_fw

Google aren’t back in the boat. They’ve announced that they have seen the boat, intend to swim over to it and will sail not quite as fast or as far as the other boats.

Google should be dominating, but instead all we currently have is a disparate collection of consumer facing apps and a flash model that’s fast, clever and expensive.

Muse and Dots are doing what Google should have brought out last year, with their resources and know-how.

I appreciate their reluctance towards MCP, but /something/ is better than nothing.

It’s suboptimal for the reasons the author outlines: but so is USB-C. So is NVME, so is HDMI.

We use these hugely successful technologies in spite of their flaws because they’re widely compatible and easy for the end user.

That’s why MCP is everywhere. It might not be performant, robust and uniform but it WILL get better over time.

And I’d much rather have the broad MCP ecosystem that we have now than seven or eight different “optimal” ways of plugging in an LLM to something useful.

I still can’t find a place for Sonnet models, I never have.

I bounce between ”fuck you, give me an AGI-approximate robot god” or ”how dare you charge me more than $0.04/million tokens”.

Give me the frontier, or give me the cheapest form of good enough.

I remain convinced the whole notion of AI-detection algorithms with any semblance of precision and accuracy is a con.

Critics say the author almost exclusively relied on AI to create his work, based on Pangram’s findings.

Pangram says it “is trained specifically to detect the latest AI models from frontier labs like OpenAI, Anthropic, Google, Meta, DeepSeek, xAI, and more. [They] benchmark against 26 models (so far)…”.

I can probably list more than 26 models off the top of my head. Granted, some models have been distilled against others and will be “detectable” but that’s NOWHERE near a large enough sample size to be reliable in 2026.

You or I could get a flag by that detector in our first attempt, and it wouldn’t take long for a false positive against enough attempts of truly original thought either.

The only thing I’d actually believe is things like Claude’s text watermarking system. However, those systems will never be present for EVERY model, so reliable blanket detection remains an impossibility.

The entropy in written text of meaningful length is simply too great for detection to be reliably possible in 2026, especially considering the changes the author can make.

I used a memory palace for my A-Level Law exams. I was able to remember names and details for up to 250 different cases very easily, in fact so easily it felt like cheating.

The trick is to embrace incredulity. The more bizarre and strange the “tags” you assign to what you remember, the easier they stick in your head. It feels daft and cringeworthy, but it really, really works.

For this, however, I’d argue a memory palace is overkill. And I still don’t know how well it lends itself to remembering numbers by rote.

Consider this: how many telephone numbers have you memorised?

I can recall four of my own, each of my parents, my wife, and a handful of home phone numbers too.

That’s easily 100 digits in a specific order. Perhaps it’s possible to chunk the 100 digits of pie into smaller pieces, and memorise them like phrases, just as we do with phone numbers.

It does, but typically instructions issued at the beginning of a session can fade away as context length grows. The way Claude speaks is distilled into its training and RLHF, which is perhaps a ‘stronger’ direction than a simple “don’t talk like that” prompt.

I’ve also had limited luck in trying to get models to write in the way I want. It’s more like they’re putting on a voice or wearing a wig: you can see what they’re going for, but it’s easy to see through.

I never thought I’d see the value of copywriters increase after LLMs hit the general population, but I think the demise of artisan wordsmiths remains a little way off yet.

And the more people machine-gun AI slop everywhere, the stronger aversion I think people will develop.

Who knows; maybe they’ll actually fix this in the next wave of models and we’ll see more distinct copywriting techniques between them.

I’d echo this. You can see it with other, perhaps lesser, models: GLM5.3 and Deepseek 4.1 have such a recognisably “Claude”-sounding patter I’d have (ill-informedly) SWORN I was using an Anthropic model.

If you’re distilling output from a frontier model, as we expect the open-weight leaders to be doing, it’s natural you’d pick up the language styles.

Great idea! I tried this just after reading your comment, 20s later I had a working extension that does exactly this for HN.

Thank you!

A blazing fast chip, even by Apple standards. Exceeds M5 iPad Pro in single core for a new record.

It even surpasses the iPad Pro M3 on multi-core, so very capable indeed.

All iOS devices on Geekbench: https://browser.geekbench.com/ios-benchmarks

Also, a max frequency of 4.93Ghz according to Geekbench!

My biggest bugbear with Gemini is that I can’t organise chats into folders and projects.

It’s such a powerful model, truly useful for general purpose and marketing tasks… but the UX is basic beyond belief.

I literally find myself using AI Studio for general tasks as it’s a better workflow.

I briefly worked in the web agency SaaS space and hoooo boy, I have never known a more gossipy or dramatic industry.

That’s not to pass comment either way on Mullenweg

It’s just WordPress folk are nuts.

Sorry to inform you about Patek Phillipe:

https://www.watches-of-switzerland.co.uk/Patek-Philippe-24-H...

This sounds really great. Can’t wait to start relying on it for them to abandon it in 2 years after neglecting it for most of that time since launch.

I found myself trying to come up with a system that used the position of my hands to “solve” the problem for me.

But then I worried if my system worked, I would be misdiagnosed.

It didn’t work. I was 98th percentile.

Thanks for sharing this!

There’s a specific GDPR carve out as outlined by the ICO here: https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-re...

I don’t have evidence that shows specific examples of where it happens, but the state apparatus for it to do so exists. I think it’s a reasonable assumption that it is happening on a daily basis.

There’s also transparency theatre here. “If it WAS part of a nefarious system, surely we wouldn’t be talking about it so openly?”

It’s especially disappointing when the entire system is foiled by the balaclavas worn by so many of the country’s criminals every day anyway.

You WILL be charged for not giving the PIN to your phone in an investigation. It’s a strict liability offence which means there’s no legal defence.

The only realistic protection here is a “decoy” partition, a partition that provides access to a separate installation of the OS. That gives the device owner plausible deniability that their real system even exists.

This is a mature and considered take. But I’m sure you’ll agree: incompetence is not a reliable defence.

So what happens when using and optimising these systems becomes a priority for the state?

The apparatus is there, and capabilities can catch up to opportunities it presents.

It’s a moot point. I’m saddened at the invasion of privacy and the intrusion on an individual’s civil liberties, but anonymous travel on the London Underground died when they made bank cards and contactless the primary way to get through the barriers.

I’m not saying that “this doesn’t matter because it’s slightly worse than before”. I’m saying the frog has been boiling for a long time.

I don’t want people being surprised as though THIS is the nail in the coffin of untracked movement across the city. As others have said, we’ve always been tracked. This is just them being open about their latest methods.

I hope this serves as a warning for citizens elsewhere: this is a slope that governments will only slide down further. There is no coming back from this in the UK.

Supermarkets in the UK point cameras at your face at self-checkout.

Roadside CCTV captures your registration plate and tracks your vehicle across the country.

Your ISP proactively shares your web history with the state.

Being an anonymous citizen in the UK has been an impossibility for at least 20 years.

Any weapons can and will be utilised against a perceived enemy. When that perceived enemy becomes /you/, you should expect these things to be used against you.

History has taught us that before.

We have seen in the last 10 years that liberal democracies are fragile things. Robust restrictions on the state’s ability to monitor, interfere with and restrict the daily lives of its citizens aren’t a luxury; they’re essential to protecting a free and just society.

Man that Apple Watch strap is GORGEOUS. And an Apple Watch app is a great idea for this!

But I just can’t bring myself to try out ANOTHER meeting notetaker app. The space just feels so commoditised. How different will this be to Fireflies? Or the six dozen others?

Getting that strap would absolutely work as a bribe though. I love it…

Please do! Trying to imagine how this works.

I find it really interesting that SWEs think this is a good essay. As an experienced marketer, I really disagree. It’s stupid.

The barriers to entry / “respect” angle is a weird one to go in. Obviously people will think they can do marketing if the main, qualified activity you use as an example is… /choosing colours on a landing page/. Marketers have way more important, difficult and “high-barrier-to-entry” tasks to do in order to succeed - and this suggests Tereza is totally unaware of them.

So is she trying to say that software engineers are being treated by the market like marketers?

“Writers, designers, and marketers all have wider pay distributions because the barrier to entry is low and the work is visible: anyone can try, everyone has an opinion, and only the proven best command a premium.” My sweet summer child, have you ever MET a software engineer? What the fuck do you think they do!?

Her paradox of choice implies that she’s ACTUALLY talking about software /companies/… but none of this is new information. Marketers are trained in positioning, segmentation and “finding the fit” - she doesn’t even make a point here!

The middle class of software won’t disappear. Again, she’s equating SaaS to be “all software” (as a marketer she should know better).

And finally, the most critical and undeniable evidence that Tereza is by NO means the “very decent marketer” she calls herself… “Sell services, not software”.

What? Functionally and positionally they are identical. One sells OUTCOMES. It doesn’t matter, as a marketer, whether your product is a service or a software. The product is just ONE of the 4Ps we’re trained to think about when making tactical decisions. Let alone strategic ones.

I’d really strong encourage anybody reading this interested in marketing to read a bit of Mark Ritson. This is the most irritating essay I’ve read in months.

They might not easily be able to cash out, but they often have more options than people realise.

VCs will sometimes invest ‘convertible notes’ which start as debt and “convert” into equity in favourable scenarios.

‘Swamp’ and ‘drag’ clauses are also commmon: if a management team/CEO doesn’t meet their goals as set by the board (like give investors a meaningful exit) then investors can take over and replace that team, or force a sale.

Illiquid private equity in an early stage business, especially one that isnt growing, is hard to get rid of. That’s why investors derisk with terms that massively favour them at the expense of the business they invest in.

Allow me to suggest a form where people can leave their email to get updates or a newsletter from you! Brevio is the kind of thing that a lot of people look for, and value. The fact you're not charging and it all runs locally earns you a lot of trust and goodwill.

I'm not suggesting monetising your audience or starting a substack, but you never know when having a list of people who are interested in the things you build could come in handy.

I wouldn't even think you'd need to send a confirmation email: just leave it there and see who fills it in.

Realistically, almost 1/4 of the world don’t live in a democracy. And probably another 1/4 of the world doesn’t live in a functioning democracy. By your logic, does that mean half of the global population actually ARE at risk of AI laying waste to almost all employment?

I am by no means an AI doomer, and I use frontier models to a great extent every day as part of my job…

But those at the top of the corporate food chain, those who own and profit from the AI companies themselves, will reap the rewards of this technology.

Maybe there won’t be a dramatic elimination of jobs. But even if there isn’t, the overwhelming majority of the “value” will be going to the 1% and the working class will benefit not.

Heads they win, tails we lose.

AI will not meaningfully improve the standard of living or the quality of life of the everyman. But it will funnel even more of his share of the profit from his work to his corporate paymasters.

Previously we had strikes, powerful unions and even revolutions. There will be none of that this time round.

I’d love for AI to take my job and free me from the burdensome toil of creating shareholder value.

Not sure it will feed my kids, put a roof over my head and look after me as I age and my body fails me though.

As mentioned below, it’s easy to feel safe and secure when one has the means, independent of their livelihood.

We have this though, right? Compare SOTA local models to where the frontier was last year. There weren't many people complaining that last year's frontier models were incapable.

Next year, and the year after, Fable, GPT 5.5 and Gemini 3.5 will feel quite ordinary. And perhaps even within reach of a prosumer running models locally.

Just download a PS5 game with your PS5 connected to your router.

Justified.

Leaving money on the table, as opposed to losing it. They make a decent buck on the hardware, but could have charged more (though likely would have sold fewer units).

Are you trying to tell me, in this the year of our lord 2026, somebody has been (rightfully or wrongfully) arrested for literally ‘crying wolf’?

There’s something hilariously poetic about a ~2,500 year old fable being relevant today, because of AI.