iiinova — where your digital self keeps evolving

Reposted by Ryan Gallagher

Duncan Lamont @duncanlamont2.bsky.social · 20/05/2026

A much more micro example but the change to allow tapping-on multiple London busses and it count as a single journey was a wonderful evolution In Edinburgh I am charged for multiple journeys which, if there’s 2 of you, often means a taxi is cheaper. That doesn’t feel right

5664

Reposted by Ryan Gallagher

Ciaran Martin @ciaranm.bsky.social · 14/04/2026

This paper represents a small but deeply impressive and genuinely important achievement by the much maligned British state in what is probably the most important global issue of our era. Hear me out ( 🧵) 1/ www.aisi.gov.uk/blog/our-eva...

aisi.gov.uk

Our evaluation of Claude Mythos Preview’s cyber capabilities | AISI Work

We conducted cyber evaluations of Anthropic’s Claude Mythos Preview and found continued improvement in capture-the-flag (CTF) challenges and significant improvement on multi-step cyber-attack simulati...

4228135

Reposted by Ryan Gallagher

Sarah O'Connor @sarahoconnorft.ft.com · 26/03/2026

In spite of all the talk of Claude Code and Codex meaning the end of humans writing code, software job adverts are actually going up, according to @jburnmurdoch.ft.com's crunching of millions of job ads for this week's The AI Shift www.ft.com/content/7325...

20565174

Reposted by Ryan Gallagher

Ed Zitron @edzitron.com · 17/03/2026

hearing microsoft is reorganizing its AI team under the banner of "the Copilot System." Also hearing that teams are under pressure to *reduce* AI token use, remit is that there needs to be "fiscal responsibility in AI ops" and that Claude Code usage is being reduced in favour of Copilot CLI.

31867108

Reposted by Ryan Gallagher

Ben Collins @bencollins.bsky.social · 19/02/2026

look at my country man, we’re getting out-No Kings’d by the fuckin’ british

446246905092

Reposted by Ryan Gallagher

Jack Blackburn @hackblackburn.bsky.social · 30/01/2026

This one is surely a still from an ITV drama called “The Rig” about oil workers solving a murder in the middle of the North Sea.

85910

Reposted by Ryan Gallagher

Dan Snow @thehistoryguy.bsky.social · 17/01/2026

Britain is one of the best places in recorded history to be born, live, work and grow old. Lots of room for improvement obviously but combat the pessimism, it’s the forerunner of fascism.

12704135

Reposted by Ryan Gallagher

Sarah O'Connor @sarahoconnorft.ft.com · 30/10/2025

2nd edition of our The AI Shift newsletter is out. The question today: could AI be making us LESS productive? @jburnmurdoch.ft.com www.ft.com/content/2480... At the individual level, it's clear we're not reliable witnesses on this Q. At an organisational level, it gets even more interesting...

511145

Reposted by Ryan Gallagher

BeijingPalmer @beijingpalmer.bsky.social · 29/09/2025

This is my regular reminder to everyone that jstor is open to the general public now; a free account there will give you access to 100 papers a year.

8333791201

Reposted by Ryan Gallagher

Katie Martin @katie0martin.ft.com · 05/06/2025

Lolololol

27614

Reposted by Ryan Gallagher

Alan Beattie @alanbeattie.bsky.social · 28/05/2025

RESOLVED: no one to be quoted by large-circulation media outlets about overall employment levels until they have convincingly demonstrated to a panel of economists that they understand the lump of labour fallacy.

77419

Reposted by Ryan Gallagher

Chris Giles @chrisgiles.ft.com · 21/04/2025

It’s time to retire the phrase, “when America sneezes, the rest of the world catches a cold” on.ft.com/3RoUozD The US, not others, will feel most pain from its economic mistakes

on.ft.com

The US, not others, will feel most pain from its economic mistakes

[FREE TO READ] America is not the powerhouse it once was

611532

Reposted by Ryan Gallagher

Rachel Coldicutt @rachelcoldicutt.bsky.social · 12/04/2025

Quite wonderful find by this FT team. A Chinese small business that makes MAGA merch on.ft.com/4cAH6tE

While many Chinese businesses are reeling from the trade war, Qi, the Maga merchandise maker, brushed off any suggestion his might suffer.
Trump supporters, he said, were willing to pay any price for items bearing the image of their beloved president - and US suppliers were making such a huge profit on them that they could afford to partially absorb the tariff impact.
A Trump baseball cap, for instance, cost only Rmb7.50 ($1) to produce. Tariffs might raise that cost to Rmbo, but the caps were being sold for $50 in the US.
"American sellers could even use the tariffs as an excuse to raise the price to $60 - yet the extra cost will still be borne by the US consumers,
" said Qi.

16734354

Reposted by Ryan Gallagher

JW Mason @jwmason.bsky.social · 11/04/2025

Funny juxtaposition of headlines in this morning's Financial Times.

15731152

Reposted by Ryan Gallagher

Rolling Stone @rollingstone.com · 10/04/2025

According to a source, when Musk privately messages associates and confidants about reports from federal staffers about how their lives have been wrecked, the Tesla CEO is known to react with laugh-crying emojis. Story: www.rollingstone.com/politics/pol...

2502550981

Reposted by Ryan Gallagher

Robin Wigglesworth @robinwigglesworth.ft.com · 05/04/2025

Goldman Sachs is warning that that America’s famous “exorbitant privilege” might be killed by “negative trends in US governance and institutions”. 👀https://on.ft.com/42cMYnT

814146

Reposted by Ryan Gallagher

Adam Payne @adampayne26.bsky.social · 01/04/2025

Intriguing @zoecrowther.bsky.social spot: Government has quietly joined Reddit, where it's posting in subreddits inc Royal Navy & cars It's part of No10's bid — driven by the New Media Unit — to step up comms online, where it feels voters increasingly get news www.politicshome.com/news/article...

34916

Reposted by Ryan Gallagher

Taras Grescoe 🚇 @taras-grescoe.com · 27/02/2025

"TfL’s analysis shows that within a km of an Elizabeth line station in London, the number of new houses is 8-14% higher, and nearly 400,000 jobs have been created since 2015." A smashing success: it's carried half a billion passengers since opening in 2022. www.theguardian.com/business/202...

521643

Reposted by Ryan Gallagher

JW Mason @jwmason.bsky.social · 05/02/2025

The most obvious parallel to me seems like the end of the USSR, the classic "revolution from above." That's the clearest modern example of a small group seizing control of a state not in order to exercise power through it, but to dismantle it.

1044585

Reposted by Ryan Gallagher

Mayor of London, Sadiq Khan @london.gov.uk · 11/01/2025

London has been named the world’s No. 1 destination for 2025 🎡🇬🇧 From iconic cultural landmarks to a food scene that’s second to none, there’s no place like London. Proud to see our city shine on the global stage—come experience it for yourself. www.timeout.com/news/the-wor...

timeout.com

The world’s top destinations for 2025 have been named by TripAdvisor

Can you guess which capital claims the crown?

1735752

Reposted by Ryan Gallagher

Dr Charles Smith @smith-brocca.bsky.social · 09/01/2025

My letter in today's Guardian. Just asking, like. But I think we know the answer.

363170643079

Reposted by Ryan Gallagher

Ed Morrish @edmorrish.com · 06/01/2025

tech bros: we built a machine that can tell you stuff loudly and confidently! us: cool. is it right about stuff? tech bros: we cannot stress enough how loud and confident it is!

664142729

Reposted by Ryan Gallagher

Scott Hanselman 🌮 @scott.hanselman.com · 06/01/2025

Seriously this guy is destroying his brain with drugs

10662

Reposted by Ryan Gallagher

Rosamundi @rosamundi.bsky.social · 06/01/2025

And you might make several people's days better. I wrote an email to TfL complementing one of their bus drivers, and the customer service person was genuinely pleased to receive a positive message and to be able to pass on positive feedback to the bus driver's manager.

2523

Reposted by Ryan Gallagher

Stephen Bush @stephenkb.bsky.social · 06/01/2025

A small life hack that improves other people's lives but also often your own is to write really short thank you notes to people's managers: maybe one time out of 20 you get a freebie and hopefully more than that you help the person out.

1920537

Ryan Gallagher @zero-g-ddr.bsky.social · 06/01/2025

Elon Ketamusk wants the US to invade Britain.

Narcissist billionaire Elon Ketamusk holds a poll on Twitter with the question: "America should liberate the people of Britain from their tyrannical government" with the options Yes and No.

030

Reposted by Ryan Gallagher

JWexTheSpa @jwsidders.bsky.social · 06/01/2025

This is a superb piece of writing and analysis by @stephenkb.bsky.social in which he takes apart Tory opportunism and pandering to Elon Musk over the grooming gangs and compares it to Farage’s consistent red line on Tommy Robinson. Compulsory reading.

ep.ft.com

Good morning. I hope you were able to have a good restorative break. My thanks to Jim, Robert, Miranda, Lucy, Jen, Jude, Chris and Georgina for minding the fort while I was on holiday.

10230127

Ryan Gallagher @zero-g-ddr.bsky.social · 03/01/2025

I'm sure nothing will go wrong in any way.

130

Ryan Gallagher @zero-g-ddr.bsky.social · 02/01/2025

LinkedIn humblebrags are just too much. I shouldn't be taking three-quarters of the way through to find that I'm not reading a parody post.

000

Ryan Gallagher @zero-g-ddr.bsky.social · 01/01/2025

Being a millennial in 2025 means incorrectly signing and dating written documents with dd/mm/200x for the 6th year running.

000

Ryan Gallagher @zero-g-ddr.bsky.social · 01/01/2025

Claude Sonnet 3.5 pattern match fail. When asking it: "how did Alonzo Church's definition of algorithm differ from Godel's recursive functions?" Claude astonishingly claimed that recursive functions align with imperative programming. I often find that AI models perform poorly with comparison Qs.

A screenshot of a conversation with Claude AI Sonnet 3.5 model. The output from Claude states: "The impact of these different definitions continues to influence computer science today. Church's lambda calculus became a foundation for functional programming languages, while Gödel's recursive functions more closely align with imperative programming paradigms.
Would you like me to elaborate on any particular aspect of these approaches?"

040

Ryan Gallagher @zero-g-ddr.bsky.social · 30/12/2024

Addy Osmani's article on AI is full of insight on the trade-offs involved - especially the chasm between how senior devs and juniors approach AI: "Juniors.. accept the AI's output more readily, leading to what I call "house of cards code" – it looks complete but collapses under real-world pressure."

an extract from Osami's article 'The 70% problem: Hard truths about AI-assisted coding'. "When you watch a senior engineer work with AI tools like Cursor or Copilot, it looks like magic. They can scaffold entire features in minutes, complete with tests and documentation. But watch carefully, and you'll notice something crucial: They're not just accepting what the AI suggests. They're constantly:
Refactoring the generated code into smaller, focused modules.
Adding edge case handling the AI missed.
Strengthening type definitions and interfaces.
Questioning architectural decisions, and
Adding comprehensive error handling.
In other words, they're applying years of hard-won engineering wisdom to shape and constrain the AI's output. The AI is accelerating their implementation, but their expertise is what keeps the code maintainable. Junior engineers often miss these crucial steps. They accept the AI's output more readily, leading to what I call "house of cards code" – it looks complete but collapses under real-world pressure."

020

Reposted by Ryan Gallagher

Ryan Gallagher @zero-g-ddr.bsky.social · 30/12/2024

I agree with this. AI is great at solving problems that have already been solved (pattern matching against solutions that are a part of its dataset), but less so at novel problems that are specific to your requirements and codebase. AI is also not the best at creating maintainable abstractions.

041

Ryan Gallagher @zero-g-ddr.bsky.social · 29/12/2024

I often use large language models for learning, as an efficient explainer of new concepts. But AI seems to suffer more hallucinations than usual when Haskell is the topic. For example Gemini confused syntax from Visual Basic for Applications with do notation inside a Maybe monad.

Image is a screenshot from a conversation with Google Gemini, a large language model. The response states: "You're absolutely right to call that out! It was indeed bizarre of me to introduce "IIf" in that context.

While IIf is a perfectly valid concept (as we established), it's more prevalent in VBA and database languages. Most general-purpose programming languages do prefer the ternary operator (condition ? true_value : false_value) for similar functionality."

030

Reposted by Ryan Gallagher

Ryan Gallagher @zero-g-ddr.bsky.social · 21/12/2024

The problem with the idea of non-tech persons prompting AI devs to build software, is that the non-tech person needs deep understanding of software architecture, tooling and abstractions, which they are not likely to hold. Gen AI doesn’t care if its code breaks your other working code.

011

Reposted by Ryan Gallagher

Ryan Gallagher @zero-g-ddr.bsky.social · 25/12/2024

This is the thing that disturbs me the most about generative AI, is where it gives inaccurate information while using an assertive and highly confident tone. It’s a technique that con artists do their best to master.

021

Ryan Gallagher @zero-g-ddr.bsky.social · 02/12/2024

My Xcode appears to be suffering an existential crisis right now.

Xcode (a programming language integrated development environment), is generating the following output on compilation: Can't find or decode reasons. Failed to get or decode unavailable reasons. Can't find or decode reasons. Failed to get or decode unavailable reasons. Can't find or decode reasons. Failed to get or decode unavailable reasons. Can't find or decode reasons. Failed to get or decode unavailable reasons. Can't find or decode reasons. Failed to get or decode unavailable reasons.

080

Reposted by Ryan Gallagher

Stephen Bush @stephenkb.bsky.social · 29/11/2024

A terrific theory that pairs well with the Economist's terrific and terrifying chart of Musk's tweeting.

A chart showing Elon Musk's near-24 hour tweeting.

552019630