Very nice! I’ve built something similar for Berlin[1] which is based on a digital surface model to also accurately cast shadows from trees, hills and bushes.
Working on making an old Nokia 3210 use 4G because 2G is being shut down within the next two years in most European countries.
As it currently looks I need to design my own PCB with a modern LTE module and make it connect to all the peripherals. Thought it would be a nice project to get into electronics, soldering etc. Also really enjoying the constraints of an 84x48 LCD screen and how to implement "modern" technology like GPS navigation or design the UI for a music player.
> It’s a disciplined practice of observation, experimentation, critique,
revision, and aesthetic judgment that serves to develop creative ideas
I wonder where an increasing share of AI-generated entertainment vying for attention will take our skill for discerning aesthetics and creativity. Not only does it require the creator to actually care enough and go the extra mile to create something novel and maybe even outside of the generative realm, but it also requires enough people to be focused and attentive enough to then consume this with a similar discipline without solely relying on summaries and generated shortcuts (e.g., go to an art gallery, look at a painting, be present with it and define one's own meaning instead of watching a 15-second clip about it with AI-generated voice over and explanation text). In that vein I commend the project's constraint for the ideation phase to stay as analog as possible.
Thank you for putting into words how a lot of AI prose makes me feel. 'Persistent mic drops' is the most accurate way I have seen it described and I would even argue the 'this-not-that writing' is part of it. Setting up a chain of ever increasing seemingly powerful phrases to then conclude in a final statement which may or may not actually explain the core concept.
It also made me think of the Feynman Learning Technique[1] or learning by teaching in general: being able to accurately explain in simple terms requires to fully grasp the concept and noticing gaps in the explanation is an insight to the explainer to evaluate and understand those parts again at a deeper level. An AI arguably does not have this understanding and will happily drive home any point we have asked it to make.
In a way I think it's tied to the other discussion from today[2] where it was posited that domain knowledge makes you better at prompting LLMs and, consequently, would enable you to also better structure and evaluate explainers. But then again, explaining well is a different skill altogether.
This reminds me of LiveCode[1] (aka Revolution and MetaCard) which I’ve used quite a bit about a decade ago. Was also inspired by HyperCard. Just checked their site again and it seems they’ve discontinued the open source version and have pivoted to an AI-assisted drag and drop UI builder.
I'm using LiveCode for a dozen personal applications (and for just the fun of coding; I'm a retired s/w engineer). It's no longer free, but I started with HyperCard and LiveCode is an excellent replacement, at least for me.
Thanks for letting me know, maybe I need to get back into it. I always enjoyed the speed with which I could prototype new ideas and even get them installed on a phone or tablet.
Nice writeup and I really like the minimal aesthetic of your site!
On unique users: I wonder if there is a way to set this up to work similar to TelemetryDeck's double-hashing anonymization[1]. Did you look into that before deciding to go purely session-based?
Based on my understanding of the GPDR, this scheme wouldn't be anonymous, if the salt is stored anywhere. Using the salt(s) and the full list of all your users emails, you could find out which one matches the hash, thereby linking the analytics to the original PII. Of course that's a rather ridiculous idea, but AFAIK the GDPR doesn't put any qualifiers or limits on "it's not anonymous if the PII can be recovered with extra data".
It wouldn't if you own the full chain, i.e. have knowledge of both the on-device salt and hash AND the server salt and hash. However, in the scenario of using a third party like TelemetryDeck you as the developer would not be able to link the device hash to the server hash because you never see the latter.
So, my reading is: if there would be some mechanism (or a third party) which obfuscates the server hash so that you cannot tie it back to the device hash and only tells you "we've seen this device already" it could be anonymous. Of course, it also heavily depends on what other data you associate with that.
We are giving our dog Fluralaner/Bravecto[1] against ticks which paralyzes and kills them when they bite, drastically reducing potential exposure to Lyme disease. Unfortunately not approved for humans and therefore an interesting inverse: humans can get vaccinated against TBEV but not against Lyme disease whereas dogs can get vaccinated against Lyme disease but not TBEV.
This is exactly what I am working on for my city (Berlin). Right now it gives you sun/shade times for outside areas of cafes and restaurants as well as heatmaps for public parks and playgrounds.
The idea is to eventually offer a sun or shadow mode and the app will then surface areas, squares, and streets with more sun or shade and eventually offer walking and biking routes with more or less sun.
[1]: https://news.ycombinator.com/item?id=49589455