Querydash
@querydash.bsky.social
NYC person. UX and behavioral researcher.
Pinned
Querydash
@querydash.bsky.social
· Feb 4
Putting this XKCD comic here for OH I DON’T KNOW, NO REASON.
Reposted by Querydash
New from the @nytimes.com Visual Investigations team: Detailed, second-by-second video analysis of all of the video footage of the killing of Renee Good in Minneapolis.
www.nytimes.com/video/us/100...
www.nytimes.com/video/us/100...
Video: Video Analysis of ICE Shooting Sheds Light on Contested Moments
Newly available videos and existing footage synchronized and assessed by The Times provides a millisecond-by-millisecond look at how an ICE officer ended up shooting and killing a motorist in Minneapo...
www.nytimes.com
January 16, 2026 at 4:11 AM
New from the @nytimes.com Visual Investigations team: Detailed, second-by-second video analysis of all of the video footage of the killing of Renee Good in Minneapolis.
www.nytimes.com/video/us/100...
www.nytimes.com/video/us/100...
When someone in a crowd yells “pedophile protector” and you instantly think “hey that’s ME, I better respond!”
…well, that seems problematic for you.
…well, that seems problematic for you.
January 13, 2026 at 10:07 PM
When someone in a crowd yells “pedophile protector” and you instantly think “hey that’s ME, I better respond!”
…well, that seems problematic for you.
…well, that seems problematic for you.
It’s boring.
And it made me think about the after-effects of intimate contact with burlap, which I resent.
And it made me think about the after-effects of intimate contact with burlap, which I resent.
January 13, 2026 at 6:29 PM
It’s boring.
And it made me think about the after-effects of intimate contact with burlap, which I resent.
And it made me think about the after-effects of intimate contact with burlap, which I resent.
I couldn’t fall asleep last night because I didn’t want to wake up and discover we’d invaded Greenland.
Most of the time it’s safe to sleep on weekends since no one’s working, but in this era, it’s the most fraught 48-period because HE DOES CRAZY SHIT WHEN EVERYONE IS ASLEEP. He prefers weekends!
Most of the time it’s safe to sleep on weekends since no one’s working, but in this era, it’s the most fraught 48-period because HE DOES CRAZY SHIT WHEN EVERYONE IS ASLEEP. He prefers weekends!
January 12, 2026 at 2:50 AM
I couldn’t fall asleep last night because I didn’t want to wake up and discover we’d invaded Greenland.
Most of the time it’s safe to sleep on weekends since no one’s working, but in this era, it’s the most fraught 48-period because HE DOES CRAZY SHIT WHEN EVERYONE IS ASLEEP. He prefers weekends!
Most of the time it’s safe to sleep on weekends since no one’s working, but in this era, it’s the most fraught 48-period because HE DOES CRAZY SHIT WHEN EVERYONE IS ASLEEP. He prefers weekends!
I’d hew closer to “it’s not necessarily ready for some of the applications it’s used for, it absolutely does NOT need to be implemented in the ways companies are enforcing usage, and it has worrying implications for jr engineers building mental models” but agree that blanket anti-AI isn’t useful.
January 11, 2026 at 6:18 PM
I’d hew closer to “it’s not necessarily ready for some of the applications it’s used for, it absolutely does NOT need to be implemented in the ways companies are enforcing usage, and it has worrying implications for jr engineers building mental models” but agree that blanket anti-AI isn’t useful.
Look, we all wish Greg the best of luck at his new gig.
January 11, 2026 at 6:14 PM
Look, we all wish Greg the best of luck at his new gig.
Ah, fair clarification. Off the cuff posts are different.
Then I guess I’m bewildered by the “no one experiences anymore” claim. Because it feels like the OP wants to stand on the idea that incremental (though rapid) improvement somehow invalidates ALL ongoing, real, though shrinking errors.
Then I guess I’m bewildered by the “no one experiences anymore” claim. Because it feels like the OP wants to stand on the idea that incremental (though rapid) improvement somehow invalidates ALL ongoing, real, though shrinking errors.
January 11, 2026 at 6:09 PM
Ah, fair clarification. Off the cuff posts are different.
Then I guess I’m bewildered by the “no one experiences anymore” claim. Because it feels like the OP wants to stand on the idea that incremental (though rapid) improvement somehow invalidates ALL ongoing, real, though shrinking errors.
Then I guess I’m bewildered by the “no one experiences anymore” claim. Because it feels like the OP wants to stand on the idea that incremental (though rapid) improvement somehow invalidates ALL ongoing, real, though shrinking errors.
But that feels like a mismatch: academia has pretty long cycles. AI has notoriously short ones. It’s unlikely an academic paper is going to reflect AI changes that occurred recently, and it’s hard to incorporate hypothetical performance to satisfy an ask that the paper be more current. /x
January 11, 2026 at 6:05 PM
But that feels like a mismatch: academia has pretty long cycles. AI has notoriously short ones. It’s unlikely an academic paper is going to reflect AI changes that occurred recently, and it’s hard to incorporate hypothetical performance to satisfy an ask that the paper be more current. /x
So I’d assume the OP’s complaint is that academics are writing papers about specific examples of AI where the extent of the dataset and additives are known, but that the academic conclusions should also reflect the hypothetical impacts of AI advancements since the focus event? /1
January 11, 2026 at 6:03 PM
So I’d assume the OP’s complaint is that academics are writing papers about specific examples of AI where the extent of the dataset and additives are known, but that the academic conclusions should also reflect the hypothetical impacts of AI advancements since the focus event? /1
Can you expand on this - what are LLMs incorporating outside of their dataset?
January 11, 2026 at 5:52 PM
Can you expand on this - what are LLMs incorporating outside of their dataset?
A lot of my research has been on the tech side and what spooks me is that the prolific use of AI in the junior ranks is currently being failsafed by senior engineers who spot errors out of muscle memory, but the juniors aren’t developing that reflex.
And I think that’s happening across the board.
And I think that’s happening across the board.
January 11, 2026 at 5:43 PM
A lot of my research has been on the tech side and what spooks me is that the prolific use of AI in the junior ranks is currently being failsafed by senior engineers who spot errors out of muscle memory, but the juniors aren’t developing that reflex.
And I think that’s happening across the board.
And I think that’s happening across the board.
I’m actually more impressed by AI performance in your field than in others, but I’m also aware that AI errors are caught relatively quickly by practitioners because their extensive experience makes it obvious when something is off. They view AI as an eager but sometimes ignorant intern.
January 11, 2026 at 5:37 PM
I’m actually more impressed by AI performance in your field than in others, but I’m also aware that AI errors are caught relatively quickly by practitioners because their extensive experience makes it obvious when something is off. They view AI as an eager but sometimes ignorant intern.
And that’s where I’m puzzled by the original post - are the academics talking about something that truly doesn’t happen? Or are they talking about something that DOES happen, and has significant downstream impacts, but the OP believes the rate of AI improvement for some reason invalidates the point?
January 11, 2026 at 5:35 PM
And that’s where I’m puzzled by the original post - are the academics talking about something that truly doesn’t happen? Or are they talking about something that DOES happen, and has significant downstream impacts, but the OP believes the rate of AI improvement for some reason invalidates the point?
So I’d have two questions here:
1. Is ChatGPT Pro NEVER hallucinating, or does it just do it a lot less?
2. Are the academic papers referencing stuff like hallucinated precedents in legal filings, which would price in some lapse between occurrence and academic analysis of real-world outcomes?
1. Is ChatGPT Pro NEVER hallucinating, or does it just do it a lot less?
2. Are the academic papers referencing stuff like hallucinated precedents in legal filings, which would price in some lapse between occurrence and academic analysis of real-world outcomes?
January 11, 2026 at 5:28 PM
So I’d have two questions here:
1. Is ChatGPT Pro NEVER hallucinating, or does it just do it a lot less?
2. Are the academic papers referencing stuff like hallucinated precedents in legal filings, which would price in some lapse between occurrence and academic analysis of real-world outcomes?
1. Is ChatGPT Pro NEVER hallucinating, or does it just do it a lot less?
2. Are the academic papers referencing stuff like hallucinated precedents in legal filings, which would price in some lapse between occurrence and academic analysis of real-world outcomes?
What would academics be highlighting as a problem using the 7-9 fingers example?
January 11, 2026 at 5:24 PM
What would academics be highlighting as a problem using the 7-9 fingers example?
… this is not a gotcha question, BTW. I’m a UX researcher working steadily in the AI space.
January 11, 2026 at 5:21 PM
… this is not a gotcha question, BTW. I’m a UX researcher working steadily in the AI space.
What AI things are you referring to, that “no one experiences anymore”?
January 11, 2026 at 5:18 PM
What AI things are you referring to, that “no one experiences anymore”?
What do you want to bet he’s on the “synthetic user” train. Why bother with actual humans who are slow and messy and give difficult to analyze responses when you can fire rapid questions at a digital human persona who doesn’t require an honorarium for their time?
January 11, 2026 at 2:52 PM
What do you want to bet he’s on the “synthetic user” train. Why bother with actual humans who are slow and messy and give difficult to analyze responses when you can fire rapid questions at a digital human persona who doesn’t require an honorarium for their time?
The last decade’s been a horrifying UX research slide - I agree with your definition of “interview”, and also know this galoot thinks any contact with an end-user that generates unstructured response counts.
And let’s not forget SYNTHETIC USERS. Tons of bullshit interviews with them per hour!
And let’s not forget SYNTHETIC USERS. Tons of bullshit interviews with them per hour!
January 11, 2026 at 2:49 PM
The last decade’s been a horrifying UX research slide - I agree with your definition of “interview”, and also know this galoot thinks any contact with an end-user that generates unstructured response counts.
And let’s not forget SYNTHETIC USERS. Tons of bullshit interviews with them per hour!
And let’s not forget SYNTHETIC USERS. Tons of bullshit interviews with them per hour!
I’m genuinely having a hard time falling asleep because I don’t want to wake up to news that we’ve invaded Greenland and are at war with NATO, which sounds nuts but already happened last week with Venezuela. So.
They’re going to have to create a whole new classification of insomnia for this era.
They’re going to have to create a whole new classification of insomnia for this era.
January 11, 2026 at 6:04 AM
I’m genuinely having a hard time falling asleep because I don’t want to wake up to news that we’ve invaded Greenland and are at war with NATO, which sounds nuts but already happened last week with Venezuela. So.
They’re going to have to create a whole new classification of insomnia for this era.
They’re going to have to create a whole new classification of insomnia for this era.
Pick the romance books. You can rent romance books. There is no library for yarn or, god help you, a loom.
January 10, 2026 at 5:08 PM
Pick the romance books. You can rent romance books. There is no library for yarn or, god help you, a loom.
“You have a collect call from….MOMI’MATTHEFRIENDLY’SCOMEGETMEANDCANYOUDRIVEHEATHERHOMETOOSHELIVESONCRANBER….
Do you accept the charges?”
Do you accept the charges?”
Without revealing your actual age,what's something you remember that if you told a younger person they wouldn't understand?
January 10, 2026 at 5:01 AM
“You have a collect call from….MOMI’MATTHEFRIENDLY’SCOMEGETMEANDCANYOUDRIVEHEATHERHOMETOOSHELIVESONCRANBER….
Do you accept the charges?”
Do you accept the charges?”
Reposted by Querydash
So @hcrichardson.bsky.social has a post reviewing WWII War Department pamphlets teaching troops about how to spot and resist fascism. They seem awfully relevant. Give it a read, but mind Heather doesn’t steal your wife.
heathercoxrichardson.substack.com/p/january-9-...
heathercoxrichardson.substack.com/p/january-9-...
January 9, 2026
Beginning in 1943, the War Department published a series of pamphlets for U.S.
heathercoxrichardson.substack.com
January 10, 2026 at 3:40 AM
So @hcrichardson.bsky.social has a post reviewing WWII War Department pamphlets teaching troops about how to spot and resist fascism. They seem awfully relevant. Give it a read, but mind Heather doesn’t steal your wife.
heathercoxrichardson.substack.com/p/january-9-...
heathercoxrichardson.substack.com/p/january-9-...
Reposted by Querydash
Jesse Welles doing that thing he's really good at
January 8, 2026 at 5:16 PM
Jesse Welles doing that thing he's really good at