How they're measuring ROI I can't say, but they do seem happy with the rate of codebase progress.
September 13, 2026 at 3:42 PM
How they're measuring ROI I can't say, but they do seem happy with the rate of codebase progress.
That makes sense. We're a large org and already have doc teams, travel budgets, etc. Management is always on the lookout for additional aids to productivity. With LLM's they've come to see a benefit, but they did go through a cost shock when they were deployed to initial teams.
September 13, 2026 at 3:41 PM
That makes sense. We're a large org and already have doc teams, travel budgets, etc. Management is always on the lookout for additional aids to productivity. With LLM's they've come to see a benefit, but they did go through a cost shock when they were deployed to initial teams.
We ran into budget constraints initially, but management has been flexible since we were able to demonstrate better results. Still I'm jealous of one person I know whose personal token allowance is $10k per month!
September 13, 2026 at 1:56 PM
We ran into budget constraints initially, but management has been flexible since we were able to demonstrate better results. Still I'm jealous of one person I know whose personal token allowance is $10k per month!
I'm at 48 years, but I find that I'm more excited than I have been in quite a while. I certainly feel more productive. Curiously the amount of code at the end of the day hasn't increased much. But I try out many more ideas and am happy to have a "colleague" to debate with.
September 13, 2026 at 11:32 AM
I'm at 48 years, but I find that I'm more excited than I have been in quite a while. I certainly feel more productive. Curiously the amount of code at the end of the day hasn't increased much. But I try out many more ideas and am happy to have a "colleague" to debate with.
Welp, guess we're doomed now...
September 7, 2026 at 11:55 AM
Welp, guess we're doomed now...
C2 to C3. I'm a bit surprised to be there because I'm an old dog, but apparently I can learn a few new tricks.
May 16, 2026 at 12:57 PM
C2 to C3. I'm a bit surprised to be there because I'm an old dog, but apparently I can learn a few new tricks.
My view as a programmer is that it has and will continue to change the nature of the job in a commercial setting, just as high-level languages did. I program by hand when I'm doing it recreationally. I use agents heavily in my job where I'm being paid to be productive.
April 25, 2026 at 3:13 PM
My view as a programmer is that it has and will continue to change the nature of the job in a commercial setting, just as high-level languages did. I program by hand when I'm doing it recreationally. I use agents heavily in my job where I'm being paid to be productive.
I feel sorry for people who listened to that and forced themselves through a CS program only because they thought they could make money. The ones who love it though are in a fantastic position.
April 14, 2026 at 2:58 PM
I feel sorry for people who listened to that and forced themselves through a CS program only because they thought they could make money. The ones who love it though are in a fantastic position.
A bit similar in spirit to gift wrapping perhaps. en.wikipedia.org/wiki/Gift_wr...
Gift wrapping algorithm - Wikipedia
en.wikipedia.org
April 4, 2026 at 4:39 PM
A bit similar in spirit to gift wrapping perhaps. en.wikipedia.org/wiki/Gift_wr...
To be fair, it will also be very useful for scientific and high performance computing. I work on an EM simulator and am looking forward to the higher memory bandwidth compared to the Blackwell GPUs. AI gets all the headlines, but simulation has really benefited too.
February 22, 2026 at 2:28 PM
To be fair, it will also be very useful for scientific and high performance computing. I work on an EM simulator and am looking forward to the higher memory bandwidth compared to the Blackwell GPUs. AI gets all the headlines, but simulation has really benefited too.
I'm pretty boring to anyone who's very smart. To me an LLM is a bit like writing in a notebook. I'm not trying to get it to think for me; I want to get my thinking into a clearer form. Writing alone helps, but it's useful when my pen asks good questions, even if there's no real thinking behind them.
December 28, 2025 at 1:00 PM
I'm pretty boring to anyone who's very smart. To me an LLM is a bit like writing in a notebook. I'm not trying to get it to think for me; I want to get my thinking into a clearer form. Writing alone helps, but it's useful when my pen asks good questions, even if there's no real thinking behind them.
I think I saw that episode of Star Trek
December 4, 2025 at 12:46 AM
I think I saw that episode of Star Trek
That's really cool! What was causing the problem before where the translation was off, if it's not too detailed to explain?
November 26, 2025 at 12:32 PM
That's really cool! What was causing the problem before where the translation was off, if it's not too detailed to explain?
When using the framework do pure rotations work, and do pure translations work? If one of those is broken it might at least simplify figuring out where the problem is... Good luck!
November 25, 2025 at 10:14 AM
When using the framework do pure rotations work, and do pure translations work? If one of those is broken it might at least simplify figuring out where the problem is... Good luck!
She wrote code, so I misunderstood
November 25, 2025 at 3:06 AM
She wrote code, so I misunderstood
Coding, not math. Or at least I never tried them for anything math-related.
November 25, 2025 at 2:50 AM
Coding, not math. Or at least I never tried them for anything math-related.
I don't know. I was dismissive last year. But recent agent-based flows make me feel like there's something there, and I don't think I'm alone.
November 24, 2025 at 7:58 PM
I don't know. I was dismissive last year. But recent agent-based flows make me feel like there's something there, and I don't think I'm alone.
For people like me who treat both platforms as effectively a read-only news feed (almost no posting, don't read replies to whatever others post), twitter is much better. I follow a few people on bsky and maybe 10x more on twitter. I only use bsky due to a few such as yourself who moved.
September 17, 2025 at 11:17 PM
For people like me who treat both platforms as effectively a read-only news feed (almost no posting, don't read replies to whatever others post), twitter is much better. I follow a few people on bsky and maybe 10x more on twitter. I only use bsky due to a few such as yourself who moved.
I just realized that this is what we kids used to pick and eat during the summer when visiting my grandparents. We just called them grapes, but I always wondered why the normal grapes were nothing like them.
August 12, 2025 at 8:03 PM
I just realized that this is what we kids used to pick and eat during the summer when visiting my grandparents. We just called them grapes, but I always wondered why the normal grapes were nothing like them.
Someone told me the ending was a bit subtle. I thought I remembered it and had no idea what they meant. But when I went back and read the *whole* book, I understood.
March 28, 2025 at 11:56 AM
Someone told me the ending was a bit subtle. I thought I remembered it and had no idea what they meant. But when I went back and read the *whole* book, I understood.
Right up there with the alien face hugger
March 23, 2025 at 8:07 PM
Right up there with the alien face hugger
But if I knew a spatial slice at x0 for all times with the same amount of info (number of rhs), I think I could solve for all other x in an analogous manner. That's probably clear as mud, but that's my current intuition.
March 11, 2025 at 4:34 PM
But if I knew a spatial slice at x0 for all times with the same amount of info (number of rhs), I think I could solve for all other x in an analogous manner. That's probably clear as mud, but that's my current intuition.
In that form, the distinction seems irrelevant. It appears you'd need exactly the same amount of info no matter what. So while we do perceive time as special, I suspect it doesn't carry more info. We think about time slices because we usually specify "now" and want to know what will happen. 3/
March 11, 2025 at 4:29 PM
In that form, the distinction seems irrelevant. It appears you'd need exactly the same amount of info no matter what. So while we do perceive time as special, I suspect it doesn't carry more info. We think about time slices because we usually specify "now" and want to know what will happen. 3/
But if I expand the linear system to include trivial equations for the boundary conditions, I could instead write the problem as finding a unique u such that Au = 0. The u has everything related to both boundary conditions and "outputs". 2/
March 11, 2025 at 4:23 PM
But if I expand the linear system to include trivial equations for the boundary conditions, I could instead write the problem as finding a unique u such that Au = 0. The u has everything related to both boundary conditions and "outputs". 2/