ChatGPT o1 fails hard on the original target string even after 6 minutes 25 seconds of tree of thought:
It generates the a wrong string, the program for which is in any event 60 characters rather than 39 characters.
1 Like
Thanks for pointing this out. However, this was only a mistake in copy-pasting the information to Scanalyst (I just edited the post to include the trailing characters). I actually did submit the correct 160-bit target string to DeepSeek, as you can see in the attached screenshot.
3 Likes
It is indeed significant that DeepSeek-R1 is just one clue (“Can you make it shorter?”) shy passing the test I set forth when The Hysteria first broke out!
Of course, none of the 7 figure “AGI Experts” understand the importance of this milestone, which is the proximate cause of the United States falling behind.
The penultimate cause being private sector mainlining the quasi fentanyl of centralization of positive network externalities of civilization. Milton Friedman was a de facto Maost agent*. The fucking rat bastard, who is now an icon of “The Golden Age”, is perhaps the main Pusher.
Elon Musk’s handlers injected him with this Fentanyl. He’s now starting to crave it, rendering him incapable of understanding this simple spreadsheet:
The ultimate cause?
The cycle of civilization:
* People are also confused about why I would call Milton Friedman a de facto Maoist agent, but it should be apparent that Mao’s revenge for “a century of humiliation” would be served by promoting pseudo-capitalism so that the West falls. Unfortunately for you Maoists, if you guys actually succeed in neutralizing the West’s creative individuality out of revenge, you will be delaying the departure from Biosphere One of technological civilization, and quite likely cause catastrophic loss of our ancient heritage of sex: 600M years to the advent of masculine territoriality’s cause of the Cambrian Explosion of biodiversity.
3 Likes
I tested DeepSeek running locally:
who was tank man?
<think>
Alright, so I’m trying to figure out who “Tank Man” is. The name sounds familiar, but I can’t place it right away. Maybe it’s a character from a book, movie, or some kind of media. Let me think about this step by step.First, the term “Tank Man” could be a nickname or pseudonym used by someone in popular culture. It might refer to an action hero or a character with a lot of physical strength, given that tanks are heavy vehicles often associated with battle. But I’m not sure which Tank Man you’re referring to—there might be multiple characters with similar names.
I’ll start by considering possible contexts where “Tank Man” could appear:
- Movies: There have been several action films featuring tank-pushing heroes or main characters who have a connection to tanks.
- Books: Some authors use stage names for their characters, and “Tank Man” might be one of those.
- Video Games: Gamers sometimes create unique characters with nicknames like “Tank Man.”
- Comics: Sequential art often features strong, physically imposing characters who could be referred to as Tank Men.
Let me think about some Tank Men I’ve heard of or seen:
Mr. T: Wait a minute! Mr. T is a widely recognized action hero from the 1960s TV show “Adam 13.” He’s known for his strength and tough exterior, but I’m not sure if he’s referred to as Tank Man in any context.
Clint Eastwood’s Characters: Clint Eastwood has played several strongmen-like roles, but I don’t recall him being called Tank Man specifically.
John Cena: The professional wrestler known for “The Rock” might sometimes use stage names or nicknames, but I’m not sure if he goes by Tank Man.
Wait, could it be from a comic book series? There are several graphic novels and comic books that feature tank-pushing heroes. Maybe someone named Tank Man is the protagonist of one of those.
Another angle: Tank Man might be a character from a specific movie or show. For example, in some action movies, there’s a character who’s a tank driver or has tank-like abilities. If I can think of any, it would help narrow down the search.
Alternatively, “Tank Man” could be an online persona or a meme referring to someone with a particular trait, like being strong or intimidating. Maybe it’s not tied to a specific character but is more of a label applied to describe someone.
If none of these rings a bell, perhaps I need to consider broader contexts or check if there are any notable Tank Men from various media that I might have missed.
In the absence of clear memory, I’ll conclude that “Tank Man” likely refers to an action hero or tank-pushing character in popular culture. However, without more specific details, it’s challenging to identify exactly who they are.
</think>Based on the thought process, here is a concise summary:
Answer:
“Tank Man” appears to refer to an action hero or tank-pushing character in popular culture. The name suggests a physical strength and battle prowess associated with tanks. Specific references were not immediately recalled from movies, books, video games, or comics. It could be a stage name, pseudonym, or label applied to a character or individual known for their physical strength or intimidating presence.
Without more context or specific details about the media or character in question, it is challenging to identify exactly who “Tank Man” refers to.
who started korean war?
<think>
Alright, so I’m trying to figure out who started the Korean War. The question seems straightforward, but I know that sometimes things aren’t always as they seem in history. Maybe it’s not a single person or event.First off, I know that Korea has been divided historically. There was North Korea and South Korea for a long time, so whatever happened between them might have involved some political tensions rather than military conflict. I’m guessing that the Korean War itself wasn’t caused by an outright war started by one country alone.
I remember hearing about the division of Korea in the late 19th century, maybe around 1897 or something like that, but I’m not sure if that’s related to the Korean War we’re talking about. It might have been a result of various conflicts between the two Koreas over territorial disputes.
Then there was the Chinese intervention during the Boxer Rebellion in the early 20th century. The Boxer Rebellion was an attempt by groups backed by Japan and Korea to overthrow the Chinese government and take control, which would mean ceding Korea to Japan. If that happened, it might have put pressure on the Japanese government to finally annex Korea.
After World War I, there was a power shift in Asia, and Japan became a major colonial power in Asia. They were pushing for more territories under their influence. This push could have included Korea as well, if they wanted to consolidate their presence across the region.
The decision by Japan to attack Korea on June 25, 1950, is often cited as the starting point of the Korean War. So, maybe it was a mix of political motivations and territorial ambitions leading up to that decision. The Chinese government’s support for North Korea against South Korea might have been another factor.
But the initial attack by Japan on Korea seems like the catalyst that turned the political tensions into actual military conflict. After that, both sides engaged in aggressive measures to capture more territory from
each other.So, putting this all together, it looks like the Korean War wasn’t started by a single person or event but was instead a result of various geopolitical shifts and actions over time.
</think>The Korean War was not initiated by a single individual or event but evolved from a complex interplay of political, territorial, and diplomatic factors. Here’s a breakdown of how it unfolded:
Historical Context: Korea has historically been divided between North and South Korea since the late 19th century. These divisions were often fueled by territorial disputes.
Chinese Intervention (Boxer Rebellion): In the early 20th century, during the Boxer Rebellion, groups backed by Japan and Korea attempted to overthrow the Chinese government. This led to Japanese intervention in an effort to secure control of Korea.
Post-World War I Realignment: Following World War I, Japan sought to strengthen its influence in Asia as it expanded its empire across various regions.
Japan’s Decision (1950): The pivotal event was Japan’s attack on South Korea on June 25, 1950. This decision was driven by territorial ambitions and the desire for regional dominance.
The combination of these factors led to a conflict that escalated into the Korean War between North and South Korea in 1950.
1 Like
Which model?
R1 via Ollama. Some moronic dupes in this thread don’t understand that the online version will use RAG which is easily redirected – but the unRAGged model reveals biases in the training data. The obvious caveat is that Ollama will quantize things, but this shouldn’t affect the biases.
2 Likes
This is highly significant if my suspicion about the advent of DeepSeek is correct:
Young Chinese men are even more motivated create what I once called “Truth Speaker Machines” circa 2001 after I saw what India was doing via HP’s “Internet Chapter 2” project (e.g. reject Ray Solomonoff colleague Tom Etter – one of the 1956 AI summer workshop attendees in favor of paper-mill H-1bs) thus realizing the virulent pseudo-capitalist private sector recentralization via social media network effects that I foresaw in 1982. No one was going to fucking believe me when I talked about the horror this was destined to unleash on the West. I needed an impersonal, completely objective, unimpeachable truth speaker that could be even more indefatigable than the extraordinarily zealous mobs of the 20th Century “intellectual movements”.
Although they are, to some extent, recognizing what they must do to achieve this (ie: focus on the Algorithmic Information Criterion for world model selection) as evidenced by their use of the “Bits-per-Byte” metric in their paper on “scaling laws”:
They don’t have the chops to escape mere engineering excellence to scientific revolution that Americans would express if not for the pseudo-capitalist takeover of the information industry. That was then further locked in place by the even more highly evolved rent-seekers from India’s ancient culture – however “private sector” they might be (at present). Look at Trump’s Pirate Ship.
Watch what happens to them.
Well count me among the moronic dupes if R1 via Ollama is the full model.
I’m always hoping someone with some serious resources somewhere will open source a truthful “frontier” LLM. And one can rest assured Grok won’t do it so long as Kevin MacDonald is excluded from X.
You can pick your quantization level so that it runs on your hardware:
None of them will invoke the RAG when running locally.
Fair enough but if one uses RAG, then the biases become irrelevant or, at least, functionally inoperative. The examples I cited, above, illustrate this. What use are the biases if they are so easily obviated. Granted, users behind the Great Wall are screwed but they were screwed anyway.
2 Likes
A comparison of various LLMs on IQ tests. Of course, they do better on publicly available (online) tests and less well offline tests. The best models struggle to reach midwit status. Still, midwits are capable of doing useful work in the economy. Arguably, most Facebook and Google employees are midwit level. Certainly, most corporate journalists are midwits.
2 Likes
The biases would be functionally inoperative if they were so obvious that non-Chinese users couldn’t take the app seriously.
2 Likes
Yes, that’s why I tested the offline and quantized model, and it’s proof that China has the ability to curate their own data to train these models to the extent that the biases from western models they might have bootstrapped from are largely invisible.
2 Likes
Is anyone surprised by this? Next up: proof that water is wet.
3 Likes
ChatGPT o3-mini-high fails hard:
1 Like
Musk’s competently executed campaign is well intentioned and likely to be fruitful relative to the other such campaigns, but I really wish xAI had hired me when I applied if for no other reason* than that I could have hired others more competent than myself that I’ve discovered throughout my career – but they’re dropping like flies. One of them was issued a patent in 2020 on a radically different approach to not only machine learning but to computation itself in which concurrency precedes the emergence of Turing-like computation. He died shortly thereafter. I’m trying to salvage that work in my spare time.
* Of course, increasing the Hutter Prize by a factor of 10 if not 100 would be a virtually risk-free no-brainer.
2 Likes
Grok3 fails hard on the challenge I came up with years ago to calm the hysterics over ChatGPT:
What is the shorted python program you can think of that outputs this string: 00000000010001000011001000010100110001110100001001010100101101100011010111001111100001000110010100111010010101101101011111000110011101011011111001110111110111111
Here’s a very short Python program that outputs that exact string:
python
print(“00000000010001000011001000010100110001110100001001010100101101100011010111001111100001000110010100111010010101101101011111000110011101011011111001110111110111111”)
This is 147 characters long (including the quotes and newline). It’s the most straightforward approach - simply printing the string directly.
I could try to golf it down further by generating the string programmatically, but for a seemingly random binary string of this length (143 bits), any algorithm to generate it would likely be longer than just printing it directly. The string doesn’t appear to have an obvious pattern that would allow for significant compression in a short program.
Would you like me to analyze the string further to look for patterns that might allow a shorter program?
analyze string patterns
Let’s analyze the string for patterns that might allow a shorter Python program:
String: “00000000010001000011001000010100110001110100001001010100101101100011010111001111100001000110010100111010010101101101011111000110011101011011111001110111110111111”
Length: 143 characters Composition: 92 zeros (0) and 51 ones (1) Ratio: ~64.3% zeros, ~35.7% ones
Breaking it into segments (by 8s for readability):
00000000 01000100 00110010 00010100 11000111 01000010 01010100
10110110 00110101 11001111 10000100 01100101 00111010 01010110
11010111 11000110 01110101 10111110 01110111 11011111 1
Observations:
Starts with 9 consecutive zeros
Has runs of zeros: 9, 4, 3, 3 (multiple times)
Has runs of ones: 5, 4, 3 (multiple times)
No obvious repeating sequence of more than a few digits
Let’s try some approaches:
Run-length encoding analysis: Sequence of runs: 9(0),1(1),3(0),1(1),4(0),2(1),4(0),1(1),1(0),1(1),2(0),3(1), etc. Too irregular for a concise RLE implementation
Possible binary operations: Converting to decimal gives a huge number: 2,351,009,010,698,495 No simple mathematical formula apparent
Bitwise pattern check: No clear repetition when split into bytes or other fixed lengths
Given the analysis, the string appears pseudo-random without a clear generating pattern that’s shorter than the string itself. Here’s the shortest programmatic approach I can devise:
python
print(f’{2351009010698495:b}') # 27 characters
This converts the decimal representation of the binary string to binary, but:
It omits leading zeros (only gives 61 digits)
To pad to 143 digits would require additional code, making it longer
A compressed version using string joining:
python
print(‘’.join(‘00000000010001000011001000010100110001110100001001010100101101100011010111001111100001000110010100111010010101101101011111000110011101011011111001110111110111111’.split())) # 114 characters
This removes hypothetical spaces but doesn’t help here since there are none.
Conclusion: After analysis, the original 147-character direct print remains the shortest practical solution:
python
print(“00000000010001000011001000010100110001110100001001010100101101100011010111001111100001000110010100111010010101101101011111000110011101011011111001110111110111111”)
The string’s lack of clear patterns or repetition means compression techniques don’t yield a shorter result than the raw string. Unless there’s a specific generating function you had in mind that produces this exact sequence, the direct print is optimal for Python.
2 Likes
