Peter Swimm
@peterswimm.com
Founder @toilville
https://www.toilville.com/
https://www.toilville.com/
the seattle times is such a company town mouthpiece maybe they should get one of the companies to pay for it
November 23, 2025 at 5:31 PM
the seattle times is such a company town mouthpiece maybe they should get one of the companies to pay for it
it’s still wild to me that i grew up with this being my corner default hot dog place
November 23, 2025 at 5:04 PM
it’s still wild to me that i grew up with this being my corner default hot dog place
wow that’s two and half von trapps
November 23, 2025 at 4:50 PM
wow that’s two and half von trapps
the main reason it fails is what kind of psycho recommends software
November 23, 2025 at 3:19 PM
the main reason it fails is what kind of psycho recommends software
i’ve tried and failed to reframe it as a way to review specific user feedback in a cherry pick way but leadership will always gravitate to “why is the score 69”
November 23, 2025 at 3:19 PM
i’ve tried and failed to reframe it as a way to review specific user feedback in a cherry pick way but leadership will always gravitate to “why is the score 69”
Reposted by Peter Swimm
You might have heard about NPS. For those of you who don't, I'll briefly explain over the loud groans and boos from the rest of the audience.
NPS is this thing.
Everything about it is stupid but we're going to focus on one way in which it is stupid: it is a trailing indicator.
NPS is this thing.
Everything about it is stupid but we're going to focus on one way in which it is stupid: it is a trailing indicator.
May 30, 2025 at 3:41 PM
You might have heard about NPS. For those of you who don't, I'll briefly explain over the loud groans and boos from the rest of the audience.
NPS is this thing.
Everything about it is stupid but we're going to focus on one way in which it is stupid: it is a trailing indicator.
NPS is this thing.
Everything about it is stupid but we're going to focus on one way in which it is stupid: it is a trailing indicator.
would accept the b option
November 23, 2025 at 2:28 PM
would accept the b option
i found out that some llms use 600x the rest call size in code to describe an api to itself
November 23, 2025 at 5:09 AM
i found out that some llms use 600x the rest call size in code to describe an api to itself
Loved the memory-as-workspace framing.
One thing I’m still thinking about: does the model encode ‘negative knowledge’ (bad paths / dead-ends) in the same way as successful ones, or does the task distribution implicitly prune them?
Curious how you’re thinking about this for generalization.
One thing I’m still thinking about: does the model encode ‘negative knowledge’ (bad paths / dead-ends) in the same way as successful ones, or does the task distribution implicitly prune them?
Curious how you’re thinking about this for generalization.
November 22, 2025 at 11:43 PM
Loved the memory-as-workspace framing.
One thing I’m still thinking about: does the model encode ‘negative knowledge’ (bad paths / dead-ends) in the same way as successful ones, or does the task distribution implicitly prune them?
Curious how you’re thinking about this for generalization.
One thing I’m still thinking about: does the model encode ‘negative knowledge’ (bad paths / dead-ends) in the same way as successful ones, or does the task distribution implicitly prune them?
Curious how you’re thinking about this for generalization.
Reposted by Peter Swimm
now i’m convinced you are a psyop
November 22, 2025 at 9:08 PM
now i’m convinced you are a psyop
Don’t kink shame John
November 22, 2025 at 5:10 PM
Don’t kink shame John