My latest newsletter is about how generative AI companies are risking the lives of the most vulnerable by releasing tools that affirm dangerous delusions -- and the potential human consequences. And it's about why nobody seems to be freaking out about it. whatwelost.substack.com/p/how-many-p...

Anyway, that's a taste of my latest (paid only, sorry) newsletter. Want to read the whole thing? Give me your money. whatwelost.substack.com/p/the-death-...

The Internet had a glorious age of culture. You need to realize what's been stolen from you. If we're to take it back, we're going to have to build a new web -- one free of the dominance of platforms that don't care about culture or creativity, or users. whatwelost.substack.com/p/the-death-...

I also need you to understand that this wasn't a mistake or an accident, but a consequence of active design choices. By acting as curators, or stewards, platforms have made it impossible to build the communities required for culture to propagate. whatwelost.substack.com/p/the-death-...

The end result is an Internet of nothing. The web is now homogenous, curated, and boring. It's a British high street, with empty storefronts, vape shops, Oxfam stores, and fuck-all else. whatwelost.substack.com/p/the-death-...

I want to make it clear that there is such a thing as Internet culture, and that it existed, and that it mattered. It was cool, and it existed primarily because the web wasn't centralized. We lost something important and cool and interesting. whatwelost.substack.com/p/the-death-...

The algorithmicization of the web hasn't just made it harder for culture to propagate organically, but has also changed the incentivies behind why and how people create culture. And that's without even mentioning generative AI (which I do). whatwelost.substack.com/p/the-death-...

So much of my writing has focused on the influence of platforms on human interactions. This post goes a step further and talks about what that means for online culture -- or, more specifically, how platforms killed Internet culture. whatwelost.substack.com/p/the-death-...

My latest newsletter is about how Internet culture died, and why the centralization of the web around a handful of large platforms is responsible for making the web a duller, more boring place. whatwelost.substack.com/p/the-death-...

Hey all, quick question: is it good that mark zuckerberg’s answer to “are we in a bubble” was “yeah, maybe, but don’t worry, it’s just like the dot com boom,” a point easily-disproved? Isn’t he meant to be the guy who has a real answer? Is this good? Hello?! www.wheresyoured.at/how-to-argue...

Y'all. @adambecker.bsky.social reviewed the new Eliezer Yudkowsky book for the Atlantic and he *does not miss.* www.theatlantic.com/books/archiv...

My latest newsletter is about how Internet culture died, and why the centralization of the web around a handful of large platforms is responsible for making the web a duller, more boring place. whatwelost.substack.com/p/the-death-...

Jeremy Kyle MP. Has quite the ring to it, doesn’t it? www.telegraph.co.uk/news/2025/09...

Maddest update to this: Facebook is apparently experimenting with deleting features, and will consider bringing them back if people will make videos talking about why said feature is important to them. www.stuff.co.nz/society/3608...

There are some other stories I didn't mention, if only because they might get me cancelled. Or render me totally, permanently, completely unemployable.

There's a story of how I turned a random Bulgarian software developer into a vegan dance music superstar, and how someone tried to bribe me to promote their crypto company, only for it to backfire spectacularly.

I'm not treating this newsletter as a proper one -- it's more of a personal blog, something that I wrote for myself and decided to push into the digital ether. Still, you may find some of the stories entertaining.

The Next Web was the first site that gave me a steady salary as a journalist. It was a place where I grew as a writer and a person. And, after almost two decades, it's closing down. I wrote a eulogy about why TNW mattered to me, and to tech media at large. whatwelost.substack.com/p/rememberin...

Google AI overviews nearly sent me driving to the wrong airport terminal (and nearly caused me to pay twice for parking). Honestly fuck everything about this.

The Next Web was the first site that gave me a steady salary as a journalist. It was a place where I grew as a writer and a person. And, after almost two decades, it's closing down. I wrote a eulogy about why TNW mattered to me, and to tech media at large. whatwelost.substack.com/p/rememberin...

Earlier today, I tried to find a post I'd written a few years ago, because it has a useful checklist that I consult regularly. You now can't search for it with the regular FB UI. The link to the URL endpoint has been removed. The *total* bastards.

A couple of months ago, I wrote about how tech products are increasingly designed to frustrate the user when trying to find stuff. Well, Facebook no longer lets you search for posts and organize (albeit algorithmically) by date. whatwelost.substack.com/p/losing-con...

Excellent thread. I think you can replace Big Tech with "big banks", "big corporations", and "extremely wealthy" and the points still apply. Long legacy of this type of stuff

Twitter reminded me today that it's been 12 years since I joined the platform. Actually, it's been longer. I joined in 2007, deleted my account around 2012, and re-created it that same year. It's not the place it used to be. A lot of people I really liked have disappeared.

Next post will be about web culture and its quiet death. It's a fun one, and it's only for premium subscribers. Want to read it? Give me your money. whatwelost.substack.com/p/the-benefi...

/fin. I'm not going to lie, this was one of the most draining newsletters I've ever written. If I could drink, I'd be pouring myself a scotch now. whatwelost.substack.com/p/the-benefi...

We've all held our noses and used a product from a company that we disagree with, or that we find immoral, simply because we liked it or because it was convenient. For tech to be held truly accountable, action can't come from the top. We all need to act. whatwelost.substack.com/p/the-benefi...

I believe that the unwarranted trust placed in these companies by the tech media has, in part, allowed them to act with impunity. There are no checks. No balances. But do we, as a society, also share some culpability for the crimes of big tech? whatwelost.substack.com/p/the-benefi...

It's hard to imagine that Instagram, part of a trillion-dollar company, had no idea that it was recommending exploitative content of children. Especially when Meta knowingly turned a blind eye to human trafficking on Facebook in the past. whatwelost.substack.com/p/the-benefi...

A Brazilian friend pointed me to a video from a journalist who did a deep-dive into how platforms like Telegram and Instagram act as marketing funnels for CSAM material. It was grotesque, and it's hard to believe that these companies were oblivious. whatwelost.substack.com/p/the-benefi...

The problem with writing about the offences of big tech is that you think you know everything, and then something shocks you. Last week, I was shocked. whatwelost.substack.com/p/the-benefi...

There are many, all of which aged like milk. The problem with this thinking is that it presumes a beneficence that doesn't exist, and simultaneously ignores the fact that these organizations are very wealthy, very powerful, and should know better. whatwelost.substack.com/p/the-benefi...

If you want to understand how we're in this situation, google "the benefit of the doubt" on any major tech publication. Here's one insane quote from 2010 by Michael Arrington that needs to be read to be believed. whatwelost.substack.com/p/the-benefi...

I think healthier relationships and societies are one based on trust and the presumption of good faith. The problem is that Big Tech has shown, repeatedly, that it's undeserving of that trust. whatwelost.substack.com/p/the-benefi...

I believe that, for too long, we've assumed that when big tech gets things wrong it's the product of human error, rather than malice. We found big tech's products useful. People like Mark Zuckerberg weren't what we imagined of corporate evil. whatwelost.substack.com/p/the-benefi...

In my last newsletter, I talked about the double standards that allows Big Tech to do things that, when done by normal people, often result in prison sentences. One thing worth exploring further is the idea of intent. whatwelost.substack.com/p/the-benefi...

This newsletter makes sense on its own, but it's technically a follow-up to this post. If you haven't already, you might want to read it first. bsky.app/profile/matt...

My latest newsletter is about how a credulous media and a trusting society emboldened Big Tech to do serious harm. And it’s about why these companies don’t deserve our trust or our forgiveness. whatwelost.substack.com/p/the-benefi...

I can’t claim to fully understand every nuance and detail, but this by @adambell.bsky.social is insanely interesting reading. unbalancingmechanism.substack.com/p/the-ghost-...

I was searching for a health thing a couple of weeks ago (don’t pretend you don’t google symptoms even though you know it’s insane) and one of the AI results was from a health centre actively lobbying against a specific female treatment. So that felt great and not at all worrying

This long read in The Verge does a remarkable job of describing how Wikipedia's editing community works, the project's strengths and weaknesses, and the threats it faces. www.theverge.com/cs/features/...

I actually had to Google this to check it was real. It is.

Sad news for @edzitron.com, who can no longer claim that Jensen Huang is the only person making money from generative AI.

So, my family doesn't use Bluesky, so I have no problems in sharing something I'm doing on Sunday that could either work really, really well, or could backfire on me spectacularly.

We're in a deadly situation where every company building software on top of AI models is already unprofitable, and that's before model companies are forced to start charging their actual costs. And if the GPU companies are losing money too, everything is doomed. www.wheresyoured.at/why-everybod...

At the heart of gen AI sits a massive economic problem: LLMs are too expensive to charge a monthly fee for, and that is what every single LLM company charges. This will never work. www.wheresyoured.at/why-everybody-is-losing-money-on-ai/#what-if-it-isnt-possible-to-make-a-profitable-ai-company

At this point, it's clear that it's impossible to build a profitable company around generative AI. Perpelxity spent 140% of its 2024 revenue on AWS, OpenAI and Anthropic, and winning that revenue requiired $27m in refunds. These are not good companies. www.wheresyoured.at/why-everybod...

When I dug in, I found a worrying fact: providing GPU compute might not be profitable, with one H100 costing around $1.46-$2.10 an hour all in. These things have to be run for years to get close to break even. www.wheresyoured.at/why-everybody-is-losing-money-on-ai/#is-it-profitable-to-rent-out-gpus

Worse still, even the GPU providers aren't making money. CoreWeave, Lambda and Nebius all carry masses of debt, funded by the GPUs they buy from NVIDIA, which they use to build data centers and buy *more* GPUs from NVIDIA. It's dodgy circular finance. www.wheresyoured.at/why-everybod...

The cost of AI is increasing, to the point that many startups pay 100%+ of their revenues directly to OpenAI and Anthropic. VCs investing in AI startups effectively feeding money directly into two companies, who in turn feed it to Microsoft, Amazon or Google. www.wheresyoured.at/why-everybod...