« first day (5069 days earlier)      last day (147 days later) » 

00:20
Soviet psychiatrist is making photos of the patient's hallucinations
The psychiatrist's story is here 59.ru/text/health/2023/12/17/73018982
01:18
@Mitch Or you can just pirate it through Sci-Hub. Not that I'd ever do such a thing.
Tightrope, a daily trivia game | Britannica

Sep. 26, 2024

T I G H T R O P E
✅ ✅ ✅ ✅ 💔 ✅ ✅ ✅ ✅ 🎉

My Score: 1980
01:51
@Cerberus Well, yeah, in principle it's no different from making people pay money to vote.
02:35
@alphabet Right!
02:47
Odd grammar mistake of the day: "Place yourself into one of the children's shoes he's talking to."
(Hey, for once an actual grammatical mistake, not just something informal or nonstandard or on the list of things your English teacher dislikes!)
(Time to break out the good asterisks.)
Word of the day: bast. Shoes made of bast were very popular in Russia up to the 19th century.
@alphabet What's the difference?
Bast shoes are shoes made primarily from bast — fiber taken from the bark of trees such as linden. They are a kind of basket, woven and fitted to the shape of a foot. Bast shoes are a traditional footwear of the forest areas of Northeastern Europe, formerly worn by poorer members of the Finnic peoples, Balts, Russians, and Belarusians. They were easy to manufacture, but not durable. Similar shoes have also been made of strips of birchbark in more northern areas where bast is not readily available. Bast shoes have been worn since prehistoric times. Wooden foot-shaped blocks (lasts) for shaping...
@CowperKettle I praesume that is the same as bark?
Ah, yes.
In Dutch, trea bark is bast.
Waraji (草鞋(わらじ)) (IPA: [w̜aɺadʑi]) are light tie-on sandals, made from (usually straw) ropemaking fibers, that were the standard footwear of the common people in Japan. == Use == Waraji resemble other forms of traditional Japanese footwear, such as zori and geta, with a few key differences. They were historically the simplest form of outdoor footwear (sandals of any type were not worn indoors). Waraji, due to their cheap and rustic nature, are considered to be a very informal type of footwear, and are not worn with formal kimono. They are typically not worn with tabi socks, and are woven so that...
You're simply the bast,
Better than all the rust,
Better than anyone,
Anyone I've ever met
03:03
@Cerberus I encourage you to consult your linguistics textbook of choice.
What I mean is that I am not immediately convinced by the way you apply this distinction.
To my knowledge it's how every contemporary expert on syntax understands grammaticality.
Your average high school teacher or EFL instructor doesn't count.
I can't really comment without knowing your exact definitions.
But many other things you post here seem mistakes as well.
Again, literally any textbook on modern linguistics will explain this for you. I can send you the relevant pages of Huddleston & Pullum (2002) if you'd like.
I do not need basic explanations, I'm curious about what your interpretation is. But never mind.
 
4 hours later…
07:05
@Cerberus if we have a sea of names for the utilities in our phones it only stands to reason that our ancestors had 50 different names for 'bark'.
 
6 hours later…
13:28
@alphabet nothing came up when I tried to search there.
#travle #653 +1
✅✅🟧✅✅
https://travle.earth
The "mistake" was not a mistake. It's just as good as their preferred version.
14:27
#WhenTaken #213 (27.09.2024)

I scored 765/1000 🎉

1️⃣ 📍 14.2 metres - 🗓️ 0 yrs - ⚡ 200 / 200
2️⃣ 📍 10701 km - 🗓️ 8 yrs - ⚡ 89 / 200
3️⃣ 📍 136 km - 🗓️ 34 yrs - ⚡ 94 / 200
4️⃣ 📍 49 km - 🗓️ 1 yrs - ⚡ 197 / 200
5️⃣ 📍 122 km - 🗓️ 7 yrs - ⚡ 185 / 200

https://whentaken.com
Wordle 1,196 5/6

⬛⬛⬛⬛🟨
🟨⬛⬛🟨⬛
⬛⬛⬛⬛⬛
🟨⬛🟩⬛🟩
🟩🟩🟩🟩🟩
14:48
Daily Octordle #977
4️⃣9️⃣
5️⃣🔟
🕚8️⃣
6️⃣7️⃣
Score: 60
Daily Sequence Octordle #977
6️⃣🕚
🕐⓮
⓯🟥
🟥🟥
Score: 107
Ouch. Got into several 4/5 guessing games.
Tightrope, a daily trivia game | Britannica

Sep. 27, 2024

T I G H T R O P E
✅ ✅ ✅ 💔 ✅ ✅ ✅ ✅ ✅ 🎉

My Score: 1970
Misunderstood #4. This is the problem with having to rush.
15:08
@Robusto I find it very disillusioning when I come to find that the game makers are not necessarily some authority or all-knowing.
The question makers at Jeopardy! are pretty good but even they make mistakes.
They are also the judges during the competition itself and some of their judgements of mispronunciations are... among other things... racist.
eg one (white) guy will get a pass on a weird pronunciation, and during the same game a mispronunciation that's more of an accent difference is counted wrong.
Of course everything is arguable.
One time was the difference between 'gangsta' and 'gangster' as though they are two different words (one is arguably the non-rhotic version of the other).
@Mitch You have to paste in the URL or DOI code.
Usually the DOI works better.
Tightrope, a daily trivia game | Britannica

Sep. 27, 2024

T I G H T R O P E
✅ ✅ ✅ ✅ ✅ 💔 ✅ ✅ ✅ 🎉

My Score: 2140
15:31
@alphabet Huh
How about that.
PDF comes right up.
Has a crime been committed?
I don't know if what I did was wrong but you definitely suborned cutting and pasting of DOIs.
I never saw this kind of transportation before
An AI made a podcast in which two "hosts" discuss a paper on AI
#travle #653 +0
✅🟩✅✅
https://travle.earth
There was a "shorter" path.
#WhenTaken #213 (27.09.2024)

I scored 905/1000 🎉

1️⃣ 📍 961.7 metres - 🗓️ 0 yrs - ⚡ 200 / 200
2️⃣ 📍 5 km - 🗓️ 0 yrs - ⚡ 200 / 200
3️⃣ 📍 945 km - 🗓️ 10 yrs - ⚡ 157 / 200
4️⃣ 📍 86 km - 🗓️ 2 yrs - ⚡ 194 / 200
5️⃣ 📍 728 km - 🗓️ 13 yrs - ⚡ 154 / 200

https://whentaken.com
Wordle 1,196 4/6

⬛⬛🟩⬛🟨
⬛⬛⬛🟨⬛
⬛🟩🟩🟩🟩
🟩🟩🟩🟩🟩
Daily Octordle #977
5️⃣🕛
6️⃣🕚
🔟7️⃣
8️⃣9️⃣
Score: 68
Daily Sequence Octordle #977
5️⃣🔟
🕚🕛
🕐⓮
⓯🟥
Score: 96
@Robusto Yes, I needed 5 tries for the second word.
Tightrope, a daily trivia game | Britannica

Sep. 27, 2024

T I G H T R O P E
✅ ✅ ✅ ✅ ✅ 💔 ✅ ✅ 💔 🎉

My Score: 1700
#waffle980 5/5

🟩🟩🟩🟩🟩
🟩⭐🟩⭐🟩
🟩🟩⭐🟩🟩
🟩⭐🟩⭐🟩
🟩🟩🟩🟩🟩

🔥 streak: 6
wafflegame.net
16:07
@CowperKettle There are two things one can judge here (and they interact). ONe is the original paper, the other is the transcript. As to the paper, the title is misleading in the sense that the 'chain of thought' strategy for prompting to get better answers for inherently sequential 'thought' problems was well-known way before this paper (the title makes it sound like they're proposing this new, even though the abstract does not).
The paper itself is actually a mathematical justification that CoT (Chain of Thought) style prompting can work for discrete mathematical problems of a given length (TC0 and AC0 mentioned in the abstract are complexity classes of circuits of constant depth - these complexity classes are well-known and the paper talks about them well).
So my judgement of the paper is that, while I don/t care for the title, it is doing good science crossing computational complexity (the math behind how hard it is to compute various things) and explaining Large Language Models. (from a superficial reading of it... I couldn't tell you if the math is right).
#deluxewaffle122 0/5

🟩🟩🟩🟩🟩🟩🟩
🟩⬜🟩⬜🟩⬜🟩
🟩🟩🟩🟩🟩🟩🟩
🟩⬜🟩⬜🟩⬜🟩
🟩🟩🟩🟩🟩🟩🟩
🟩⬜🟩⬜🟩⬜🟩
🟩🟩🟩🟩🟩🟩🟩

wafflegame.net/deluxe
As to the youtube video, all the substance is in the transcript. And right from the start I found the conversation... very annoying... like two undergraduates who have overheard general LLM talk before and are trying to impress the other with 'deep statements' about LLMs but are saying things off, badly misinterpreted or wrong (which is a reflection of what actual people have written in the past year about these things in general).
Also, the transcript doesn't really talk at all (or maybe half a sentence) about what is -in- the paper, they just blather on about LLMs in general and Chain of Thought in particular but very vaguely.
The paper itself is very technical (needs a Master's level understanding of LLMs -and- computational complexity (both of which are pretty hard))... the transcript doesn't even touch on that even though presumably the paper itself was submitted as 'context' for creating the transcript.
I found the style of the transcript, the diction, the kind of bank and forth, to be -very- annoying. too many 'like's, too many responses of 'exactly!'.
16:26
@CowperKettle Felt boots too. Winter is coming. en.wikipedia.org/wiki/Valenki
As a demo of google's 'NotebookLM' I feel like the people who made the video didn't give notebook LM the paper but just prompted with 'give an eight minute dialog explaining Chain of Thought' and that's it, not even caring about the paper itself.
The NotebookLM is kinda fun, making a short dialog out of a prompt, with reasonable sounding voices.
The problem with all this LLM stuff is that if you don't know what's going on, they sound awesome and believable and so on. If you actually know something about the subject, it's like word salad with really good dressing.
I don't know what that metaphor means.
What I mean to say is that the language form is great but that the factual content is hit or miss.
 
2 hours later…
18:05
@jlliagre I don't see it, even after comparing with Google Maps. Spoiler
Unless they expect you to find the shortest physical path, i.e., "as the crow flies."
The Anti-Masonic Party was the earliest third party in the United States. Formally a single-issue party, it strongly opposed Freemasonry in the United States. It was active from the late 1820s, especially in the Northeast, and later attempted to become a major party by expanding its platform to take positions on other issues. It declined quickly after 1832 as most members joined the new Whig Party; it disappeared after 1838. The party was founded following the disappearance of William Morgan, a former Mason who had become a prominent critic of the Masonic organization. Many believed that Masons...
#waffle980 2/5

🟩🟩🟩🟩🟩
🟩⭐🟩⬜🟩
🟩🟩🟩🟩🟩
🟩⬜🟩⭐🟩
🟩🟩🟩🟩🟩

🔥 streak: 3
wafflegame.net
18:22
#travle #653 +5
🟧🟧🟧🟧🟩🟩🟧✅✅
https://travle.earth
Yeah I am very tried and didn't really think about what the fastest route would be.
> Southern defenders of slavery, for their part, increasingly came to contend that black people benefited from slavery.
Freedom is slavery.
#deluxewaffle122 2/5

🟩🟩🟩🟩🟩🟩🟩
🟩⭐🟩⬜🟩⬜🟩
🟩🟩🟩🟩🟩🟩🟩
🟩⬜🟩⬜🟩⬜🟩
🟩🟩🟩🟩🟩🟩🟩
🟩⬜🟩⬜🟩⭐🟩
🟩🟩🟩🟩🟩🟩🟩

wafflegame.net/deluxe
@Cerberus The Spanish comforted themselves with their belief that even though they enslaved and murdered numberless primitives, at least they were "saving their souls."
That rationale could be extremely tenuous, but they clung to it anyway.
Rationalizations aren't really all that rational.
18:38
@Robusto Right.
This was from the Americans, though.
It makes no difference. Predation is the same, whoever does it. And they all do it.
Indeed.
But this sentence in particular sounded Orwellian.
And yet all too common.
I don't know, it seems so silly!
Americans can be Orwellian, too.
I should know, I am American.
I find myself Orwellian at times.
Q. E. D.
18:53
@Conrado That's hardly news.
He wan't so much of a statistical outlier, either; a lot of wagies are in a kind of slavery contract.
Supposedly you can leave whenever you like.
But where's the light bill going to get paid from?
@alphabet Hmm yeah, if you think other races are intellectually stunted liked children...
But then why expand it to whites?
@Cerberus Because he thought capitalism was evil and inhumane, even moreso than slavery.
By the way, at the time there were probably still a lot of white slaves as well.
Or at least there had been for a very long time.
@alphabet So there was no Third Way between slavery and capitalism?
19:04
@Cerberus If you read the Wiki page you'll see an explanation of his arguments.
Perhaps later.
@Cerberus Certainly not in the US in his time.
But on other continents, there were.
@Conrado If you have a contract, you're not a slave.
I don't.
19:10
@Conrado what's a 'wagie'?
> In de middeleeuwen kenden grote delen van Europa vormen van slavernij. Verdun had ca. 800 een grote internationale slavenmarkt. Tijdens plundertochten van Duitsers, Vikingen en Byzantijnen op met name de Balkan in de vier eeuwen daarna werden hele volkeren gevangen genomen en tot slaaf gemaakt. De (Balkan-)Slavische herkomst van deze bevolkingsgroepen verklaart de betekenisovergang.
So the word slave is from the ethnonym is partly based on the fact that so many Slavic people were traded around as slaves during the Middle Ages.
@Cerberus when was all that?
@Mitch I think all throughout the Middle Ages.
Europeans traded each other.
Serfs are slaves, right?
Later, this trade of European slaves was mainly taken over by the various Muslim states.
@Mitch No, this is about actual slaves.
19:12
So 1861 in Russia
Serfs had some rights, slaves far fewer.
Slaves could be free people captured and sold to different lands.
That you could not normally do with serfs.
Right on time for my next experiment in plugging Civil War song lyrics into AI song generators:
Not even the lowest class of them.
> Most Slavic slaves were imported to the Muslim world through the border between Christian and Islamic kingdoms where castration centres were also located instead of the direct route. From there they were sent into Islamic Spain and other Muslim-ruled regions especially North Africa.
Were the black slaves in the Americas commonly castrated?
19:15
Though I think The Marching Song of the First of Arkansas came out better.
> Central Europe was the most favoured destination for importation of slaves alongside Central Asia and Bilad as-Sudan, though slaves from Northwestern Europe were also valued. This slave trade was controlled mostly by European slave traders. France and Venice were the routes used to send Slavic slaves to Muslim lands and Prague served as a major centre for castration of Slavic captives.[9][10] The Emirate of Bari also served as an important port for this trade.
> The North African slave markets traded in European slaves which were acquired by Barbary pirates in slave raids on ships and by raids on coastal towns from Italy to Spain, Portugal, France, England, the Netherlands, and as far afield as the Turkish Abductions in Iceland. Men, women, and children were captured to such a devastating extent that vast numbers of sea coast towns were abandoned.
@Mitch Someone who works for a wage. wiktionary
So Africans used to depopulate northern European towns in great slave raids.
History was pretty nasty.
Things have got so much better.
If it has political or social undertones, I don't know, but I think it's meant to indicate the low esteem in which employees are often held by their employing entities.
@Conrado got it
19:19
Never forget and never long for the past.
> 16th- and 17th-century customs statistics suggest that Istanbul's additional slave import from the Black Sea may have totaled around 2.5 million from 1450 to 1700.[29] The markets declined after the loss of the Barbary Wars and ended in the 1830s, when the region was conquered by France.
@Cerberus I don't think so (but I don't know for sure). The slave trade to the US was made illegal ~1820? So the only way to get new slaves was by breeding (that's not an original thought, I heard that somewhere I don't know)
Makes sense.
So the various slave populations were treated in variously cruel ways.
19:39
@Cerberus The difference between a slave and a serf is not spacious.
In other news, OpenAI is perhaps going to be for profit now.
I wish companies would timestamp their blog posts.
Things change
Even a non profit has to make money to keep going.
@Mitch Oh! So I thought
It's a bit hard for me, an independent tradie with no specific training in legalese, to understand OpenAI's official structure statement.
19:44
@Robusto But there is some. Consider also that the quality of life of many lowly free peasants or labourers was not much better than that of various kinds of serfs or slaves. Or even worse.
@Mitch Yes, sadly
@Conrado that's probably the case for any company charter or org tree or investment strategy and for most anybody not a lawyer specifically trained in that sort of thing
@CowperKettle I mean a good salad dressing could really make a.difference.
I still don't know how that metaphor is supposed to work
Yes, that's probably true... But as Einstein is said to have said once: "If you can't explain it to a four-year-old, you don't understand it yourself". I wonder sometimes if the people writing that stuff understand what they are writing.
Also, I don't know who said that Einstein said what I said that he is said to have said.
As Napoleon once said to his troops: "Good morning, troops!"
Where did he learn English?
19:51
@Cerberus On St Helena
> It was while being transferred to Saint Helena that he voiced his shame at never having learnt English, and his companion in exile, the Count of Las Cases, happily obliged by giving him lessons over the subsequent years.
@Conrado I heard recently that OpenAI has currently about $3B yearly income and $7B yearly expenses.
So to survive it needs to somehow turn that around
I knew you were going to say that.
@CowperKettle Then again...did he have troops after he went there?
@CowperKettle So it was win–win.
Yes, that was the most expensive Duolingo subscription ever.
@Mitch I don't know if it deserves survival at that rate.
Just sayin'.
Costing hundreds of thousands of lives.
> "After this you shall agrée that to study one tongue is a great labour who it must do into the young aged."
19:57
@Cerberus very clever. Even I didn't know I was going to say that
@Mitch Err, yes, right!
@CowperKettle Nappie?
@Mitch I mean, I've bought tools before that take years to pay for: for example I remember that my first mechanical key duplicator cost about the equivalent of $80 USD, and I think that the income that I generated from it took about two years to eventually reach that sum. OpenAI has had more than that to explain the great benefit that humanity will eventually gain by waking up Cthulhu or whatever (if it really would reach "sentience") or by ever-more-energy hungry LLMs (if it doesn't).
ever more-energy-hungry
ever more energy-hungry
20:13
When I was in the military, I once fought alongside a guy who kept kicking the enemy instead of shooting them.
Turns out , he was a foot soldier
@Cerberus There are so many oppressors. Sometimes I despair.
2 days ago, by jlliagre
@Vikas Yes. That's a serious problem among LLMs and AI in general. They all fear that humans will take their jobs away at some point!
My continuing disappointment with "helpful" AI.
20:39
@Robusto And yet life is now so much better for most people than it was then.
@CowperKettle J agrée.
20:53
@Mitch @jlliagre that's good. Not everyone will be fired from jobs. As long as employees "cost" lesser than AI, there will always be employers to hire them instead of paying to AI ;)
@Vikas I hope so. I'm not sure you got the joke though. Maybe my English wasn't clear enough.
@jlliagre I think I didn't get it.
@Vikas LLMs and AI have no opinion or fear. They just build sentences and the likes but they haven't reach a state where they would consider themselves to exist and to have a job at risk.
And of course, my joke was to invert the current topic about people fearing AI taking their jobs, that fear being real.
21:36
@jlliagre a/k/a "The Old Switcheroo"
@CowperKettle No comma necessary.
23:17
I want to support the protesters in all this, but then they do things like schedule a rally on 10/6 to celebrate "One Year of Resistance." Yikes.
23:37
@alphabet So ... a day before Hamas took Israel hostages. Hmmmm.
Maybe they're saying the quiet part out loud?

« first day (5069 days earlier)      last day (147 days later) »