I believe it's just worse battery life (10-14 vs 30 days), no heart rate monitor, worse IMU, and no speaker (but still a mic). That seems to be about it. DPI and the screen are basically the same.

The full specs are further down on this page

As a side note, this is why I'm a bit interested in wherever @teal.fm or other WIP music atproto systems end up. The features that exist in bsky can easily translate to different mediums. Labellers, feeds, starter packs, and so on would all would work great on a music-centric platform.

The existing clouds aren’t going to go without a fight though. I worry about regulatory burden potentially reducing the capability of individuals from running a competitor.

Also, we haven’t seen any malicious licensing or encoding of public PDS data yet, which makes interoperability impossible.

Yep, definitely. It’ll be open sourced if/when I finish it.

I personally worry more about reliance than I do deskilling. I found that I became more "tiktok brained" when I started using LLM tab-completion. Rather than thinking, I just shunted that to the LLM and hit tab.

The CLI tools are better for me, since I need to put more effort into controlling them.

I still have to write a script for it. Shouldn't really be that hard but I have to work out how to 1. Not spam anyone listening for records and 2. Normalize the records in the same way Piper does.

I haven't decided if I'm going to go through Piper or just directly write to my PDS either.

For reference, I had about 82k minutes of Spotify this year that I’m looking to backfill. A lot more over the last decade as well 🙂

Also shout-out to foundationdb-exporter for saving me the trouble of parsing the status into a Prometheus format. Really great included Grafana dashboard as well.

Being fair here, I likely gave it poor instructions and a problem pretty far outside of it's training set, so maybe this result is to be expected.