iiinova — where your digital self keeps evolving

Jon Ayre @jonayre.uk · 8h

There are so many options that are better than Palantir. They should never have been in the mix in the first place.

000

Reposted by Jon Ayre

Alejandra Caraballo @esqueer.net · 19h

Solid chance Openai goes belly up and takes softbank and oracle with it. Couldn't happen to more deserving companies.

1342866492

Jon Ayre @jonayre.uk · 21h

Mass production of standard widgets to an established specification using existing tooling Or, small changes to an existing stable system to reflect a change that has already happened in the business using the system So there are cases, but only at the tail end of the engineering lifecycle

jonayre.uk

000

Jon Ayre @jonayre.uk · 01/10/2026

Yeah. That's a revolut issue.

000

Jon Ayre @jonayre.uk · 01/10/2026

A real extract from the terms and conditions for Meta's cuddly Muse and OpenAI's fuzzy Dots: 7.1.2 Don't get them wet 7.1.3 Don't feed them after midnight

011

Jon Ayre @jonayre.uk · 30/09/2026

Why would it do that?

000

Jon Ayre @jonayre.uk · 30/09/2026

Okay. But will it have a fuzzy butt?

000

Jon Ayre @jonayre.uk · 30/09/2026

That's the thing with Software, Grady. Every generation starts again from scratch. I'm looking forward to them rediscovering iterative development (again) in a decade or so. It's almost as if engineering as a broader discipline doesn't exist to them.

020

Reposted by Jon Ayre

Sarah Murphy @13sarahmurphy.bsky.social · 29/09/2026

Burnham made Farage seem “angry, thin-skinned… absurd. What we witnessed today was a masterclass in political communication which makes Farage look like an amateur” Finally. The confidence that comes with emotional intelligence. It could make all the difference. iandunt.substack.com/p/burnham-sp...

iandunt.substack.com

Burnham speech: An astonishing moment of radicalism and commitment

It’s easy to be cynical, but it would be false. Something extraordinary is happening here.

491157247

Jon Ayre @jonayre.uk · 29/09/2026

It is very much bollocks

010

Jon Ayre @jonayre.uk · 29/09/2026

Dear Americans. To be taken seriously in the UK you have to say maths and Lego, not math and Legos. Yes, we're that petty - it makes us happy 😉

010

Jon Ayre @jonayre.uk · 29/09/2026

Man! You just rebooted Catholicism!

020

Reposted by Jon Ayre

A.R. Moxon @juliusgoat.bsky.social · 29/09/2026

To people who don’t believe that Kermit the Frog is real: What is the basis for your morality? Where does your moral standard come from?

963619663

Jon Ayre @jonayre.uk · 29/09/2026

The agentic loop motto: If at first you don't succeed, try try try try try try try try try try try try try try try try try try try try try try try try try try try try try try try try try try try try try try try try try try try try try try try try try try try try try try try try try try try again

010

Jon Ayre @jonayre.uk · 29/09/2026

Exactly this. Long horizon automation using LLMs is a dead end because there is nothing in the loop to make considered decisions. You're basically following a random walk towards a destination and the further away that destination is, the lower the probability is that you'll reach it.

000

Jon Ayre @jonayre.uk · 28/09/2026

Funny how the markings on the cat changed over the two years and the house grew an extra bit of wall to the right of the fire place - and that's just for starters. Blocked.

020

Reposted by Jon Ayre

Dragan Stepanović @dstepanovic.bsky.social · 27/09/2026

If you think the effects of dissolving the team's mental model of the system, by removing humans (authors and reviewers) from the path to production, will show up soon enough to reverse course safely, I have news for you.

162

Jon Ayre @jonayre.uk · 27/09/2026

Absolutely. Now you can hack a whole range of sites with your AI agents and instead of going to jail you just say "oopsy! My bad" and everyone invests in your product.

010

Jon Ayre @jonayre.uk · 27/09/2026

It's hype to boost investment in a product that is failing to deliver even a fraction of the benefits it has to deliver to justify the market inflation. The real existential risk of AI is to the economy and everyone's pensions.

160

Jon Ayre @jonayre.uk · 27/09/2026

There's a question I keep asking and so far no-one's answered it. If AI wrote the test and you didn't review it, how do you know it's testing the right thing? Without an answer to that question, all your software factory hype is exactly that. Hype.

021

Jon Ayre @jonayre.uk · 26/09/2026

No. The answer is no. You are now simply known as "Magazine".

010

Jon Ayre @jonayre.uk · 25/09/2026

Amidst all the noise around AI, it appears the previous thing everyone got excited about is having another go. Meta have announced their new VR goggles. However, there are still those pesky seven hurdles to overcome before VR has a chance of going mainstream

lnkd.in

Meta VR Glasses hands-on: Hand and eye tracking is so much better than a controller - Engadget | Jon Ayre

000

Jon Ayre @jonayre.uk · 24/09/2026

It's easy. They're known as Tommo, Tommy and Thomas. Simples.

010

Jon Ayre @jonayre.uk · 23/09/2026

Just like real life 😉

010

Jon Ayre @jonayre.uk · 23/09/2026

For 8 years we've been told that the UK is lagging behind other countries in productivity. Turns out we never were. The ONS was just doing the maths wrong. 8 years of people investing real effort into fixing a problem that never existed

000

Jon Ayre @jonayre.uk · 22/09/2026

I think just ban all reporting on technology tbh. They're terrible at it.

000

Jon Ayre @jonayre.uk · 22/09/2026

I immediately adopted the obvious tactic of deploying myself in the opponents goal mouth and couldn't understand why no-one passed the ball straight to me. Obvious scoring opportunity! What's an offside rule? Never heard of it.

020

Jon Ayre @jonayre.uk · 21/09/2026

Literally the same picture

000

Jon Ayre @jonayre.uk · 21/09/2026

In fact, everything here is your opinion and I consider that bias and choose to challenge it. The fact that you find that challenge insulting in no way makes the challenge wrong.

000

Jon Ayre @jonayre.uk · 21/09/2026

And then you double down by claiming "there are many in the tech community who think that writing skills are unimportant". There are certainly some, but there's no evidence to suggest there are proportionally more than in other professions. It's not a fact - it's just your opinion.

100

Jon Ayre @jonayre.uk · 21/09/2026

Did you notice that in your original post you didn't include yourself. You could have included "computer scientists" but instead went for mathematicians and engineers, and you included historians in your list of good programmers (which you happen to have studied). This looks a like unconscious bias

100

Jon Ayre @jonayre.uk · 21/09/2026

I didn't call you a bigot. I did suggest you were applying a bias to your conclusion (as all humans have a tendency to do), and as your statement was a published one I also asserted that it should come with evidence. You find that offensive. I believe it's a fair challenge.

100

Jon Ayre @jonayre.uk · 20/09/2026

Any statements that publicly disrespect an entire demographic should be backed up by concrete evidence or be recognised as the prejudice that they are. There are no "okay" groups to stereotype in a negatively prejudiced way.

120

Jon Ayre @jonayre.uk · 20/09/2026

I'm not partial to any posts that stereotype people tbh. I've seen great programming and bad programming from people from all backgrounds. I've also seen excellent and poor language skills from people from all backgrounds.

111

Jon Ayre @jonayre.uk · 20/09/2026

You don't need a watermark to know this is GenAI. Just look at the bottom of the box and how it aligns with the edges of the shelf. That's not how right angles work.

010

Jon Ayre @jonayre.uk · 20/09/2026

Nowt, unless you consider overly decorated presentations, clichéd writing that contains little information, and code that no-one wants or needs to be valuable.

110

Jon Ayre @jonayre.uk · 20/09/2026

Calculate what? Final amount paid? Monthly payment? Remaining balance? Etc etc etc.... Also, mortgages don't calculate, calculators do, so the first rather than the second as a starting point.

100

Jon Ayre @jonayre.uk · 20/09/2026

Essentially, they need to become the publishing platforms they've worked so hard to avoid being. Their business model will be their ultimate undoing.

010

Jon Ayre @jonayre.uk · 20/09/2026

Fair point, but some of us use "it's just a tool", in a derogatory way, as part of the narrative you give here. It's not mysterious, it's not intelligent and it's certainly not going to produce anything useful on its own. It's just a tool - so how can it ever be worth all that wasted energy & money?

111

Jon Ayre @jonayre.uk · 20/09/2026

Truth is "they" never knew software engineering at all.

000

Jon Ayre @jonayre.uk · 17/09/2026

Where there is no intelligence at work entropy increases. Where there is consistent decrease in entropy, intelligence is present. Left to their own devices, LLMs increase entropy. When a human is involved, entropy can decrease.

120

Jon Ayre @jonayre.uk · 17/09/2026

I'm going to write a blog post about an idea I've had, but right now I'm on holiday so it'll have to wait. The concept is this: intelligence is difficult to define but easy to detect. Intelligence reverses entropy.

110

Jon Ayre @jonayre.uk · 15/09/2026

So half a second every second... like Spain? (Which is a good thing by the way, and not a slight in any way on Spain)

010

Jon Ayre @jonayre.uk · 13/09/2026

I love the Asimov robot stories. TBH I pretty much grew up on them, but anyone who thinks the three laws are in any way implementable needs to try coming up with an unambiguous definition for "harm". Even Asimov made this point in his later stories.

000

Reposted by Jon Ayre

Jason Gorman @jasongorman.bsky.social · 12/09/2026

The theory's almost complete. Set your bar for shippable code to even just one-9's reliability, and the human has to stay very much in the loop. Set it to two 9's, and the human *is* the loop. codemanship.wordpress.com/2026/09/12/o...

codemanship.wordpress.com

Out-Of-Distribution Checks – How Human Intelligence Stabilises Agentic Workflows

Back to my “Special Theory of Autonomous Agent Reliability” (STAAR). The reliability of a step (a model interaction) in an agentic workflow: R = 1 – (1 – C)(1 – P) Whe…

0165

Jon Ayre @jonayre.uk · 13/09/2026

"AI went rogue". Nope. Hiding behind the machine was ways going to be the strategy:

jonayre.uk

010