iamwil
@interjectedfuture.com
Tech Zine Issue 1: LLM System Eval https://forestfriends.tech
Local-first/Reactive Programming ⁙ LLM system evals ⁙ Startup lessons ⁙ Game design quips.
Longform: https://interjectedfuture.com
Podcast: https://www.youtube.com/@techniumpod
Local-first/Reactive Programming ⁙ LLM system evals ⁙ Startup lessons ⁙ Game design quips.
Longform: https://interjectedfuture.com
Podcast: https://www.youtube.com/@techniumpod
Pinned
iamwil
@interjectedfuture.com
· Mar 22
We already have digital necromancy in the form of Tupac holding concerts every few years.
The cocktail party question, "Who would you want to have dinner with, alive or dead?" would be wish fulfillment.
Heck, with AR, the dead can accompany you to the cocktail party itself.
x.com/vanschneide...
The cocktail party question, "Who would you want to have dinner with, alive or dead?" would be wish fulfillment.
Heck, with AR, the dead can accompany you to the cocktail party itself.
x.com/vanschneide...
November 13, 2025 at 9:30 PM
We already have digital necromancy in the form of Tupac holding concerts every few years.
The cocktail party question, "Who would you want to have dinner with, alive or dead?" would be wish fulfillment.
Heck, with AR, the dead can accompany you to the cocktail party itself.
x.com/vanschneide...
The cocktail party question, "Who would you want to have dinner with, alive or dead?" would be wish fulfillment.
Heck, with AR, the dead can accompany you to the cocktail party itself.
x.com/vanschneide...
Brand guidelines are a prompt.
November 13, 2025 at 9:22 PM
Brand guidelines are a prompt.
I wished the hand controllers were also keyboards. Then this could be a work machine.
store.steampowered.com/sale/steamf...
store.steampowered.com/sale/steamf...
Steam Frame
VR and non-VR gaming
store.steampowered.com
November 12, 2025 at 8:24 PM
I wished the hand controllers were also keyboards. Then this could be a work machine.
store.steampowered.com/sale/steamf...
store.steampowered.com/sale/steamf...
How often do ya'll rewind an LLM conversation to regenerate a response, as opposed to just plowing forward with corrections and keep the history of mistakes?
I find I go up and regenerate, to have a cleaner context and not confuse the LLM. Are current models not as susceptible?
I find I go up and regenerate, to have a cleaner context and not confuse the LLM. Are current models not as susceptible?
November 12, 2025 at 7:00 PM
How often do ya'll rewind an LLM conversation to regenerate a response, as opposed to just plowing forward with corrections and keep the history of mistakes?
I find I go up and regenerate, to have a cleaner context and not confuse the LLM. Are current models not as susceptible?
I find I go up and regenerate, to have a cleaner context and not confuse the LLM. Are current models not as susceptible?
One mental muscle I've noticed atrophying through AI use is the muscle for generating options. Intellectually, I know it's rare for AI to give you out-of-band options. In the moment, it's easy to be mentally lazy in this regard.
November 8, 2025 at 8:00 PM
One mental muscle I've noticed atrophying through AI use is the muscle for generating options. Intellectually, I know it's rare for AI to give you out-of-band options. In the moment, it's easy to be mentally lazy in this regard.
I currently don't see a bigger unlock than a designer who can get mainstream internet users to get used to owning and managing cryptographic keys.
November 6, 2025 at 6:00 PM
I currently don't see a bigger unlock than a designer who can get mainstream internet users to get used to owning and managing cryptographic keys.
When making stuff for others, you have to wear two hats: someone behind the interface, and someone in front of the interface. It's hard but necessary to remember to be the latter.
x.com/unconed/sta...
I'll echo the sentiment for DC metro's ticket machines.
x.com/unconed/sta...
I'll echo the sentiment for DC metro's ticket machines.
November 4, 2025 at 6:00 PM
When making stuff for others, you have to wear two hats: someone behind the interface, and someone in front of the interface. It's hard but necessary to remember to be the latter.
x.com/unconed/sta...
I'll echo the sentiment for DC metro's ticket machines.
x.com/unconed/sta...
I'll echo the sentiment for DC metro's ticket machines.
Surprised there aren't more "chain-mail" type of poast for engagement farming, where people propose you ask GPT to tell you something magazine quiz-like, based on everything it knows about you. Like:"Based on everything you know about me, what do you think my porn name would be?"
November 1, 2025 at 1:00 AM
Surprised there aren't more "chain-mail" type of poast for engagement farming, where people propose you ask GPT to tell you something magazine quiz-like, based on everything it knows about you. Like:"Based on everything you know about me, what do you think my porn name would be?"
I've started blocking accounts that consistently post obviously dumb opinions who aren't young people. If I find I have the urge to correct the tweet not from a place of sharing, but from a place of anger, it's probably engagement bait. Don't feed the trolls.
October 31, 2025 at 11:00 PM
I've started blocking accounts that consistently post obviously dumb opinions who aren't young people. If I find I have the urge to correct the tweet not from a place of sharing, but from a place of anger, it's probably engagement bait. Don't feed the trolls.
A new Sarah Paine episode arrived! Dr. Sarah Paine, more than any one else, was my gateway drug into the topic of geopolitics. Can't say I really cared about the topic before, but her lectures are riveting. I recommend them over random short-form videos.
www.youtube.com/watch?v=RH_...
www.youtube.com/watch?v=RH_...
Sarah Paine — How Russia sabotaged China's rise
In this lecture, military historian Sarah Paine explains how Russia—and specifically Stalin—completely derailed China’s rise, slowing them down for over a ce...
www.youtube.com
October 31, 2025 at 7:00 PM
A new Sarah Paine episode arrived! Dr. Sarah Paine, more than any one else, was my gateway drug into the topic of geopolitics. Can't say I really cared about the topic before, but her lectures are riveting. I recommend them over random short-form videos.
www.youtube.com/watch?v=RH_...
www.youtube.com/watch?v=RH_...
Would this be considered the enshittification of ginseng? The economic shape for cultivation doesn't allow ginseng to fully express its wild self. The taste we commonly get is an expression of the unit economics, not the ecology.
www.perplexity.ai/search/what...
www.perplexity.ai/search/what...
October 31, 2025 at 6:00 PM
Would this be considered the enshittification of ginseng? The economic shape for cultivation doesn't allow ginseng to fully express its wild self. The taste we commonly get is an expression of the unit economics, not the ecology.
www.perplexity.ai/search/what...
www.perplexity.ai/search/what...
I hate that it's the prevailing narrative, as it's a betrayal of craft. But this is the market dynamics when making new market categories. Only in mature markets that become infra does the downstream effects of code quality seem to matter to customers.
x.com/lennysan/st...
x.com/lennysan/st...
October 30, 2025 at 11:00 PM
I hate that it's the prevailing narrative, as it's a betrayal of craft. But this is the market dynamics when making new market categories. Only in mature markets that become infra does the downstream effects of code quality seem to matter to customers.
x.com/lennysan/st...
x.com/lennysan/st...
It's a missed opportunity by Anthropic where if you click the button, you can't continue the existing conversation someone shared. You might have other questions. Instead of comments on web pages, this might be more helpful.
October 29, 2025 at 5:00 PM
It's a missed opportunity by Anthropic where if you click the button, you can't continue the existing conversation someone shared. You might have other questions. Instead of comments on web pages, this might be more helpful.
I suspect why devs have wildly different reports on the viability and experience in vibe coding, is because some of us do wildly different things from the mainstream. The more some specific details matter the less AI can help us out without a lot of scaffolding.
October 26, 2025 at 5:00 PM
I suspect why devs have wildly different reports on the viability and experience in vibe coding, is because some of us do wildly different things from the mainstream. The more some specific details matter the less AI can help us out without a lot of scaffolding.
tag functions are nice. I didn't even know this was a thing in javascript. Well applied to writing raw SQL. I think this can do more.
bsky.app/profile/int...
bsky.app/profile/int...
October 24, 2025 at 9:19 PM
tag functions are nice. I didn't even know this was a thing in javascript. Well applied to writing raw SQL. I think this can do more.
bsky.app/profile/int...
bsky.app/profile/int...
Vibe coding is for experts. I think the Surgeon analogy is apt. Especially if you think software should be written only if you can't get it off-the-shelf.
Geoffrey Litt just proposed a new analogy for working with AI coding tools that I really like: you are the surgeon, staying in command and doing the most challenging work - the AI tools are your support team and surgical assistants simonwillison.net/2025/Oct/24/...
A quote from Geoffrey Litt
A lot of people say AI will make us all "managers" or "editors"...but I think this is a dangerously incomplete view! Personally, I'm trying to code like a surgeon. A …
simonwillison.net
October 24, 2025 at 3:36 PM
Vibe coding is for experts. I think the Surgeon analogy is apt. Especially if you think software should be written only if you can't get it off-the-shelf.
October 23, 2025 at 11:00 PM
patchedDependencies is such a nice feature. Even better that Claude found the bug and made the patch. This is the rare time when I feel like vibe coding saved me a lot of headache.
October 23, 2025 at 9:05 PM
patchedDependencies is such a nice feature. Even better that Claude found the bug and made the patch. This is the rare time when I feel like vibe coding saved me a lot of headache.
"I was using local-first software in the 80's!" quips a greybeard on HN.
No you were not. A common misconception about local-first software is that it's just software that doesn't use the network. Local-first is inherently multi-device and collaborative with other users.
No you were not. A common misconception about local-first software is that it's just software that doesn't use the network. Local-first is inherently multi-device and collaborative with other users.
October 23, 2025 at 3:32 AM
"I was using local-first software in the 80's!" quips a greybeard on HN.
No you were not. A common misconception about local-first software is that it's just software that doesn't use the network. Local-first is inherently multi-device and collaborative with other users.
No you were not. A common misconception about local-first software is that it's just software that doesn't use the network. Local-first is inherently multi-device and collaborative with other users.
Reposted by iamwil
i mean… jay literally tried to redefine it to mean something broader. still one of my favorite articles even though it didn’t catch on
Web3 is Self-Certifying
There’s been a lot of discussion lately about what Web3 is and isn’t. Here’s my definition: Web3 is user-generated authority, enabled by…
jaygraber.medium.com
October 22, 2025 at 4:06 PM
i mean… jay literally tried to redefine it to mean something broader. still one of my favorite articles even though it didn’t catch on
Reposted by iamwil
incidentally, this is how I learned everything in 3D
I had an image I wanted to make, and no idea how to get from A to Z, but you just start stumbling through stuff and learn a bit of B or C, or a bit of LMNOP and then back to E or F, and suddenly it starts resolving, solidifying out of the fog
I had an image I wanted to make, and no idea how to get from A to Z, but you just start stumbling through stuff and learn a bit of B or C, or a bit of LMNOP and then back to E or F, and suddenly it starts resolving, solidifying out of the fog
The old path: Learn Lean (6 weeks), study abstract algebra (8 weeks), understand group theory (4 weeks), finally attempt your proof.
The new path: Start with your exact problem. Generate a tutorial for it. Backfill concepts as you hit them.
The new path: Start with your exact problem. Generate a tutorial for it. Backfill concepts as you hit them.
October 20, 2025 at 9:12 PM
incidentally, this is how I learned everything in 3D
I had an image I wanted to make, and no idea how to get from A to Z, but you just start stumbling through stuff and learn a bit of B or C, or a bit of LMNOP and then back to E or F, and suddenly it starts resolving, solidifying out of the fog
I had an image I wanted to make, and no idea how to get from A to Z, but you just start stumbling through stuff and learn a bit of B or C, or a bit of LMNOP and then back to E or F, and suddenly it starts resolving, solidifying out of the fog
Reposted by iamwil
this is the way to learn with ai. you can start anywhere and backfill. you have to remain curious and careful and not just passively eat up plausible explanations. but if you put in the effort and the model is good, it’s powerful
The old path: Learn Lean (6 weeks), study abstract algebra (8 weeks), understand group theory (4 weeks), finally attempt your proof.
The new path: Start with your exact problem. Generate a tutorial for it. Backfill concepts as you hit them.
The new path: Start with your exact problem. Generate a tutorial for it. Backfill concepts as you hit them.
October 20, 2025 at 7:10 PM
this is the way to learn with ai. you can start anywhere and backfill. you have to remain curious and careful and not just passively eat up plausible explanations. but if you put in the effort and the model is good, it’s powerful
New post: Most people think you need foundations before attempting hard things. Prerequisites as requirements.
But that's just an artifact of industrial-scale curriculum design. Prerequisites are activation energy, not gatekeepers. 🧵
But that's just an artifact of industrial-scale curriculum design. Prerequisites are activation energy, not gatekeepers. 🧵
October 20, 2025 at 7:03 PM
New post: Most people think you need foundations before attempting hard things. Prerequisites as requirements.
But that's just an artifact of industrial-scale curriculum design. Prerequisites are activation energy, not gatekeepers. 🧵
But that's just an artifact of industrial-scale curriculum design. Prerequisites are activation energy, not gatekeepers. 🧵
Learned a new word: "chiaroscuro"
Good for image prompting, I guess. I have yet to find a use for "erinaceous".
Good for image prompting, I guess. I have yet to find a use for "erinaceous".
October 18, 2025 at 6:00 PM
Learned a new word: "chiaroscuro"
Good for image prompting, I guess. I have yet to find a use for "erinaceous".
Good for image prompting, I guess. I have yet to find a use for "erinaceous".
It's the same debate going on when I was in school. I agree then as I do now: it's better for the students to learn the theory in school. Glad my school didn't bow to the pressure to teach Java in the classes.
x.com/GuptaAnkitV...
x.com/GuptaAnkitV...
October 17, 2025 at 8:30 PM
It's the same debate going on when I was in school. I agree then as I do now: it's better for the students to learn the theory in school. Glad my school didn't bow to the pressure to teach Java in the classes.
x.com/GuptaAnkitV...
x.com/GuptaAnkitV...