When I'm using coding agents and LLMs extensively in some tasks, I feel way more resistance to grappling with difficult, uneven problems in others. As if I get recalibrated to expect ease, effortlessness, velocity in everything. Not good!
Curious how you deal with this, if you feel it too.
But if we draw a parallel to architecture, faster and cheaper building has not lead to better buildings. To give new tools the right shape you need appropriate context, and oh mate that’s difficult
Me too! Kinopio but also your approach to building a tool (and your write-ups) has been a great inspiration for me.
Someone said hey let’s read ‘70s papers and stuff, and then there was a sudden explosion and everyone started running around and squeezing the next second because hey, the singularity is tomorrow at 11.35. Disclaimer: more feeling than clear thinking.
Is it obvious? My hypothesis was more based on the large misbalance between some nodes in the network, making the larger nodes extract from the smallest... reliably...
Slide: We ended up in casinos Not by choice, but by design platforms built to extract attention, maximise engagement, and serve private profit above all else.
It's not just that we live under private government of the monopolies, but it's private government by casinos. #Rebuild
I think we’re going back to Lindy anytime now
Do you make drag-and-drop interfaces? Or do you wish you did, but they're just too hard?
We're working on a new way to drag & drop! Wanna try it out? Join our study – we'll ask you about your drag-and-drop experiences, and get your hands on the prototype.
Join here: forms.gle/J82tFtu6DqS8.... 🙏!
Yes. That’s why I feel we could build alternative ways to approaching the codebase that are not dependent on the FS arbitrary-ish choices. For example user flows, specs, patterns…
Specs, design choices, general patterns... these are a great guide from any contributor, whether they are an LLM or a human, to reach the specific areas to modify.
You could argue that the directory structure is often based on arbitrary choices, or aesthetic trends. Some people put hooks together, some don't, for example. User flows are not necessarily clear when you see a codebase, but need to be understood to make changes to the code.
But... a codebase is not a tree www.patternlanguage.com/archive/city...
10m documentary film on the local-first software movement!
www.youtube.com/watch?v=10d8...
Do you do so based on your feeling on how things are going? I ask because these mindsets are sometimes opposed to each (for example, let it crash VS degrade gracefully) other and depending on the context one may go for one or the other.
Ah, and it also applies to agents interrupting agents.
I mean, sometimes you’d rather work in parallel with short touch points. Without too many interruptions, so you can focus. It feels that this would also apply to working with agents. Hence my comment.
But even with agents, if we want to stay sane we might need to have them asynchronously do their work without interrupting us as they go. So non-realtime could also be great with agents
I do see a parallel to being permanently communicated. Sometimes it can be a blessing *not to be able* to pursue an idea at a given time.
In the press release they talk about a partial bus connection so I'd guess they're operating with the same trains, only rerouting those with mixed gauge
Oh! I found the right quote: "Putting a concert hall beside an opera house is almost never practical to either audience, and only a consequence of that simple-minded part of us which puts things with the same name in the same basket" C. Alexander "A city is not a tree" bp.ntu.edu.tw/001/Upload/1...
That I can relate to. But I wonder if there are not better metaphors. For example, instead of placing a thing in folder A or B, why not place it closer to place A, or in between places A and B. Dots on space. Or things tied to other things.
"Work with what we have", but it's a surprising limiting factor. Files and folders has always been the wrong metaphor, following closely a physical reality where you cannot place something in more than one place.
Changing the s130’s interiors to add couchettes and minicabins
They’d need to repurpose the S130. Or buy new trains, but I don’t know if they are ready for that, plus it seems, like everyone, are unsure about the internal configuration.
I would start with a Madrid branching hub, as you mention. But HS line usage needs to be fixed, as well as gauge changing locos.
Yo también viajo así regularmente y mejorará cuando haya trenes pasantes por Madrid, pero a partir de las 5h de viaje, el tren AV es caro y demasiado largo, y la gente se va al avión. Esas relaciones son las que se pueden hacer en tren nocturno.
En el horizonte 2035, con las últimas ampliaciones de AV, hay muchas relaciones transversales e internacionales para las que un servicio AV diurno de 6h+ roza el límite. No dejar otra alternativa que el avión no es una opción. Si queremos tren para muy larga distancia hay que darle una vuelta.
Mucho avance 👏 pero seguís necesitando gente buena en diseño de información y experiencia de usuario. Las interfaces podrían ser muchísimo más claras. Sería una pena que eso empañara el trabajo hecho.
Y'all, I'm super excited to announce that a workshop we've been planning is finally slated to happen at this year's CHI. The topic is the Science and Technology of Augmenting Reading (STAR) 📖 💫 chi-star-workshop.github.io
Back to iron
