Always searching

Pfifferlinge, Rote Johannisbeeren, Butterbrot,

Turtle Quilt by Laurraine Yuyama

Like when people argue whether warhammer 40k is fascist propaganda that says that fascism is good, or if it's actually a satire of fascist propaganda saying that fascism is bad. When it isn't really either of those things. That's not what it's going for and that's not what it's doing. It's got satirical elements floating around in there in varying degrees depending on era and author, but the point of them isn't to instruct the audience whether fascism is bad - it's for drama, and mood, and weird interestingness, and humour, and all the other reasons why things are in art other than political and moral instruction. It portrays a bunch of fucked up stuff connected by in-setting circumstance and lets you make your own mind about it, or just let yourself think "damn that's fucked up" and leave it at that.

It's not even, like, complex, by almost any literary standard. It's the same thing with them being unable to understand marvel movies, literally the most accessible movies ever made, when they crack out the most basic #101 nuance for teenagers where a villain has a sympathetic motive or a hero makes a mistake, and they can't really understand it, thinking that because the villain has a sympathetic quality then the movie is trying to demonise that sympathetic quality, because they assume that the film is operating at the level of basic propaganda, when it's, like, even slightly above that.

"That everyone has their perfectly fitting other? Like puzzle pieces?"

"No" I would sigh "We have many gloves that fit our hands perfectly, each different and special, none like the others"

"It's been years you see, that I've not been looking for girls, but for a woman. That I could listen to for hours. That I could understand with no effort. That I can count on like I count on my feet to carry me. That I could cherish and treasure better than I do myself. I am taking the bet that that's you, but even if it's not, this adventure is worth its while in present"

How many woo-woo-to-fascism pipelines do we have to create before we accept that any system in which we agree to prioritize what makes people feel good and comfortable over objective, observable reality is inherently prone to radicalization. How many times are we going to act shocked that people who refuse to accept scientific evidence in regards to crystal energy and repressed memories then move on to refusing to accept scientific evidence in regards to vaccines and chemotherapy. How many times are we going to hold people's hands and tell them of COURSE it's okay to believe things just because they Feel Right and not because you have even a shred of evidence and then act shocked when they start believing the people who tell them that all their problems are caused by minorities?

See now that is going too far in the opposite direction.

Looking at this, to be in the global top 1% of income earners, you needed an annual PPP adjusted pre-tax, post-pension income of just over 250k Dollars in 2025:

A median full-time income in the US puts you just shy of the threshold for top 10%, but I think that is without pension scheme deductions, and also this means that about half of US full-time employees earn less actually, and not everyone does or can work full-time. If someone makes less than 30k per year, they are actually below the global average.

One could even argue that if you live in a big city, your dollars are actually worth less than PPP dollars. And we could go into how wealth inequality is actually worse than income inequality, and the top 1% of both don't necessarily overlap completely, because capital gains are usually not counted as income.

After demonstrations were planned in 18 locations across the Strip, the streets were unusually empty after organisers had called on Hamas to "disarm and to transfer civil administration of Gaza to a transitional-governing authority".

Hamas gunmen were seen patrolling the Strip after several days of crackdowns leading up to the day dubbed June 26 Revolution.

Multiple arrests, beatings and death threats for those suspected of inciting or participating in protests appeared to quell the unrest.

One Palestinian in Gaza told The Telegraph: "People are afraid and exhausted. Torn between the desire for change and the risk of being accused of plotting a coup or collaborating with Israel, any popular protest movement appears to have little chance of succeeding."

Streets in areas such as Khan Younis Port, Al Qarara and Al Nasr, were eerily quiet on Friday. "People are scared," another Palestinian told The Telegraph.

In advance of the protests, a fatwa was issued by the Association of Palestinian Scholars, an influential Palestinian group, warning that demonstrators would be accused of collaboration with Israel, a crime punishable by death under Hamas rule.

Another source on the ground in Gaza told The Telegraph anonymously on Friday morning that there were Hamas supporters and troops positioned at key intersections.

The source added: "However, they are clearly avoiding any overt or formal deployment for fear of being targeted by Israel.

"What Hamas appears to be most concerned about is the possibility that armed groups could exploit the protests to expand their control over additional areas of Gaza, or that Israeli forces could use the demonstrations as an opportunity to target Hamas operatives.

"In any case, according to activists, this is precisely what they hope to avoid. Any intervention by Israeli forces or armed militias would provide Hamas with a pretext to crack down on peaceful protesters."

According to Gaza-born Ahmed Fouad Alkhatib, a Palestinian humanitarian activist who lives in exile in the US, the Gaza hospital that became the focus of international condemnation after a deadly explosion early in the war is now being used by Hamas as a centre for interrogating, threatening and detaining opponents.

He claimed friends had been summoned to Al Ahli Arab Hospital, also known as the Baptist Hospital, where they were questioned by Hamas security officials about suspected involvement in the demonstrations.

"They were told explicitly by al-Qassam Brigades operatives, Hamas police, and internal security officers that if they post anything supportive of the protests on Facebook or offer any help to protesters, they will be executed under 'revolutionary conditions' and treated as collaborators with Israel; no trial, no process, just immediate death," he wrote on X this week.

He alleged that some were subsequently interrogated, threatened and placed under house arrest inside the hospital itself, "the same hospital that grabbed global headlines early in the war after a faulty Islamic Jihad rocket fell and caused an explosion which killed hundreds of people".

He said: "That hospital is now a central hub for Hamas's intelligence, militancy, and internal repression.”

On Friday morning, Mr Alkhatib also claimed Hamas had deployed security forces throughout Gaza to prevent any demonstrations.

"Right this moment, across the Gaza Strip, Hamas's police, intelligence units, and al-Qassam Brigade militias are fully deployed to crush the June 26 protests," he wrote.

Last year, protesters against Hamas took to the streets for several days in a row, with the Iran-backed terror group accused of beating a man to death and leaving him on his family's doorstep in a warning against any further protests against its rule in Gaza.

Uday Al Rabbay was kidnapped shortly after taking part in the protest. His body, covered in blood, was returned to his family's home days later.

Earlier this week, organisers appealed to governments, journalists and human rights organisations to monitor events closely, warning that participants had already become targets of intimidation.

"The right to peaceful assembly and political expression is a foundational civil liberty," they wrote. "The population of Gaza is entitled to exercise this right without fear of violent reprisal."

The movement accused Hamas of circulating threats through affiliated social media channels before the protests had even begun.

"These statements explicitly identify protesters and journalists as targets, indicating a coordinated effort to suppress dissent prior to its public expression," it said.

On Thursday, the Fatwa Committee of the Association of Palestinian Scholars, which operates across the Palestinian territories, sought to deter participation by issuing a religious ruling declaring the demonstrations forbidden.

The fatwa described the planned protests as "a criminal movement", accusing participants of assisting "the occupier" [Israel] and advancing objectives Israel had failed to achieve through military force.

Posting my result for posterity's sake, but one drawback with this quiz-and-compass approach is that it can't really account for personal uncertainty in the outcome space. I genuinely don't know whether capabilities will level off at the point where we get "Global GDP keeps growing at 3% per annum, yawn" or it dramatically changes absolutely everything. This in turn leads directly into my assessment of AI risk: it could be sci-fi nonsense or we could be dead already; it all depends where we are on the sigmoid and a priori I can't see that— the nature of exponentials and sigmoids is that tiny changes in initial conditions lead to wildly different outcomes. Both seem obviously possible to me, and my only firm epistemic commitment is that I disavow anyone who's sure it's one or the other. Capability agnosticism, I guess.

The booking software I have at my restaurant job has a field now where AI tries to summarize the buying patterns of our customers.

Turns out, and wait for the amazing insights gained by this amazing, world-changing tool, that the thing most customers buy is alcohol.

This is a feature that fundamentally doesn't function, and which anybody can see doesn't function just by looking at it. The fact that Claude or whatever is very good at python is not relevant here, hell, I bet you can get Claude to explain why this tool isn't useful without much prompting.

As a civilian there is something intensely frustrating about this. It's all promises from demonstrably untrustworthy people, and the cost of computing has massively shot up to provide tools that, for the most part, don't function.

I keep waiting for Google's AI feature to tell me something that I need to know which is both accurate and not already on the first page of a Google search and I've never gotten it to work.

It's just straightforwardly (And confidently!) wrong whenever I ask it something like that.

"Oh, well, you're using the wrong AI and doing it wrong".

I'm less and less convinced that people in the tech world, who often use advanced models and don't listen to much of the marketing because they know what it is already, understand exactly how insanely AI is being marketed to the rest of society.

Like, the moment you built an AI tool to analyze restaurant customer behavior and saw that 99% of the output was,

"Most ordered items: Coffee, Wine, Cocktails"

Why didn't you just scrap the whole thing?

It's all very "Apple Newton" to me.

The Apple Newton didn't fail because handheld personal organizers and handwriting recognition were bad ideas that fundamentally don't function, it failed because it didn't fucking work.

A lot of things that visibly, trivially, obviously don't fucking work are being sold or added to existing products simply because they are AI.

Personal computing has gotten massively more expensive so that huge data centers can process customer data to explain to me that many of our patrons drink wine.

See, the thing is, AI has gotten to the point where it's genuinely useful in software. Not "replace all the workers" useful, but useful enough that it seems possible that it might get there someday. And one of the things you'll discover about software people, if you hang out with them long enough, is that they have a really hard time conceptualize the world outside of software.

In my experience this is most severe with database people, but I have yet to meet someone who lives and breathes software and can genuinely understand that most of the world doesn't work that way.

You see, software is incredibly error tolerant. Yes, I know, you misplace one semicolon in a thousand line code and it breaks, but the debug mode finds that and you fix it. I'm talking about the way it can take huge amounts of bad inputs and just chug through them in loops to still provide a generally useful output. Because of this, every software person I've ever met has this idea that error rates that would make the eyes of anyone in the real world water are no big deal and that any software could do "the job" better than a person. Maybe I'm generalizing a bit, but not as much as I should be.

And, unless I'm mistaken, that's what you're seeing here; most of the bullish arguments I see about AI are coming from software people who are seeing it transform their own field and not understanding why it wouldn't be the same in fields that can't tolerate the same level of error.

As far as what you're seeing at work, yeah, that's a bunch of people who need to make their money back who have bamboozled a bunch of executives into shoving AI into an application that could be done on a Commodore 64 or even just by actually talking to the people you already employ. You're completely right that that part is a huge bubble that's going to burst soon, probably in the next year or two if I had to guess.

That part is fairly straightforward, though, what really interests me is what comes next. When the bubble pops, a bunch of people are going to be able to buy datacenters and intellectual property for pennies in the dollar. The next wave of AI companies won't be racking up trillions in debt to make their product and they won't have to make desperate moonshots to make it pay off.

I agree with you a fair amount with people totally not understanding what The World Outside Software is like. I saw someone saying that they thought robots might take over typical factories in six months. Six months isn't enough to get a machine qualified to handle my job's fragile, fiddly, expensive products even if the machine is totally ready to go. Maybe if it comes with a small army of consultants and field engineers to set it up and test it.

I think the term you're looking for isn't "error tolerant", but that re-doing things is cheap and frictionless and scalable. Some kinds of software can kill people, but even then you can often simulate it. Work in the real world, with a very limited number of exceptions, is always time-consuming and expensive and failures have a cost to recover from.

I am hugely sceptical of AI in general, have heard about some niche application in which it is extremely useful (custom AI programs churning through huge amounts of text in academia to return with statistics and citations, diagnostic tools in laboratories, etc.), have heard alarming reports of the dangers to young or mentally unstable people, but do not believe it is an existential danger.

My uninfomed opinion: Definitely overhyped new technology, some benefits, some dangers. The dangers are exaberated by the hype.