@categorically.me
Girl Dad Founder @ No Discipline - Transdisciplinary problem solving with systems approach. CTO @ Hikma health - building health tech for refugees and migrant populations worldwide https://categorically.me https://elsa.health https://nodiscipline.org
After 15+ years of teaching, architecting & delivering CSS for global organisations & startups, I’ve collated that experience into an extensive course which will take your CSS skills beyond the next level.
Complete CSS has taken nearly a year to produce and it’s officially live 🎉
complete-css.com
I’m struggling to get reliable street level env air quality data - Working on health+env models with clinics serving unhoused populations and recently learned that they are often kicked out of their neighborhoods and only return for clinic days … not sure where to start … help?
#data #climate
This is why I'm appreciating ai;dr as a sibling to the usual tl;dr.
Nice!
Well, we shipped it!
@preactjs.com is joining the party with an official @tanstack.com Query adapter 🎉
Huge shoutout to @thvdnta.bsky.social, who drove this effort 🙌
Welcome, @preactjs.com 👋
I leave with 7 books, come back with 14 …
Oh yeah this was a great talk!
TigerStyle.dev
A new, visual home for TigerStyle -- the software engineering methodology developed by TigerBeetle to produce safer, faster software in less time.
"Do the hard thing today to make tomorrow easy."
"If you want to be remembered, be remarkable."
[doing things I’m bad at]
“ugh, I suck at everything”
[doing things I’m good at]
“this doesn’t count because it’s easy for me”
Ouch, yeah… sounds about right 🙃
2006: Less is more, don’t repeat yourself
2016: Beautiful code prevents technical debt
2026: trillions of lines of generated spaghetti code that generate other spaghetti code will save us
Ha! Yes!
It’s not hyperbolic to say AI is a multiplier. A multiplier multiplies. I’m seeing more people just getting better are “meta-work” and large volumes of mediocre work.
If you don’t know where you are going you will just go further in random directions. The real problems don’t go away.
When using LLMs, it's very important to give clear instructions. Spec-driven development is the future.
For example, a clear prompt to add two numbers could look like:
ADD a TO b
ON SIZE ERROR
DISPLAY "Error"
This specifies the overflow as well.
A toilet that has — and I can't believe I'm writing this — latte art (classic leaf) in it
vibe coding
hey guys, just to confirm this is my fundraiser. i didn't want to ask for help, but i do need it.
my cancer is back. i am going through a lobectomy next week, the same surgery Walter White undergoes. i can't express how grateful i am. thank you so much everyone.
www.gofundme.com/f/support-ge...
life is wild right now!
@todoist.com , can we get description for projects? This could be a game changer! Often projects need a short description (or a long one), and we have to put it in a doc elsewhere… and have to juggle tools.
I wouldn’t need google docs for project spec documents, my team can just live in Todoist!
Writing is thinking
Outsourcing the entire task of writing to LLMs will deprive us of the essential creative task of interpreting our findings and generating a deeper theoretical understanding of the world.
I'm convinced understanding Geometric #Algebra requires unlearning almost *everything* from years of Linear Algebra exposure.
My brain is actively rejecting GA - especially in applications.
This transition in thinking frameworks is worse that when I gave up Frequentism for the church of #bayes.
It feels like eating but not getting satiated, and feeling more empty.
But it’s a great business because it’s much harder to work without it after using it - skill loss is 100x easier than skill gain. Feels like for the first time ever neural plasticity is a handicap
A screenshot showing the words on my mute list. Agentic, agent, GPT, grok, ChatGPT, claude, Opus, gemini
This mute list feature is pretty good. Finally the good stuff in my feed is not swept away by threads on how good vibe coding is.
Sorry for repost, forgot the alt text
Yes, human programmers can create problematic code, too.
But not at that speed.
Bsky went from ‘Using LLMs is a mortal sin & you’ll burn’ to ‘you must use LLMs or else you are a dumb ignorant Luddite, let’s all pile on you’.
Maybe it’s just my feeds …
Both suck! Luckily in #bsky I can just change who I follow and my algo 😊!
Pls recommend #science peeps who do creative work!
I see what you mean! On the other hand, can we say generative AI saves us time making digital art, so we think for longer, and ship less digital/media? (Probably not)
For many the issue was never the code generation, it was the excuse not to ship - this group still won’t ship (also including me)
I see your point, but only if weird is measured as “not what most others are doing” and not as a negative quality. I still don’t like using LLMs, simply because they take away the parts I like and rob me of AHA moments in the details. Which, to your point, may make me weird 😊!
Ha! I would only disagree that your kid’s crayon drawings could provide some common ground, express empathy or vulnerability which could strengthen a relationship … your AI-generated clip-art, not so much.
Also, love the pod! Thanks for the content!
Amazing! Thanks!
