I've been thinking about this over the weekend and I can't put my finger on exactly where (or if) I disagree, so I'm gonna write to sharpen my thinking, if you'll humour me.

Maybe the biggest disagreement is that I don't understand what classifies as world-changing to you. How can you both believe NFTs are already changing the world when the largest use case is improving the bottom line of specific luxury brands through destroying their grey market, yet not seeing LLMs make a dent in the universe?

I mean - are LLMs world-changing to the tune of the $200B put into it or to NVDIA's market cap? Hell no. Is AGI around the corner? Also no. Are LLMs world-changing at all? Well, I'd put it at 'somewhere between YouTube and smartphones'. They'll make a bunch of things worse, they'll make a bunch of things better, like any powerful tool. But because it's a tool that can so directly influence the core of knowledge work, I find it very hard to believe it won't change things at all like you seem to suggest. Whether that means the big corporations are making smart decisions about investments AI is an exercise left to the reader.

It's not that I'm not aware that LLMs are, to a large and arguably frightening degree, a bullshit machine. You know how much I hate Tesla's """self-driving""" for the exact same reason. I will not trust LLM output for anything with any serious consequences, just like I will never have Musk take the wheel so I can nap. But as a co-intelligence? Microsoft was at least somewhat on point by naming it Copilot. The copilot can do useful things but I'm still piloting this thing, I'm still making the decisions to the degree that I want to make them.

The guy behind NotebookLM makes the point that the useful thing about LLMs isn't just the model, but also its ability to combine that with a context window that is getting so large it's exceeding what most of us are capable of. It can see a needle in a haystack but, more importantly, it can see the entire haystack. Hallucinations are much rarer when the text is right there, so it's also markedly more accurate and can cite shit.

I was discussing LLMs with my FIL the other day. He has been writing reports on construction failures for almost three decades now. All his work, and that of his colleagues going back five decades, is digitized into an archive of searchable text. But he said it's hard to use in his desk research because he can only search on keywords and those keywords change a lot over time. My FIL also has a hard time getting started with a new report - not because he doesn't know his shit, but because he's not the best at structuring his thoughts into a logical, linearly ordered line of reasoning. I'm 100% sure his company would pay a lot for a suite of LLM tools to enable him and his colleagues to supercharge their desk research. Not to replace their expertise but to enhance it, by having an LLM surface decades of knowledge in a way that isn't possible now, combining it with relevant information in a case, and then also helping them make use of it.

Now multiply this use case for all academics and knowledge workers around. We still need professionals in the loop, just like I still need to be in control in a Tesla. We need expertise to tell bullshit from fact. That doesn't mean LLMs are doomed to be useless toys - even when I, too, use it as a toy now from time to time.

It's not just that I can conceive a universe where I want LLMs to be useful, it's that I and many normies are using it as a tool in basic ways and that there are a bunch of more advanced ways to use it that feel just out of reach. I genuinely don't believe I'll use these tools less in two years than I do today. So for me it's not in the "maybe it'll be useful someday" realm, it's already useful to me now and would be a lot more useful if the ways to use these models weren't so godawful half of the time.

I have in the past weeks used LLMs to:

- write a working, hosted, version controlled and github-linked web app in less than an hour that pulls data from APIs and displays them in a custom way (which I had spent multiple afternoons on before, not getting even 20% as far)

- help me figure out how to write complex Excel formulas

- write the first draft of a Request for Information in such a way that I could skip over the easy questions and hone in on the hard, important questions to ask suppliers

- assist me in figuring out how to diagnose filament extrusion problems I had with my 3D printer

- give me spoiler-free yet highly specific information on a video game I wanted to play

- explain a difficult concept to me in multiple ways from a field I don't know but just want the Pareto 80/20 gist of it

- do boring-ass restructuring of a large .txt tabulated file

- replace at least 30% of all Google searches because Google's search results are already garbage, the SEO slop is already most of it and I can avoid it with an LLM (which feels like garbage on top of garbage but it does the job)

For me, LLMs are mind-blowing when I see a completely novel way of arranging the tool and its inputs, when it opens a new avenue of use cases for me, like how recently multi-agent stacking has got me thinking about what else I can do.

That's different from if I can ever use it in that way. I've never found a personal use for NFTs, but I can conceive of a world where they are useful to me. The "I guess we're doing AI now" is exactly the same as "I guess we're doing NFTs now" was then; soulless corps gonna soulless corp. But just because there is a hype cycle around it all, does not necessarily mean it will recede back into obscurity like NFTs and nerd helmets did. It won't change the world but it's already had a larger impact on my day to day work than crypto has (and we've been talking use cases for a decade on that). How's that not worth something?

Isn't the argument not about Trump and his team being better, but about them being more effective? I don't think they need to be any better than last time to wreak more havoc, just simply not being inhibited from almost every branch of government due to subservience will probably be enough.

To draw a parallel to our extreme-right Wilders government coalition - 3 out of the 4 coalition parties are objectively incompetent at governing. The Farmer's Party managed to be so incompetent they gave away €20B in farmers' subsidies during coalition negotiations in exchange for nuthin'. The party who ran on better governance managed to admit to hide documents in their third week.

It is absolutely a clusterfuck, it absolutely leads to stupid and ineffective policies. There's a debate now about whether we can impose border controls again, which we can't and everyone knows it but Wilders wants his version of The Wall anyway. There, however, will also be inhumanely policies because that's what we're now facing: things like taking away passports from second or third generation immigrants, things like changing visums to be for a few years at most, things like making refugee camps so sober the line between camp and prison is almost entirely blurred.

I thought we couldn't go this inhumane, this low. I hoped their incompetence would prevent them from hurting too many people. But we can go this low, they'll dare you with how low they can go, and there's little we can do but watch in pain and hope.

And the general public doesn't seem to mind, or is happy that something is being done even if it doesn't work, will never work, will be hard to undo and repair. So I'm not so sure about that 'unpopular' part of the clusterfuck.

You know, WSJ, that article would’ve been a hundred times more interesting had you asked power users, researchers, people actually at the bleeding edge that question instead of Tom who likes to admit he’s cutting corners for his daughter’s wedding speech.

Two years feels like too long to really see where we’re going. To draw a parallel to smartphones, I’m undecided whether we’re still in the novelty iBeer phase of this technological change or have reached the Fruit Ninja phase of largely mass adoption.

The end stage, largely awful but hard to avoid TikTok stage of LLMs is still beyond reach I think. Even if the models themselves are already plateauing or would stop improving, there will be steps forward that I expect will be made and ways in which this will end up improving / enticing / sucking that we can’t foresee yet. I don’t believe this is it. How could it be it when we’ve only so recently unlocked the ability to mass-produce, mass-fake whatever we want in nearly every medium we want?

I recently played with NotebookLM, which has the ability to generate a listenable podcast out of any text you throw at it. I gave it a long, detailed report I know dearly and had just made a presentation about that week. It picked the same framing device that I did in my presentation, without said framing device ever being hinted at in the source material. That blew my mind. Even a fucking Suno song from the newest model version I heard the other day sounded passable, instead of peak cringe that must be exterminated with fire. Just today I saw some Midjourney photos that fooled me.

When’s the last time we were talking about six fingered people? Because it’s been a while, and I’ve resorted to accept that even the best of us won’t be able to tell the difference most of the time sooner rather than later. Right now it requires a modicum of thought and expertise and a boatload of tolerance for shitty UIs to get really compelling results, but that bar is going to lower and lower until even Judith can make candle labels that don’t trigger my AI slop spidey senses.

Whether that all translates into the professional world is harder to guess. I’m already seeing the beginnings of enshittification in, say, consultancy work which I expect to spread to all other types of knowledge work. I think the next push will not be from better models, but from better strings of models. That’s how NotebookLM works behind the scenes; it’s a string of models that first summarize, then stylize, then add uuh’s & aah’s, and then feed it into a speech synthesis model. Similar stacks will become normal for knowledge work. You plug your clients demands into an expensive and slow architecting LLM trained on all earlier reports, it writes a detailed outline for cheaper/larger context models to write, each part is fed to cheaper models to iterate over and then it’s all fed through a half dozen style, language, prose, models. Rawdogging ChatGPT is for losers (sorry, WSJ), prompt engineering in the right model is easily the best way to double the quality, but stacking multiple well-prompted models will be the next flurry of development if I were to bet on it.

I think one of the biggest risks will be the inherent biases in every model, the way its misdirection will be subtle as a whisper. A thought I had the other day - to what degree does, say, Squarespace exist in transformer models? Because their omnipresence in every podcast and YouTube essayist under the sun has to be a factor, the same way Whisper’s text-to-speech will hallucinate “don’t forget to like and subscribe” often because it’s been trained on YouTube transcriptions. Squarespace is a harmful omnipresence, but it’s not a far stretch to imagine all the other ways in which dominant groups or interests might influence models which in turn influences how we think or write or interact with each other. It could be that our thinking gets pushed a dozen …-ist ways at a time. It could be that laziness becomes the norm in everything we write and think. I’m not looking forward to finding that out after we now ostensibly fuck around.

There might be a way in which AI models will separate experts from the rest of the world. Why would any knowledge work based job hire an intern anymore? Why would they hire someone fresh outta college? I can think of a long list of work I did in my first years that I see completely being overtaken by AI already, or eventually. Which means that future experts need to either find a way to grind domain knowledge on their own, or spend much longer in academia / training. Because how else are you gonna learn how to discern the truth in your field from hallucinations?

I’m curious whether your thoughts have changed about the longer term consequences of this - what does this look like in a decade to you?

The best argument you can make about Trump’s first term is that there was a constructive tension between his disinhibition and the constraints of the staff and the bureaucracy and the institutions that surrounded him. Yes, some of his ideas were bad, dangerous and unconstitutional. But those mostly didn’t happen: They were stopped by his aides, by the so-called deep state, by the courts, by civil society.

But now the people around Trump have spent four years plotting to dismantle everything that stopped Trump the first time. That’s what Project 2025, and the nearly 20,000 résumés it reportedly vetted, is really all about. That’s what Trump’s inner circle is spending its time and energy doing. Don Jr. told The Wall Street Journal, “We want people who are actually going to follow the president, the duly elected president, not act as sort of unelected officials that know better, because they don’t know better.”

From this overly long Ezra Klein essay.

We’ve seen how it plays out, but isn’t there a pretty solid argument to make that this time’s different? That the GOP inhibitors are replaced with Yes Men this time around? Because the cabinet picks sure seem to me to be that way. Donald “let’s get the army to shoot protestors” Trump doesn’t seem to have people who will stop him at his worst anymore. Donald “I like the oil industry because they gave me money” sure seems to pave the way for special interests to screw everyone else over.

I don’t think it’ll get to mass deportations, but I highly doubt Trump will be able to resist his inhibitions to ruin the country in favor of golf this time around - because it won’t be as frustrating for him as the last time.

(3) They've seen it before, and they still want this.

I've been discussing this with friends the past week. There's a bunch of "if only she would've..." takes but I don't buy any of them. The 'every incumbant has lost this year' is going around a lot, and I think it is part of it, but it also feels like a nice economic scapegoat for libs - a safe haven of logic to avoid facing what I think is the harsher reality, which is that people do want this, whatever their imagined version of the next four years of this is.

And when people are scared, and people are tired, and people are worried, they choose simplicity.

Over here, after the dust of the election had settled, the consensus of the PVV's victory comes down to people voting for the extreme-right because they want stronger immigration. It's as simple as that. In previous elections the PVV were ostracized because of the whole far-right thing, but this election the neolibs said they wouldn't ostracize the party any longer. Suddenly, a PVV vote wasn't wasted anymore, so anyone who wanted to put their anti-immigration vote to good use flocked to the PVV.

I think there's a faint parallel to Trump's victory here - the simple answer could be that people hated the past four years (case in point: Biden approval ratings), and with the GOP now magawashed/normalized you're not gonna have a fight anymore with your family for voting President Chump.

---

Personally I am also pondering if I should re-adjust my belief that people vote for what's best for their country, instead of what feels best for themselves. Sigh.

From one extreme-right-elected country to another: good luck, you're gonna need it.

I'm still left to wonder why the Harris turnout was so bad - according to CNN she did worse than Biden did in every district in the country.

Got married! Travelled across Spain! So many good vibes that I still have a hard time comprehending it all.

As a surprise gift, we got an envelope with a date from each of the guests. Inside the envelope, to be opened on that date sometime in this first year, is a fun task to undertake for us. It is such a thoughtful gift and such a great way to celebrate our marriage, like when we “had to” dine with sushi and wine this week because it’s a go-to combo from a good friend.

Good luck all yall with the dumpster fire (/ballot box fire??!) that is next weeks election. Feels like the whole world is watching.

We kind-of-accidentally made the wedding a 5 day thing, which turns out to be a phenomenal thing. Pretty early on we made the decision to have our civil marriage not on The Day but a few days early, giving us much-needed breathing space on The Day itself. We invited only our core families and the witnesses for it. Since our municipality allows you to have your civil marriage in your own garden we took that opportunity and ran with it. So we're now officially marriaged, although we have yet to be religiously married and have saved all the extravagancies for The Day. What I did not expect was the civil marriage to already feel so much like marriage - but speaking our vows in front of the people most dear to us and them gracing us with kindness too was just remarkable. We're now in a lovely inbetween place where we can look back on a part of the marriage already wildly exceeding our expectations, look forward to The Day, and gently float on cloud nine for a few days while we sort out the final to-do's.

Next Pubski I'll be in Andalusia with a ring on my hand. Prolly won't be around much for a while, as I am using the trip to detox from my phone more.

Two weeks on the dot until the wedding and it's all we're thinking about. Is there a life after? Who knows? My mood cycles from excited to nervous to already-grateful to nervous at least half a dozen times per day. I know it'll all be wonderful but there's some yak shaving to do to get there.

What about Siemens? The previous owners graced our kitchen with all-Siemens appliances, including the induction plate. Doesn't light up at all and I have maybe one? gripe with it.

You did get me to wonder what my Siemens oven has that I don't know about... it has a pizza mode, and it has a superspeed heat mode, which leads me to wonder if it secretly also has convection in it or sumthin' because it draws a clean 5-6 kW when I put it in said mode.

It's nice, but my god I can't wait for the eighties nostalgia wave to wane and for the nineties/zeroes to begin any day now.

I feel like we've touched this subject a few times before. This article in particular I found an interesting reflection on the clock.

I'm about halfway through The Song of the Cell by Siddhartha Mukherjee. Have read the Dhammapadana.

Yesterday I started reading The Gone-Away World by Nick Harkaway after enjoying Titanium Noir a while ago. So far I like the latter more than the former, but it's a slow burn supposedly.

I am dipping my toe into buddhism and sailing. We had a day out with our department where we went sailing on a lake for the afternoon. It was a thorougly enjoyable experience because it reminded me of when I was learning gliding eight years ago. You're on a vessel, subject to the whims of the wind going (vertically/horizontally), and have a few axes of control (flaps/sails). How well you fare is something you can only partially control; it's really a challenge of sensing what the wind does and being adaptable enough to lean into that. Some colleagues were frustrated by the fitful winds but I never was; the winds just are.

The interest in buddhism comes from watching an unreasonable amount of Dr. K videos and livestreams over the past year. I find his blend of clinical, (neuro)scientific and yogic/buddhist perspectives on mental health fascinating, and some of his videos have genuinely changed our lives. In one of his streams he shared his understanding of dharma and it made me curious to learn more, so now I'm reading the Dhammapada.

Picked out my wedding tux today. I'm so happy I finally found a skilled tailor that has a fantastic eye for style and that I like working with, so it's essspensivo but so worth it IMO.

Armenia and Estonia keep popping up in my head days after the Eurovision finals

If I had to pick a winner, it'd be "France, if it wasn't off-key at times"

with of course Ireland as an honorable mention

Booked a trip to Italy today! We’re going by Nightjet night train, with the remarkable timetable of taking an 8pm train and being in Italy at 9am without a single high speed train involved.

I’m quite sore from landscaping our garden today and the past weekend. The work is very fulfilling - urban planning and garden planning are both design challenges in the real world that I like. It also vaguely reminds me of my years in Minecraft as a teenager because I’m paving with brick in a pixel-like pattern of squares and doing landscaping, lol.

Had a fantastic day with my sister and her kids the other day. I can honestly say I don’t think I’ve ever been closer to her, it’s great to be a better brother and uncle than I was years ago. Most of the time we do end up discussing our oftentimes difficult parents and upbringing - glad we turned out okay despite it all.

Did a bunch of gardening over Easter. We took out most of the stone bricks and I started repaved parts where we want to extend the brickwork. Quite fun actually to lay down brickwork like that, it’s easy to do but a bit of a physical challenge as I’m using shouldering back muscles that almost never get to work. I’m hoping the weather helps a bit the coming weeks so we can get the garden plant-ready asap.

Maybe I read over it, or it’s hidden incorrectly in the ‘low self-esteem’ or ‘task aversion’ category, but what I see with myself and my peers is that the biggest reason for procrastination is the perceived ability to do the task. “I need to write an essay, but I don’t know where to start” or any other task where the steps from now to finished thing is fuzzy, unclear, tricky, scary, or all of the above. I see so many people struggling with procrastination actually struggling with generating the activation energy necessary to start and subsequently not losing steam. None of that seems present in this meta-analysis in a way that I find matching my experiences.