With one difference. After merging with text.com they switched to a model with on-device bridges so that the decryption on their servers is no longer needed. Ofc it’s similar if the matrix instance and the bridges run on the own server.

To be fair though, moving personal to institutional knowledge was always a challenge and rarely works really well. While I value apprenticeship a lot (I do science in Central Europe where that is pivotal) I wonder whether it is also a way to move personal knowledge from person to person without ever becoming institutional knowledge. Management didn’t just bury the legacy of Ben, they missed making sure that Ben and Sarah were leaving a manual which cannot burn. We know similar problems because we, as in the scientific endeavour, keep telling people that doing core developments and writing papers about it on half-year contracts in different institutions half a globe apart for a decade is about excellence and learning to become senior rather than a lack of commitment. And we have done so for a long time. But at least juniors, dreaming of becoming a sailor on the research vessel, keep coming.

And after watching ML is exacerbating existing problems in other fields for some years, we start (!) debating whether it might be slowly replacing us, too. But rather than challenging LLMs writing papers so that other LLMs can summarize them for us, we are still thinking about the next paper and how it will be cited most because that is how it always was and will always be.

So it is not just about greed. It is the idea of ever performing better to death and the way we define success. The same reason we self-optimize and love that fitness watch and the paid subscription so much because it helps us building habits and being strong and fit and better rested so that we can work even better. In the end, we must acknowledge that we have been part of it all along and the rest is a mirror of what societies’ rules have become.

What software stack are you running on it though? Asahi doesn’t seem to support the M4, yet.

I don’t find this convincing. Everyone knows how easy it is to click “yes” without actually knowing what’s behind it even reading it at all. There’s a reason that “I have read the terms and conditions” is the most common lie on the web. Also, no one will read a lot of information. So adding AI sounds great, doesn’t it? While this is a technical opt-in, it is still really close to an opt-out.

And then the ML generated code. I am aware it’s anecdotal evidence and might not mean anything but every time I tried Ubuntu, I deleted it after a few days because I found the bugs way too annoying after hitting too many road blocks. So if a buggy prone desktop environment additionally accepts ML code, it makes me suspicious and wanna stay away from it.

Exactly at the speed of light, the γ-coefficient would be infinite and so would be the time dialation. The eigen time of the moving person would thus be infinitely slower than the non-moving person. From the perspective of the stationary person, the time of the moving person would stand still and thus the person would never say anything. Very close to the light speed, when the coefficients are large, this problem eases but persists. The stationary person would have to wait for very long (and use a massive Doppler shift of the moving signal) to perceive something. At the end of the conversation, it will have lasted much longer for the stationary person, spending years on this. The twin paradox would basically kick in as well. If the moving person is at a speed too close to the speed of light, the stationary person might die before the conversation is over—assuming the stationary person is not immortal. That is kind of a very slow motion, yes. What a dedication, spending a lifetime on a person who can’t slow down ;-) Funny enough, from the perspective of the moving person, the effect is reversed.

Threema is great, I’m using it privately a lot. Several German authorities have set up and published variants of Element X, using the Matrix protocol in the backend.

The only thing that really got me going was small applications I had some interest in. Writing games I will not play never kept my interest up for long. So I’ve been building mini tools I used for teaching numerics classes in meteorology (Python, Julia, Fortran, C) or code I would be using for some tinkering with microcontrollers or similar (Python, C++) when using new languages. For me personally, some iconic projects were a CO2 sensor with an attached display, or very simple internal gravity wave ray tracers. But that’s likely not what you’d be interested in. So without trying to suggest specific applications for you (many good examples in the responses :-) I’d advise to do something you’ll have fun with. Get yourself a small project that generates added value for you specifically (and fun is great added value in my eyes).

Link to app repos, both contain the disclaimers: https://github.com/eu-digital-identity-wallet/eudi-app-ios-wallet-ui https://github.com/eu-digital-identity-wallet/eudi-app-android-wallet-ui

Maybe the security expert could read the readmes in the repos first. From the iOS app repo:

The initial development release has reduced security, privacy, availability, and reliability standards relative to future releases. This could make the software slower, less reliable, or more vulnerable to attacks than mature software.

And further:

If you’re planning to use this application in production, we recommend reviewing the following steps: […] The Pin storage configuration matches your security requirements, or provide your own by following this guide Pin Storage Configuration […]

So the text hints not at design flaws but at facts that are already stated in the readme. <irony> Plus, the major source for the article is Pavel Durov, who’s messenger is of course a standard in security and privacy. </irony>

So there seems to be no news but a lot of speculation by Durov instead.

Clouds.

• polar stratospheric clouds play an important role in creating the ozone hole
• the highest clouds on earth are about 80 km high
• in the mid-latitudes most rain is cold rain, that means it leaves the clouds as ice and melts on the way down
• pure water droplets without an aerosol inside (cloud condensation nucleus) freeze at about -40°C, sea salt aerosols make cloud droplets freeze at about -38°C, …

And there’s much more to be found.

To be honest, I simply wouldn’t feel comfortable storing this data on the third party servers hosted on hyper scaler infrastructure. That probably works for most but I’m not keen on a Telekom / AWS combination there. The regulation itself is a totally different discussion. There’s arguments both in favor and against and I don’t really wanna judge that here. I’ll say though, that IMHO the OS level is totally the wrong place to do it. Just gives large non-European companies a powerful bonus datapoint.

Where are the photos of the document processed though? And if not on the phone itself, is the server backend open source as well? Can I self-host it? And is the data which is used to generate certificates deleted immediately or stored in the backend? I have questions.

/rant Advertising broke the internet in so many ways since it became the number one way of monetizing “free” online services. Most popular means of communication (email, messaging, …), news distribution, or any other digital interaction (social media, neighborhood platforms, …) are driven by either selling users stuff they didn’t want from the service in the first place first place or by helping others do that through gathering huge amounts of data not needed for the service. So participation in the digital society partially forces us to play the game as non-profit open source solution typically only cover smaller portions of society. I’m so sick of it. /rantoff

I was wondering if blocking ads and switching off location services would even limit the impact on Webloc much. It’s always surprising how much information ca be pulled from seemingly harmless interactions.

That’s a much reduced statement. Construction noise is indeed a massive disturbance, which is why there’s a lot of efforts on mitigation it. Once the plant is running, however, it has been observed to work as an artificial reef increasing the local populations. To increase the effect there’s some work on adding boulders in between the monopiles. Another reason is that power boats are (depends on country) often forbidden to enter which actually reduces the noise disturbance within the area. There’s also effects on some species (lower trophic levels) due to the local reduction in wind speed in the wake which modifies the amount of mixing in the surface. That is, however, an active field of research. Long story short: fields of wind power plants are depending on the current life stage a disturbance but have in some sites been observed to be the opposite once operating. A conclusion to only build them on land could hardly be drawn at this time.

I was always wondering the same and found the related choices always somewhat inconsistent. My own anecdotal experience was that Ubuntu was always to buggy to actually use it in production (every time I tried, I ended up removing it after a few days only due to driver road blocks and os-related bugs). Moreover, using Debian, as many said before, is a question of feature stability and extensive testing. That’s great for server deployment but not so great for consumer electronics. Thus trying to base a distro on Debian while trying to fulfill the expectation of receiving recent developments in the software seems counterintuitive to me.

Disclaimer: I’ve made a lot of very good experiences with Arch and derived distros and use them in production for more than a decade. Given my not so good experience with Ubuntu I’m certainty on the biased side here.

Nice. I wasn’t aware they managed to assemble all parts with Linux drivers. From all I read (not an expert at all) that seems quite tough. What remains is a runtime environment/infrastructure for everyone who’d like to run Android binaries/packages on the phone.

SFOS is not an Android fork. As many classical Linux distros, it controls sessions through systemd, the compositor is Wayland and the standard c lib is glibc. However, compatibility with the drivers of many hardware vendors and certainly also running binaries build for Android require Android libraries and abstraction layers to be present. Sadly some Android libraries or suitable replacements where available are absolutely necessary to run a phone nowadays. Both hardware and software producers are so focused on Android that drivers / binaries for alternative systems are not offered. That also underlines that this is indeed a niche product. Disclaimer: I ordered one of the phones.

I’m glad to hear that lack of time/resources for code reviews are more common. Also for clarification: I was the author and requested reviews by my colleagues. In reverse, I did not receive requests to review PRs so far. Tbh. I would really like such a review culture as it is already standard in scientific publishing and it would have avoided some obvious bugs we did encounter in the past. Having that said, as I did not receive any review at all and I would appreciate low(er) quality reviews better than none.

It was more a question of finding the time within the daily business of doing science. I suppose it is also a culture issue, the priority to do a review is usually low. In other words, there was de facto never a review.

From a scientific modeler perspective: Always trying to do 5 (or 4), but I’m having difficulties getting a culture of reviewing each other’s codes going. Many times I was asked to “just merge” months after submitting a PR. In the context of operational or large community codes, 5 is usually strictly enforced. Weather services don’t appreciate broken code.