Computers are neat and people are neat. 🧑💻 @Droplet.io 🦣 @jergason@mastodon.xyz 🦋 @hi@jamison.dance 🎙️ @SoftSkillsEng
ᕕ( ᐛ )ᕗ
Joined March 2009
- Tweets21.2K
- Following893
- Followers8K
- Likes23.4K
Normalize adding the adjective "earth" to things. "this is my earth job", "please review my earth PR", "my earth wife love that song"
Jamison Dance retweeted
every product i use should just be a headless mcp, that feeds into an agent, where i never have to think about anything underneath.
except for our product, of course.
in our product, customers love the ui and love us very much.
we're special that way.
Jamison Dance retweeted
When you make something with AI -- whatever it is, software, a design, an essay, an explainer -- superficially it looks good.
And it's unsurprising that it does because it was trained to be accepted - to get you to hit that lgtm button. It therefore of course hits all the triggers in your brain that say "nice": tasteful rounded borders, warm greys, modern typography, unit tests, whatever.
But if you force yourself to really engage, you will likely find a million little things you want to change.
The way I do it is I just start at word one. Or method one in a type. Or top left corner of the UI. I tune out the big picture that is tempting me to say "🤷♂️ - lgtm" and force myself to consider just the very first, smallest piece:
"Is that really the intro sentence I want? Is it communicating what I want?".
Almost invariably, I will find I have an opinion about it that I couldn't see before. Then I move to the next bit. My job is chief critic.
You may say "but then what is the benefit of the AI!? You're slowing it down!"
But to me this is exactly the benefit. The same power that allows AI to make progress super fast can also be used to make things incredibly high-quality:
* I can change my mind a million times. AI doesn't get grumpy or tired.
* I can work on a million things in parallel.
* I can revisit ideas I've already tried.
* I can work when I'm not even at my computer.
Slop or craft -- it's up to you.
Jamison Dance retweeted
“While the act of creation is inherently satisfying, there’s a danger: one might find satisfaction in the labor itself and mistakenly believe that a high-quality piece of work has been created,” he said.
very insightful comment from miyamoto
Nintendo’s Shigeru Miyamoto, the acclaimed designer behind Super Mario and Zelda, has explained why he believes generative AI won’t make human creativity obsolete.
"There will always be a role left for us to play."
videogameschronicle.com/news…
Jamison Dance retweeted
I think this is like indisputably true for rewrites now.
For new things, I think you still have to choose between something like: (1) a bottleneck on human judgment, or (2) give up control, stop caring about any of the details (may not even realize this is happening).
“Knowing what a fuzzer is” is now a prompt away from fuzzing your codebase. This is a way that good engineering is better post-LLMs. This wasn’t worth it for 99% of projects. Now, fuzz all the things?
Jamison Dance retweeted
A decade ago, after struggling on a hard problem in React Fiber, I spent several days designing and iterating on a fuzz tester that eventually uncovered a bunch of subtle cases that helped me converge on a solution.
Today, a fuzz tester simply appeared in my codebase. Poof.
Ah crud someone did it earlier and better than I :( nitter.cf/pogarithm/status/21101…
I have little data for this but it *feels* correct to me.
Every few months I go rip out a bunch of manually-supplied context that newer models and harnesses don’t need anymore.
Jamison Dance retweeted
a weird inversion with LLMs is the models improve faster than the tinkerers
when i see people with custom workflows and setups they're all addressing problems that don't exist anymore
the person naively using vanilla codex is more likely to be experiencing state of the art
Jamison Dance retweeted
A framework is an intelligence cache.
You could waste tokens reinventing the ORM, cache, queues, auth, routing, etc, or you could just pick a world-class framework.
