate-limiting, the focus is on optimizing resources. Yet, standardization remains a hurdle; the lack of portability between git forge repository configurations serves as a reminder that even our most basic developer tools still suffer from fragmentation.

For those looking to test their own knowledge amidst these rapid changes, the latest Nerd Quiz offers a chance to measure your "inner geek" against the everyday challenges of the modern tech world.

Featured Articles

10-000-000th-fibonacci-number-efb6b201

I calculated the 10 millionth Fibonacci number using code from this post. The certificate is auxiliary data that makes it faster to confirm that the original calculation was correct. Sometimes you can get a certificate of correctness for free. Other times computing the certificate takes a substantial amount of work.

Keywords: seconds verifying, seconds certificates, verify fibonacci, 000th fibonacci, fibonacci number, millionth fibonacci, certificate efficiently, seconds verify, number fibonacci, fibonacci numbers
Source: johndcook.com

bitcoin-mining-difficulty-f390ed21

The previous post looked at the Bitcoin network hash rate, currently around one zettahash per second, i.e. 1021 hashes per second. The difficulty of mining a Bitcoin block adjusts over time to keep the rate of block production relatively constant.

Keywords: mining difficulty, hash rate, mining difficult, rate difficulty, difficulty mining, difficulty ratio, mining bitcoin, bitcoin mining, current difficulty, mining
Source: johndcook.com

exahash-zettahash-yottahash-219a4e72

Bitcoin mining requires computing hash functions to solve proof-of-work problems. Other cryptocurrencies use hash functions for proof- of-work as well, but they contribute a negligible amount of hashes per second compared to Bitcoin. The hashrate varies over time because the difficulty of Bitcoin mining is continually adjusted.

Keywords: hash functions, bitcoin hashrate, uses hash, hashrate varies, hashrate, cryptographic hash, hashrate relatively, number hashes, computing hash, hashes
Source: johndcook.com

forge-specific-repository-folders-6db5a9b6

Every forge that hosts git repositories has added these features through the same trick: a dot- folder in your repo root that the forge reads on push. The folder names differ, the contents overlap in some places and diverge in others. The portability story between them is worse than you’d expect.

Keywords: github workflows, file github, workflows github, configuration github, github actions, files github, github reads, paths github, advisory github, github folder
Source: nesbitt.io

how-ai-labs-proliferate-6f4fc9e8

The irony: “we’re the responsible ones” is each lab’s founding mythology as they spin out of each other. “We can’t trust any of these people with super-intelligence. We need to build it ourselves”

Keywords: ai labs, labs trust, labs, ai, competing ai, labs proliferate, lab, ones lab, super intelligence, lab founding
Source: blog.jim-nielsen.com

how-i-think-about-codex-1d628e89

Codex is OpenAI’s software engineering agent, available through multiple interfaces. Gabriel Chua (Developer Experience Engineer for APAC at OpenAI) provides his take on the confusing terminology behind the term "Codex"

Keywords: trained codex, codex models, codex model, openai codex, codex openai, codex, harness codex, term codex, codex harness, codex refer
Source: simonwillison.net

london-stock-exchange-raspberry-pi-holdings-plc-19348050

Raspberry Pi's stock price has surged 30pc in two days, amid chatter on social media that the company's tiny computers can be used to power a popular AI chatbot. CEO Eben Upton bought stock in the beaten‑down UK computer hardware firm, halting a months‑long slide.

Keywords: pi stock, stock uk, london stock, upton shares, shares raspberry, pi holdings, stock beaten, stock, bought stock, stock purchase
Source: simonwillison.net

nerd-quiz-4-0960ece7

Nerd Quiz is a single page HTML application that challenges you to measure your inner geek. Each question in the quiz comes from everyday moments of reading, writing, thinking, learning and exploring. Visit Nerd Quiz to try the quiz.

Keywords: nerd quiz, quiz community, nerd, try quiz, visit nerd, quiz comes, quiz, instalment nerd, quiz fourth, quiz single
Source: susam.net

the-future-of-migrations-c0f962f9

Migrations are a big part of real-world software engineering. They tend to be painful. Many engineers hate migrations and don't do their best work on them. I conjecture that generative AI will help us with migrations more than is commonly understood.

Keywords: test migrations, migrations longer, future migrations, migrations migrations, make migrations, migrations long, migrations, migration feel, migrations big, migration projects
Source: natemeyvis.com

the-importance-of-limiting-syndication-feed-requests-in-some-way-faedf1d9

Yesterday there were 7492 feed requests that got HTTP 200 responses, and 11941 requests that received HTTP 429 responses. If feed readers didn't do any conditional GETs and I didn't have any rate limiting, the additional feed requests would have amounted to about 3.5 GBytes of responses sent out to people. The importance of limiting syndication feed requests in some way is obvious.

Keywords: limiting syndication, syndication feeds, feed requests, compressed syndication, serve syndication, high syndication, syndication feed, syndication, lot syndication, feed responses
Source: utcc.utoronto.ca