I thought it was Autonomy. You installed a program, instructed puppies agents, logged out, and while you were offline the puppies searched through several engines. Next time you logged in the findings waited for you. That was the time of 56k modems and metered connections.

Yes, and you might also say that time-starved humans just reviewing LLM output may generate more accurate reports than having to write them from scratch in a rush. That’s until humans get complacent or are expected to do even more per minute. But there is a fundamental difference. Unlike humans, LLMs don’t understand context and don’t do sanity checks. When they hallucinate they can do so wildly, without a sense of implications, but always with confidence.

A policy I saw coming out of an NHS (UK) department mandated ‘human-in-the-loop’ which is essentially what the article mentions in the end. The risk is that over time clinicians may become complacent with ‘good enough’ and don’t bother to review thoroughly. And it may be easy to spot mistakes, but not necessarily omissions unless you keep your own notes. More so after a long session, although medical appointments are typically short and focused.

On a positive note, in my experience clinicians using LLMs do indeed spend more time engaging with service users. In an ideal world, they would be given time to engage and take notes, but this is not going to happen.

EEA and UK apparently.

In addition to wifi, Bluetooth beacons would be good too.

Seeing the same SSIDs (eg in a cinema) might also mean you are not moving, but then how can you tell you are not sitting near another train passenger with their hotspot on?

Jepster on Google Play was good but from v8.0 it won’t start if, like me, you have Google Play Store disabled. Presumably, they need that for the optional in-app purchases but they never replied to my email so I don’t know.

From FLOSS, I am experimenting with FitoTrack which looks promising. Another one is AAT.

Colota is great for general self-tracking.

I’ve been using this for a few weeks and it’s great. In addition to offline-first, it would be nice to be able to ask Colota: List my trips between date1 and date2 when I was near (ie within x meters from) point y.

I am planning to use this for a long time too, so an export/import data for when I change my phone would be nice. I see Export but not Import.

Also, being able to delete trips between date1 and date2 would be useful. Currently, you can delete 1-by-1 or recent trips only.

I run a Tor proxy on my raspberry pi and all my browsers across all my household devices have foxyproxy installed. This way we access defined websites through Tor as standard eg when we query search engines and wikipedia. I try not to overburden Tor though. Never had a problem.

Windows refugee here. I installed Debian 13 with KDE Plasma on my main machine four months ago and I am still ironing out issues. Eg CUPS was asking me to login all the time and didn’t accept my credentials. After some days researching I discovered I had to log in as root. Then, I discovered I didn’t have root credentials for some reason. I had to create them and then add my local user to a group! Just to be able to use my home printer.

Or suddenly my clock was 62 minutes off. I discovered the NTP service was never set up properly and I had to install chrony.

I don’t see how I could have avoided using the terminal. These are only a couple of examples. No deal-breakers and on this occasion I had the time and determination to resolve them. I could have easily given up.

My understanding is that sandboxing is not mandatory for Snaps, but it is for flatpaks. Some of the Snap code not being open source, and generally the technology being centralised around Canonical apparently is off-putting for some.

I share your concerns about trust. With flatpaks we can still read the source and commits, but not many will or can do this every time they install and update software anyway. In this sense, we have little choice but to trust the verified developer and the community, who may of course be compromised too, regardless of distribution method. I suppose with flatpaks we have to check permissions and make them as restrictive as possible.

This ranking is very close to how I see this. Anything after Docker/Podman is out unless I absolutely need an application in which case keeping a record of dependencies is a good idea. But I want to know the work system will absolutely start in the morning hours from a deadline. Avoiding single points of failure is another way of course (ie multiple systems, OSes, backups, password managers etc).

I remember the time applications came on floppies, 640kb of RAM was indeed enough for anyone, and people competed in writing games in one line of BASIC (yes, that was 255 characters code max). Containers feel horribly wasteful to me, but I came to accept there aren’t many realistic alternatives for the average users who need reliability with zero effort. Making a note of dependencies in case you need to backtrack is not a realistic proposition for most. But I can understand why some users will want full control and a lean setup.

I agree with the popular view that Debian Stable + KDE Plasma + Flatpaks (or Appimage, Docker) strikes a balance between system reliability and freshness in selected applications when that counts. I may be missing updates for KDE Plasma but v6 is quite mature so I don’t mind. I know storage is cheap but I am instinctively uneasy with containerisation as it’s done by Flatpaks etc because of the duplication you get with all-in. But if that’s the price of reliability, so be it. It’s just that sometimes there is only a PPA or a .deb, which is why I asked.

EDIT: I just tried distrobox for the first time. It is amazing how efficient it is. I ran Firefox on Arch and I couldn’t tell the difference in resources. Amazing really.

I have been preparing the move to Linux for years, switching to FOSS cross-platform applications on Windows and installing Linux on my secondary machines. A few weeks ago I made my work machine dual boot with the intention to remove Windows completely. I find that I never log into Windows at all already, and my Debian Trixie + KDE Plasma experience is the same in many areas (mainly because I use the same applications as before) and vastly better in others.

There were issues I had to solve but nothing major. It is true that Windows has been very stable and efficient for me, but people forget that when this happens it is the result of many years of learning, fine-tuning, decluttering and getting used to Windows. You get to that stage with Linux very quickly, and it feels much better.

True for most scenarios. Specifically with Syncthing, I find that it rarely fails and when it does there are good reasons and I need to do something about it (eg I used the wrong version config.xml recently trying to migrate between Syncthing setups).

You learn something new every day here ;) The good thing about kdialog is you can’t miss it. Thanks.

I have tried Syncthingtray and Syncthingy. I found the former did too much that I never used, and the latter was an unnecessary process doing very little while always running and adding another icon to the tray. For me periodic checking with a script is enough and more efficient for the rare situation where Syncthing crashes of fails to start.

I went for this one and it works with both notify-send and /dev/pts/0. Not sure why it is better, but I opted for the latter. Simple, lightweight and versatile, suitable for any process.

Any KDE Plasma users reading this, to enable notifications history for these you can follow the instructions here. Many thanks everyone.

enx0050b6c0f7f3 is in fact my docked ethernet. I solved the problem when I discovered it was specified ‘unmanaged’ for NetworkManager. I added a note in my original post.