The point about interaction with undo in a digital inking app is interesting. Seems like with ML handwriting recognition, we could create a handwriting "text editor" instead of a drawing app that understands the semantic structure of the text and operates at that level instead of the individual stroke level.
I had some friends who worked on this for a while but didn't get far with it. It turns out it's tricky to design a general purpose system that adapts to users' handwriting idiosyncrasies and is fast enough to work in real time with handwriting with sufficient accuracy.
Some of that work is still visible in the ChromeOS virtual keyboard handwriting mode, but it's very limited - one line at a time. https://cursive.apps.chrome/ (the built-in note-taking app on Chromebooks) has some interesting inking gestures for manipulating written words, but doesn't have any semantic-level understanding.
I don't see how "trust" is anthropomorphizing. Do I trust this bridge to hold my weight? Do I trust that this tool in my hand will perform as expected?
"as such [LLM-coded interactive] supplements are not mission-critical to the core of the paper, I again feel that the downside risk of using guided interaction with LLM agents to generate such visualizations is acceptable."
It's a tool. Good for some things but not others and generally not to be trusted.
Use it to write and edit code when that makes sense for you to do so. Just like you always have. It's an extremely useful skill to have even still.
I've had much more success with agents reviewing my code and offering inline autocomplete over LSP than I have with letting the agents write the code, which I then try to review. I end up with a much better mental model of the code and higher quality output than either I or the agent could do alone.
That might work if the software platform vendors kept compatibility with old versions, but they're quite poor at this. Windows used to be quite good at it, but Android and iOS are really bad at it, and macOS is not a lot better.
Software is, unfortunately, a living thing and when not actively maintain bit-rots and becomes unusable.
A reef knot (or square knot) cannot be adjusted or tied under load. When you tie the second crossing, the first one always loosens a bit. The lapp knot in the video can be tensioned properly after tying, which is pretty slick.
Wow there is so much wrong with this post I don't even know where to start.
Widespread car usage has caused an enormous amount of harm. It has destroyed American cities, killed thousands of people directly, tens of thousands more with sedentary lifestyle diseases, and burned a truly enormous amount of fossil fuels into CO2, and locked in the infrastructure to continue to do so for decades. It's an absolutely enormous tragedy.
Cars are also useful and have helped a lot of things, but they should not be the default choice of transportation.
Not sure where this icy, thunderstorm-prone city where it's either walk or drive with nothing in between is, but it's certainly pretty far outside my experience.
From what I've observed, the vast majority of car trips in my city could be easily replaced with a bicycle trip.
We would most certainly notice if the modest bicycle traffic in our city was replaced with car trips. The already terrible traffic would be significantly worse. "Stealing" some space for cars for a drastically more efficient means of transportation that's also 5-6 times faster than walking is a huge win both for people who want to get places reliably and on time and those of you who are always carrying cargo or can't cycle for some reason.
You wouldn't hear from me at all if it weren't for accusations like "stealing" space from cars (who in turn stole entire walkable city centers and streetcar lines!) and just made a little room for those of us who are able (which is most people!) to choose a healthy, efficient, humane form of transportation instead of hauling a 3 ton steel caged air-conditioned living room with armchairs and sofa on every trip.
I had some friends who worked on this for a while but didn't get far with it. It turns out it's tricky to design a general purpose system that adapts to users' handwriting idiosyncrasies and is fast enough to work in real time with handwriting with sufficient accuracy.
Some of that work is still visible in the ChromeOS virtual keyboard handwriting mode, but it's very limited - one line at a time. https://cursive.apps.chrome/ (the built-in note-taking app on Chromebooks) has some interesting inking gestures for manipulating written words, but doesn't have any semantic-level understanding.