Han Seoul-Oh
@laughinghan.com
Let’s build tools to augment the mind, not consume it.
Creator of http://MathQuill.com
Now working on some programming language and structured editor ideas.
https://github.com/laughinghan
Creator of http://MathQuill.com
Now working on some programming language and structured editor ideas.
https://github.com/laughinghan
Please explain the joke:
April 5, 2025 at 7:14 AM
Please explain the joke:
Did you really just cite a 7-paragraph Medium post, itself with zero citations, written by a rando whose Shift key doesn’t work?
Your misunderstandings are so basic that an LLM explains them easily, I suggest you go talk to one:
Your misunderstandings are so basic that an LLM explains them easily, I suggest you go talk to one:
April 4, 2025 at 12:10 PM
Did you really just cite a 7-paragraph Medium post, itself with zero citations, written by a rando whose Shift key doesn’t work?
Your misunderstandings are so basic that an LLM explains them easily, I suggest you go talk to one:
Your misunderstandings are so basic that an LLM explains them easily, I suggest you go talk to one:
No, that’s backwards. Abundance is the one claiming its policy approach will be popular enough to build a brand new coalition. Stancil’s whole thing is that policy always loses to propaganda, that’s how Trump won. Trump obvs doesn’t wait for things to be popular to do them!
March 22, 2025 at 8:29 AM
No, that’s backwards. Abundance is the one claiming its policy approach will be popular enough to build a brand new coalition. Stancil’s whole thing is that policy always loses to propaganda, that’s how Trump won. Trump obvs doesn’t wait for things to be popular to do them!
Hot take: if this is what “do your own research” means now, this may be a good thing? Occasional confabulations are surely far less dangerous than going down YouTube conspiracy rabbit holes
Even Grok is pretty adamant that vaccines don’t cause autism: grok.com/share/bGVnYW...
Even Grok is pretty adamant that vaccines don’t cause autism: grok.com/share/bGVnYW...
March 8, 2025 at 9:58 AM
Hot take: if this is what “do your own research” means now, this may be a good thing? Occasional confabulations are surely far less dangerous than going down YouTube conspiracy rabbit holes
Even Grok is pretty adamant that vaccines don’t cause autism: grok.com/share/bGVnYW...
Even Grok is pretty adamant that vaccines don’t cause autism: grok.com/share/bGVnYW...
> Not making anything up.
What do you think the word “Imagine” means?
In reality:
> low income NYers don’t have cars
> car owners richer than those without cars
> congestion pricing projected to help working poor 38-to-1
What do you think the word “Imagine” means?
In reality:
> low income NYers don’t have cars
> car owners richer than those without cars
> congestion pricing projected to help working poor 38-to-1
January 6, 2025 at 9:10 PM
> Not making anything up.
What do you think the word “Imagine” means?
In reality:
> low income NYers don’t have cars
> car owners richer than those without cars
> congestion pricing projected to help working poor 38-to-1
What do you think the word “Imagine” means?
In reality:
> low income NYers don’t have cars
> car owners richer than those without cars
> congestion pricing projected to help working poor 38-to-1
“Often” is meaningless and misleading, in reality people who drive private vehicles costing tens of thousands of dollars are (of course!) less poor than people who ride the bus.
As a result, congestion pricing is highly progressive: studies project it will help the working poor 38-to-1.
As a result, congestion pricing is highly progressive: studies project it will help the working poor 38-to-1.
January 6, 2025 at 9:04 PM
“Often” is meaningless and misleading, in reality people who drive private vehicles costing tens of thousands of dollars are (of course!) less poor than people who ride the bus.
As a result, congestion pricing is highly progressive: studies project it will help the working poor 38-to-1.
As a result, congestion pricing is highly progressive: studies project it will help the working poor 38-to-1.
This makes it sound like public transit is for rich downtowners and cars are for poor commuters.
That’s false. In reality, people driving private vehicles costing tens of thousands of dollars are richer than people on the bus. (Duh!) That’s why congestion pricing helps the working poor 38-to-1.
That’s false. In reality, people driving private vehicles costing tens of thousands of dollars are richer than people on the bus. (Duh!) That’s why congestion pricing helps the working poor 38-to-1.
January 6, 2025 at 8:48 PM
This makes it sound like public transit is for rich downtowners and cars are for poor commuters.
That’s false. In reality, people driving private vehicles costing tens of thousands of dollars are richer than people on the bus. (Duh!) That’s why congestion pricing helps the working poor 38-to-1.
That’s false. In reality, people driving private vehicles costing tens of thousands of dollars are richer than people on the bus. (Duh!) That’s why congestion pricing helps the working poor 38-to-1.
It’s true public transit feels more costly to the poor than rich, but private vehicles costing tens of thousands of dollars are obviously far more so, so this is all nonsense.
In reality, poor NYers overwhelmingly don’t have cars, so studies projected that congestion pricing will help them 38-to-1:
In reality, poor NYers overwhelmingly don’t have cars, so studies projected that congestion pricing will help them 38-to-1:
January 6, 2025 at 8:37 PM
It’s true public transit feels more costly to the poor than rich, but private vehicles costing tens of thousands of dollars are obviously far more so, so this is all nonsense.
In reality, poor NYers overwhelmingly don’t have cars, so studies projected that congestion pricing will help them 38-to-1:
In reality, poor NYers overwhelmingly don’t have cars, so studies projected that congestion pricing will help them 38-to-1:
January 3, 2025 at 11:03 PM
Remember that time someone shitposted that they created Emily in Paris “as a prank” and some real (bottom-barrel, but real) news outlets published it as real
www.cracked.com/article_2956...
www.cracked.com/article_2956...
December 21, 2024 at 11:52 AM
Remember that time someone shitposted that they created Emily in Paris “as a prank” and some real (bottom-barrel, but real) news outlets published it as real
www.cracked.com/article_2956...
www.cracked.com/article_2956...
Its first try is exactly what I would’ve guessed, and I think most people. I think I (and most people? Maybe?) would’ve gotten the intended answer on the second try, but if the guy next to me tried its Attempt 2, I wouldn’t have immediately questioned their intelligence.
December 21, 2024 at 8:43 AM
Its first try is exactly what I would’ve guessed, and I think most people. I think I (and most people? Maybe?) would’ve gotten the intended answer on the second try, but if the guy next to me tried its Attempt 2, I wouldn’t have immediately questioned their intelligence.
The mistakes o3 makes are also really interesting. Chollet highlights this one. Note that there’s some ambiguity; apparently this is by design, the challenge is to get the intended answer in 2 tries.
December 21, 2024 at 8:43 AM
The mistakes o3 makes are also really interesting. Chollet highlights this one. Note that there’s some ambiguity; apparently this is by design, the challenge is to get the intended answer in 2 tries.
This is actually kind of a big deal. Chollet has long been a skeptic of the idea of LLMs being able to reason, and put his money where his mouth is by creating ARC to prove it.
Tho note:
• they spent >$1k per task (inference, not pre-training)
• this is ARCv1; ARCv2 is apparently harder for o3
Tho note:
• they spent >$1k per task (inference, not pre-training)
• this is ARCv1; ARCv2 is apparently harder for o3
December 21, 2024 at 8:35 AM
This is actually kind of a big deal. Chollet has long been a skeptic of the idea of LLMs being able to reason, and put his money where his mouth is by creating ARC to prove it.
Tho note:
• they spent >$1k per task (inference, not pre-training)
• this is ARCv1; ARCv2 is apparently harder for o3
Tho note:
• they spent >$1k per task (inference, not pre-training)
• this is ARCv1; ARCv2 is apparently harder for o3
A friend pointed out the Southern Cross Cable from 2000
December 18, 2024 at 5:31 AM
A friend pointed out the Southern Cross Cable from 2000
Based on the number of ping times up and down the US West Coast that are >70% the speed of light, there appear to be significantly more direct undersea cables between NZ and the US than I would’ve expected
wondernetwork.com/pings/Auckland
Maybe it’s a common transit point for AUS<->USA?
wondernetwork.com/pings/Auckland
Maybe it’s a common transit point for AUS<->USA?
December 18, 2024 at 5:31 AM
Based on the number of ping times up and down the US West Coast that are >70% the speed of light, there appear to be significantly more direct undersea cables between NZ and the US than I would’ve expected
wondernetwork.com/pings/Auckland
Maybe it’s a common transit point for AUS<->USA?
wondernetwork.com/pings/Auckland
Maybe it’s a common transit point for AUS<->USA?
Do people think the media should’ve kept mum on scientific concerns with trans fat in the years before they were proven to be harmful? Ofc not
But RFK Jr has polarized people into attacking anything negative on seed oils, even valuable reporting on scientific concerns with ultraprocessed food
But RFK Jr has polarized people into attacking anything negative on seed oils, even valuable reporting on scientific concerns with ultraprocessed food
December 16, 2024 at 9:21 PM
Do people think the media should’ve kept mum on scientific concerns with trans fat in the years before they were proven to be harmful? Ofc not
But RFK Jr has polarized people into attacking anything negative on seed oils, even valuable reporting on scientific concerns with ultraprocessed food
But RFK Jr has polarized people into attacking anything negative on seed oils, even valuable reporting on scientific concerns with ultraprocessed food
Really clever and believable clickjacking attack found in the wild, note they went out of their way with their wording to prep you to click the “Verify” button in the security prompt.
“Transient user activation” has always been totally worthless as a security measure, it’s so utterly broken that
“Transient user activation” has always been totally worthless as a security measure, it’s so utterly broken that
December 12, 2024 at 5:38 AM
Really clever and believable clickjacking attack found in the wild, note they went out of their way with their wording to prep you to click the “Verify” button in the security prompt.
“Transient user activation” has always been totally worthless as a security measure, it’s so utterly broken that
“Transient user activation” has always been totally worthless as a security measure, it’s so utterly broken that
November 19, 2024 at 12:30 PM
In 1989, AIDS was the #1 cause of death among men aged 25-44 in NYC, and #2 nationally, “surpassing heart disease, cancer, suicide, and homicide”
The freakout was not needless
The freakout was not needless
November 19, 2024 at 12:30 PM
In 1989, AIDS was the #1 cause of death among men aged 25-44 in NYC, and #2 nationally, “surpassing heart disease, cancer, suicide, and homicide”
The freakout was not needless
The freakout was not needless
TIL: In 2019 RFK Jr visited Samoa to promote vaccine misinformation, where newborn vaccination rates had been falling to a low of 31%. Later that year they had a massive measles outbreak killing 83 children
newrepublic.com/article/1877...
newrepublic.com/article/1877...
November 19, 2024 at 11:30 AM
TIL: In 2019 RFK Jr visited Samoa to promote vaccine misinformation, where newborn vaccination rates had been falling to a low of 31%. Later that year they had a massive measles outbreak killing 83 children
newrepublic.com/article/1877...
newrepublic.com/article/1877...
Sounds like a lie intended to elicit confession tbh
November 18, 2024 at 7:46 AM
Sounds like a lie intended to elicit confession tbh
I did not claim they were peer reviewed. I am claiming they released enough details that they can be evaluated on substance rather than credentials
But here you go, another peer-reviewed article: www.pnas.org/doi/10.1073/...
All I’m saying is there’s no consensus. I easily meet that low bar.
But here you go, another peer-reviewed article: www.pnas.org/doi/10.1073/...
All I’m saying is there’s no consensus. I easily meet that low bar.
February 13, 2024 at 10:31 PM
I did not claim they were peer reviewed. I am claiming they released enough details that they can be evaluated on substance rather than credentials
But here you go, another peer-reviewed article: www.pnas.org/doi/10.1073/...
All I’m saying is there’s no consensus. I easily meet that low bar.
But here you go, another peer-reviewed article: www.pnas.org/doi/10.1073/...
All I’m saying is there’s no consensus. I easily meet that low bar.
No, multiple studies have found a conservative bias or radicalization on YouTube:
• dl.acm.org/doi/10.1145/...
• www.theguardian.com/technology/2...
• www.brookings.edu/articles/ech...
• www.bellingcat.com/news/america...
I also found 2 papers saying the opposite, but you misstate the consensus.
• dl.acm.org/doi/10.1145/...
• www.theguardian.com/technology/2...
• www.brookings.edu/articles/ech...
• www.bellingcat.com/news/america...
I also found 2 papers saying the opposite, but you misstate the consensus.
February 13, 2024 at 9:19 PM
No, multiple studies have found a conservative bias or radicalization on YouTube:
• dl.acm.org/doi/10.1145/...
• www.theguardian.com/technology/2...
• www.brookings.edu/articles/ech...
• www.bellingcat.com/news/america...
I also found 2 papers saying the opposite, but you misstate the consensus.
• dl.acm.org/doi/10.1145/...
• www.theguardian.com/technology/2...
• www.brookings.edu/articles/ech...
• www.bellingcat.com/news/america...
I also found 2 papers saying the opposite, but you misstate the consensus.