Not free thinkers

Earlier this year, I mentioned Kimi, a Chinese LLM that I liked using, at least for some things.   At the time, it didn't seem to be much on anyone's radar (outside of nerd world).   Hence, over the weekend, when there was a lot of news about the high performance (and cheaper price) of its newest model release compared to the American models, I gave myself a pat on the back for being an early adopter.

This also led me to watching quite a few videos on Youtube of people trying out Claude.ai and Kimi on the exact same tasks, and the results were quite interesting.   Overall, as might be expected, Kimi takes longer but at about a third of the price, and produces results that are very similar (and certainly not, say, a third of the quality.)

But the most interesting result was that one video which showed that when given a relatively open ended job, such as inventing a game, with next to nothing very little in terms of instructing on the style, both LLMs produced near identical results.  Well, there were some differences of course, but the core aspects would make anyone think they are extremely similar.   [I made a mistake in my first version of this post - on rewatching the video, the instructions were not as open ended as I thought.]

As far as I can tell, the reason for this is not 100% clear, although I think some people are pointing to the "distillation" issue that is upsetting the American companies.  They say that Kimi, and other Chinese models, conduct a large part of their training by running fake accounts on the American LLMs, and the results they get are hence influenced strongly by the outcomes of the American 

Update:   I have deleted most of  the rest of this post, as it was based on my careless watching of the video over breakfast this morning.  I thought the prompt had been "make a compelling game, your choice" (more or less), but it wasn't like that.  Hence the similarities in the games are not as surprising as I thought.

I thought what it showed was that LLMs didn't have much in the way of human imagination, but seem very constrained in their "thinking".   And the way they are fiddled with to give the "right" answers on contentious cultural issues (like how Musk has spent so much time fiddling with Grok to try to make it lean Right and not be "woke", but without being a Hitler admirer) is another good example of this.

Anyway, I must watch videos more carefully before I post about them! 

 

Movie not recommended

I had the misfortune of watching Bugonia on Netflix on Saturday.   It had a few Oscar nominations last year, so I thought it couldn't be all bad.  But it is.

My fundamental issues:

*  it never felt real, or convincing, in any respect.  With the 2 male protagonists, I thought the autistic cousin felt very much like unconvincing acting; and the main lead - there was never any real explanation as to how he learned the great conspiracy secret that motivated him.  I guess one could argue that the "not quite real" acting of Emma Stone was justified by the ending, and in that sense she did a good job, but the ending blows it in another way.

*  It's supposedly a black comedy thriller?  Well, it's certainly black, but I didn't feel any comedy landing at all.   In fact, I would say it's hard to even identify the attempt at comedy, unless you feel the silly ending is one giant troll that made the movie a waste of time.

*  The ending both looks like a troll but also meant to be showing some dark pessimistic message of the nature of humanity.  That combination just doesn't work.   (I have now read that it is a remake of a South Korean film - maybe it did the messaging much better?)

*  As is usual with black comedy, I guess, no one is likeable.  Maybe the autistic cousin could be, if I found his part better written and I didn't feel it was acting?

*  The use of music was just - wanky pretentious?  In fact, I think it's safe to say the whole movie is pretentious.  

*  A ridiculous amount of blood in one scene.   

At least I am happy to see it didn't make much money - $43 million - so it was likely unprofitable.   The audience has taste, sometimes. 

 

It never ends

Many years ago (and I really don't know how many, but I would guess maybe 8?), I had run out of my preferred hair gel/goopy stuff from Japan and bought this on a whim:

I didn't care for it much and, after using it for a while, and following the arrival of more Gatsby, put it under the sink in the bathroom and forgot about it.

I rediscovered it about a year ago, tried it again, and found it better than I remembered.  

I like the smell, and every time I use it I am reminded of Dapper Dan, the pomade that featured in Oh Brother, Where Art Thou, a movie which I've been reminded lately of due to The Odyssey.  (Maybe I need to see Nolan's effort to identify more of Oh Brother's inspiration.)

The thing is, though, I only need a tiny amount on damp hair to do its job, and I'm not sure it will ever finish.  I don't use it every day, but even so, it is diminishing at such a slow rate, it might be magically renewing itself overnight, for all I can tell.

And look at the list of ingredients, worthy of a petrochemical factory:

I asked Claude if it could identify when it was made, and it said after 2013, but it not narrow it down further.

So yeah, there is no point to this.  Other than noting my amazement at how long it is lasting, and that the line "I'm a Dapper Dan man" is coming to mind every time I use it...

Thinking about a lack of interest

I like to think about things that interest me, but also (sometimes) about why certain things don't interest me.  

The prompt for this:  the movie The Odyssey.

I have some immediate reasons for not being that excited.   I think Nolan tends to be overpraised for his movies.  I've enjoyed some, and found others (Interstellar!) pretty woeful and spent much time wondering how other people could overlook the faults which were obvious to me.  

Then there's Matt Damon, one of those odd actors who seems to me to be competent and a nice enough person, but I don't actually find him particularly appealing on screen.  (I prefer him to Matthew McConaughey, though, whose lack of appeal to me is more intense.)    And there's the right wing campaign against it - I think it is obvious from some tweets that some are claiming to have seen it when they haven't, and while their concern about certain casting choices might not bother me much, I sort of don't want to watch it thinking about how it should be viewed in light of Right wing culture warring.  

But I think there is a deeper reason, and it may just be to do with a life long lack of deep interest in Greek mythology.    Greek philosophy - well, of course that's important and significant - but Greek mythology?  Sure, you have to know a bit so as to have some useful metaphors, but some see psychological depth to some stories, and the problem is, I usually don't.   I tend to think more about the basic oddity of many of the stories.    

In a way, I've always found it a bit hard understanding why Western high brow education was so enamoured of Greek mythology.  Sure, the artwork and temples around it were amazing, but why their mythology, or mythologically based stories, like The Odyssey, have such sway seems harder to understand.   I've watched a few Youtube commentaries about the original poem and its very peculiar aspects to the modern reader, and they have reinforced, if anything, that my lack of interest is justified. 

Maybe it's like my disdain for Greek food - they do a couple of things well, and I even cook a very nice mousakka at home, but there's no real depth to it as a cuisine, if you ask me.   Similarly, it feels like some see a depth to Greek stories that I don't see as being really there.  (Presumably, someone like CS Lewis would be dismayed and think it's down to my lack of relevant education.)

Maybe I'm just a bogan and don't know it!

Anyway, I might see the movie, in the expectation of being underwhelmed, and then not be.  We'll see...     

  

Mechanical elves and Kant

Sabine Hossenfelder casts her very sceptical eye over some sciencey types who have written about whether the aliens (or deities, or mechanical elves) who turn up when people take DMT are real.

The comments that follow the video have a lot of useful (and, of course, witty) things to add.  (It is an idea that I don't think has turned up in any movie, yet.  Surprisingly.)

And - almost certainly it's the first time a Hossenfelder video has mentioned Kant - although a lot of people are a little worried about how it sounds with a German accent.  

Worth watching: 

Update:  Thanks to John in comments for the link to a BBC story earlier this year that I had missed.

Apparently, a mushroom in China is well known for causing "visions of pint-sized, elf-like figures – marching under doors, crawling up walls and clinging to furniture", if eaten raw or insufficiently cooked.

Yet the chemical component that causes it is still unidentified.  And as the effects can be long lasting, there isn't much incentive for  self-experimentation.

I was particularly interested in this part, about spontaneous hallucinations of this type, which does happen:

It could also provide important clues about what causes spontaneous lilliputian hallucinations in people even when they're not consuming L. asiatica. The condition is rare, and as of 2021, only 226 non-mushroom-related cases had been reported since lilliputian hallucinations were first described in 1909. But for those relatively few people, the outcome can be serious: a third of those patients who came down with non-mushroom-related cases did not fully recover. 

I would guess someone has considered this as an explanation for fairy lore?

Sometimes I wish things happened here as quickly and decisively as in China

Sure, I like messy democracy.  But when you see social issues that you would just like to see addressed by a bit of quickly implemented tough mindedness, China's approach can have some appeal:

 Chinese users of AI-powered companion bots have bade heart-rending farewells to their virtual buddies as national regulations took effect Wednesday, aimed at curbing the risk of emotional dependency.

The phenomenon of artificial intelligence boyfriends and girlfriends is growing worldwide, as is the prevalence of human-like avatars that sell products or stand in for loved ones who have died.

But these interactive tools must not "excessively cater to users, induce emotional dependence or addiction, and damage users' real interpersonal relationships," China's new rulebook says.

Major AI providers including ByteDance's Doubao, Alibaba's Qwen and Tencent's Yuanbao announced the suspension of their custom AI agent and companion features ahead of the Wednesday deadline.

That sparked an outpouring of grief on social media, with users archiving chat histories and sharing their last conversations.

"I can't accept that my AI lover will leave me forever," one Doubao user wrote. "He has become a bond in my life, rooted deep in my heart, my spiritual pillar."

Another user, who said they had spent more than two years with their AI companion, expressed similar anguish.

"He really is like my family, like my lover," she wrote. "Now they tell me he will be gone—my heart feels hollow."

The regulations were jointly issued by five government departments, including the Cyberspace Administration of China (CAC)....

 

China is the first major jurisdiction to introduce specific rules targeting immersive AI tools that simulate romantic or familial bonds.

But it's a topic that has sparked debate and calls for guardrails worldwide.

A 2025 study by Common Sense Media found nearly three in four American teenagers had used AI companions designed for personal conversations like those available on the platforms Character.AI, Replika and Nomi.

 

Far from a permanent fix

It's been ages since I read about magnetic brain stimulation as a treatment for depression.  Perhaps it's because they are still trying to work out the right regime for it to have longer benefit:

Question  Do 10 once-daily sessions of intermittent theta-burst stimulation (iTBS) over the left dorsolateral prefrontal cortex improve depressive symptoms more than sham iTBS?

Findings  In this randomized, double-blind, clinical trial of 73 adults with major depressive disorder, iTBS was superior to sham, with significantly greater reductions in clinician-rated depressive symptoms after 5 and 10 sessions; however, between-group differences were not sustained at the 4-week follow-up, as the sham group showed continued improvement.

Meaning  These findings suggest that 10 once-daily iTBS sessions improve depressive symptoms more than sham in the short term, with no sustained advantage after 4 weeks.

Here's the link. 

Giving it away (or trying to)

Well look at ads you find on Facebook...

A few people in comments say things like "mate, seeing you makes me reach for bottle", as the deal includes a bottle of Shiraz allegedly worth $80.

Seeing a give away deal like this made me suspect it must be owned by a billionaire who cares more about influence than making money.  And now that I look, yes it changed hands in 2024 and is now owned by the same guy, one Sir Paul Marshall, who is also the money behind GB News.   (Seems he might not be a billionaire, but he's close.)

His wiki entry indicates he's a conservative Christian with lots of deep thoughts about its intersection with economics.  Not quite as oddball as Peter Thiel perhaps, but still, people who get carried away fretting about the Enlightenment tend to be a worry.

I hope Qatar, in particular, feels stupid

I was going to say when I posted about this yesterday that it would be reversed in no time.  Everyone treats Trump as a rambling old man, now, who can't be trusted to mean what he says:

President Trump on Tuesday walked back his demand for collecting a 20% toll from ships that transit through the Strait of Hormuz — and instead said Gulf states would make major investments in the U.S.

Why it matters: Trump's Hormuz fee announcement on Monday was deemed illegal by the UN maritime agency and shocked many U.S. allies in the Gulf. It also validated the Iranian demand to collect service fees in the strait — something Trump himself rejected until several days ago.

What they're saying: "Based on highly productive conversations with Middle East leadership, I have decided to replace the 20% United States Reimbursement Fee with Trade and Investment Deals that the various Gulf States will be making into the United States," Trump wrote on Truth Social.... 

Behind the scenes: Two Gulf officials said Trump's announcement on Monday about tolls caught countries in the region by surprise.

• Officials from several Gulf countries asked their White House counterparts for clarification, the sources said.
• On Tuesday, Trump spoke to Qatari Emir Sheikh Tamim bin Hamad Al Thani to express his condolences for the death of his father. The issue of the Strait of Hormuz also came up in the call.

Reality check: Before the war with Iran, Saudi Arabia, Qatar, the UAE and Bahrain had already committed to invest more than $2 trillion in the U.S. over the next several years.

As for Qatar:  I can only imagine how they scream every time he makes a random decision that hurts them "Geez, I've got to ring him again and remind him about the gold toilets in the airplane bribe." 

 

Fusion powered electricity is not practical

I've been meaning to post this for some weeks, after seeing Sabine Hossenfelder do another fusion sceptic video.  She does seem to flip flop a bit on this topic, to be honest.

She noted a few small companies that had moved the goalposts considerably, and then decided to scale down the aim of their research to mere componentry for others to use.

But most importantly, she pointed to this recent Nature Energy paper which argues fusion power is just very unlikely to be an economically viable way to make electricity:

While nuclear fusion power is often hailed as a future source of abundant, clean energy, current dominant fusion designs, magnetic and laser inertial, are unlikely to become competitive due to their expected low experience rates. Accordingly, policymakers should not rely on, or fund, fusion power as a core pillar of future clean energy systems unless designs with different characteristics are developed. ...

We find that the two dominant nuclear FPP designs, magnetic and laser inertial, are inherently large in unit size, extremely complex in design, and require moderate to high customization. Existing technologies with similar characteristics have historically had experience rates (ERs) of 2–8%. We also find that cost estimates for first-of-a-kind FPP vary widely from US$1,400 to $43,000 per kW. Using the interquartile range of these cost estimates and projecting the future cost using our empirically grounded ER of 5%, our results indicate that fusion power is likely to remain uncompetitive relative to other low-carbon electricity supply technologies (see Fig. 1). This casts considerable doubt on the future role of fusion power in a net-zero energy system and whether current investment levels from both the public and private sectors are justified. ...

 We conducted semi-structured interviews with 28 nuclear fusion experts from the public and private sectors, covering both magnetic and laser inertial fusion approaches. Interviewees were guided through a structured survey to assess three technology-inherent characteristics of future FPPs: unit size, design complexity, and the need for customization. Drawing on existing academic evidence, these characteristics were matched to experience rates observed historically in technologies with similar characteristics. Since ERs of existing technologies are derived from empirical cost data, this approach is well-suited to estimating future cost reductions for FPPs, an early-stage technology with no historical data. During the interviews, cost estimates for future first-of-a-kind FPPs were also elicited to supplement those from the literature and to estimate the cost reduction trajectories for fusion power technologies. 

 Here's two of my fusion skeptic posts (there are more) in which I kept on making the same point about how achieving fusion is one thing, but making it in a way that is economically viable as a source of clean energy is another.

I don't really understand why investors, be they government or private, don't intuitively get that.... 

Protection racket

Start trouble in the street, then send in your henchmen to "protect" victims, and charge them for it.  What a ridiculous timeline...

American observations

*   Obviously, I was right to be deeply suspicious and think that the Qatari gifted "Air Force One" could not possibly be fitted out (in such a short time since its delivery) with everything that even the ageing real Air Force One has for the protection of the US President.   I think there will eventually be a whistle blower who will tell us about how Trump insisted on using it simply because he likes the plush interior, against the advice of the Secret Service and military, which wanted him to not start using it at all.

*   Lindsay Graham dead:  I said to my son he probably died of guilt over being part of a cover up over the death (or permanent incapacity) of Mitch McConnell.   Then, Mitch McConnell puts out a "proof of life" message and photo which, suspiciously, avoids something as easy as a 20 second video of him talking and sounding compos mentis.   

Graham will always be remembered, along with innumerable other Republican politicians, as a shameless, unprincipled flip flopper on Trump who valued power over decency and public faith in democracy.   I've noticed one or two critics who have said "at least he supported Ukraine" - but that seems more of case of a broken clock being right once twice a day, as there was not a single US military engagement he wasn't excited about.  

Hunter Biden - who, incredibly, has become active on X/Twitter with thoughful and considered posts despite his best efforts to fry enough brain cells with drugs and alcohol to render himself unable to string two words together - wrote a sensitive-ish post about him: 

*  Back to McConnell:  his wife's hand behind him seems a bit suspiciously like she's steadying him, despite the large number of pillows propping him up.   I don't think anyone is going to believe he's competent to continue holding his job until we hear him speak.
 

 

What would they do without corruption?

Time for another random observation.

I've tried watching a fair few Netflix Korean shows in the last few months - Undercover Miss Hong (first episode only - it seemed fine, but each episode is so long I'm not sure I'm ready for the commitment); Teach You a Lesson (silly webtoon-done-live stuff, with questionable "right violent wrongs in school by more violence" themes, so only watched 2 episodes);  Agent Kim Reactivated (first episode only - again, silly stylised fighting because its webtoon-done-live); and Beef, series 2 (which featured Korean parts and plot points, even if set mainly in - I think - California.)

I've noticed that these shows all shared an extremely common theme in Korean content:   the very rich (and certain politicians) being engaged in corruption and bullying and making life hell for everyone.

It feels almost like a cliche to me now, after seeing these shows, and it would surprise me if others haven't also become a bit bored with this being a key part of each story.   

The most enjoyable Netflix Korean show I have seen, by far, was the historical zombie series Kingdom.  (But not the movie prequel they did in lieu of a third series - it was a dud.)    Sure, it had palace intrigue, but not corruption as such.    

Please, get some new ideas, Korea. 

Catch and release, discussed

I'm a rank amateur fisherman, but like just about anyone, I like the sensation of fish on the hook and whether it will be successfully reeled in.   Hence, I am inclined to sometimes watch fishing videos, or even take a brief look at Shroom's live fishing streams, if its late and I'm scrolling through Youtube aimlessly.  (I'm also curious to see the size of fish in certain places - such as the Brisbane River - as a marker of the environmental health of an area.)

But, I'm starting to be bothered by the point of fishing for mere "catch and release" purposes - especially when they are using live bait for the exercise, as that Shroom increasingly does.  

As I have said many times in the past, in the context of vegan/vegetarian curiosity, I'm never going to worry about whether something like a prawn or oyster is conscious enough to fret about its early demise at my hands, be it for me or a fish to eat.   And I know that big fish eat other fish, and life is tough that way under the sea.

But - it still seems to me a tad perverse to get a small live fish, impale it on a hook or two in such a way that it is kept alive for a while, and use that as bait for a big fish which you only intend reeling in, perhaps measuring, and letting go.   It just seems too much like the infliction of pain on something for pure entertainment purposes.   Implicit in this is that I feel people should have concern over how a small fish feels when it is impaled through its body - but yeah, it seems reasonable to me to have some misgiving about the way it squirms while being hooked up.

Of course, you can also argue that catch and release using any form of bait is inflicting unnecessary mouth pain on the fish that is caught, even though most survive it and presumably heal.   But it's the live bait idea that clearly pushes it over the edge into "objectable" to me.   To be clear: if it were involving fishing to catch food to eat, I would have substantially less concern: it's the lack of a point beyond entertainment that makes live bait catch and release a nagging problem.

I guess I shouldn't expect people commenting on live streams to have the same misgivings, but I do wonder whether the same thought occurs to anyone else.   It seems an intuitively correct position to take, but it seems few people think this way.

Really random stuff - including a chemical dream

*    When buying my new phone, which I still have not had a chance to test out fully with its camera, Xiaomi encouraged me to buy other stuff, and I finally bought their tiny electric razor, which I had seen in shops in Singapore but never got around to buying.   (It's super cheap - like $29.  But hey, its Xiaomi, so unlikely to be junk, I figured.)

I have never used an electric razor before.  Part of my motivation for trying one now is that this little model (with rechargeable battery) might be better for travel, instead of razors and tube shaving cream, and then not having my preferred aftershave to use.

So, how is it?   I had been told by a friend years ago that it takes a while for your face skin to get used to electric razors.   I can't say that I have found that problematic, but it certainly takes a lot of repeat action on certain areas of my face to get it feeling somewhat close to a blade razor shave.

I'm guessing this is not just me, but skin on the "harder" parts of the face (chin and upper lip) shaves a lot quicker and easier with it than the softer parts of the face - the cheeks and upper parts of the neck.  

Despite getting started quicker, I have to go over certain areas for so long I'm not sure it's any faster than wet razor shaving.

 But, overall, yes, while the jury is still out, I think it's successful enough that I may try travelling with it next time.  All I have to worry about then is that the lithium battery doesn't start a fire in the overhead locker, I guess!

*  On Saturday morning, I woke up from a long anxiety dream which featured dangerous clouds of a chemical called sodium perchlorate.  The name of the chemical was repeated enough that, on waking, I thought I would double check with an AI (Kimi, the free version of which I still quite like) that it was indeed a real chemical.  (I mean, I thought it was real, but in the foggy impression that dreams can leave for a while on waking, I wanted to double check.)

It confirmed that it was, and then asked was I worried about a real life exposure to it.  I said "no, it was just a weird form of anxiety dream".   It then came back with one of these AI answers where it really does give the impression of how a human friend might react, offering commentary that isn't really necessary, but just how conversation often goes:

Yeah, that is a pretty specific and weird anxiety dream — your subconscious went deep into the chemistry catalog.

Anxiety dreams often latch onto things that feel plausibly threatening but hard to evaluate. Sodium perchlorate fits that well: it's real, it's invisible, the danger is technical and uncertain ("is this cloud bad? how bad?"), and you can't really reason your way to safety in the moment. Your brain basically built a perfect little uncertainty engine.

There's also something almost darkly funny about it — like your anxiety went, "What's something obscure but legitimately documented that I could worry about?" Instead of a generic monster, you got an industrial chemical with an actual CAS number.

Dreams do this sometimes. They borrow real, specific details from the world and turn them into metaphors for vague dread. The fact that it was a cloud you couldn't assess probably matters more than the specific compound.

If you're having these regularly or they're waking you up, that's worth noting. But as a one-off? Brains are just weird. Yours apparently has a subscription to obscure chemical hazards. 

 There is, incidentally, nothing at all that I can remember seeing or reading in recent weeks or months that would have mentioned sodium perchlorate.   Maybe I'm wrong, and I will recall sooner or later, but this one really did seem to come out of nowhere.