• 1 Post
  • 90 Comments
Joined 3 years ago
cake
Cake day: June 13th, 2023

help-circle
  • I’m somewhat glad I’m not the only one getting phantom notifications. I got one last week for a deleted event. It took me a minute to figure out what was going on.

    I’ve been debating leaving Tuta and switching to mailbox.org. I’m annoyed by the lack of support for standard protocols like S/MIME and SMTP. As far as I understand with encryption, I can only send encrypted emails to other Tuta users in practice. Yes, I can send emails encrypted with a password. But it kind of defeats the point if the password has to be sent unprotected where an attacker could just use the password to unencrypt the email themselves. The lack of SMTP support also means none of my self-hosted services can send emails.


  • There were a couple times it helped me figure out something I was struggling with if I remember. Either that or I mixed it up with something else entirely. I think it was either when I was learning to use Quickshell to make widgets, or it was something with NixOS. For Nix it wouldn’t surprise me since the documentation lives in ten different places.

    The more I used it, the more I realized that it actually wasn’t that good. Realizing that it was AI made my experience make more sense. Since there’s an element of randomness to AI output, once in a while it actually does a decent job just by chance. The more you use it, the more you realize that the average is much worse than those few lucky times. I guess I got lucky with the first couple pages I read.



  • I didn’t even think about it until you mentioned it, but I’ve had several college assignments where I’m tasked with asking an LLM a question regarding the course, and then I have to write about what I learned from it. I still have to find sources supporting or refuting the output, so we’re not expected to take the output as truth at least. And these aren’t CompSci courses either. It’s common core cultural intelligence stuff.

    When they talk about AI taking over the world, it’s always about taking over the Internet and connected industrial machines. No one told me that AI was going to take over the collective consciousness first.




  • I get that working with legacy systems often makes it hard to implement newer security measures, but at the same time I wonder if the finance sector just doesn’t understand or care about cybersecurity.

    The last couple of times I’ve had to link accounts between different institutions, the fast recommended way is to provide my username and password to the external account to verify my account. You know, the thing that almost any decent site will tell you to never ever do under ANY circumstances when signing up.

    The incompetence is infuriating.












  • The previous 40kW proposal linked in the article mentions it would allow operation where solar panels aren’t feasible, like permanently shadowed areas where water might be. There’s also the dust problem to solve with solar panels, although this would also be a problem for nuclear reactors since their radiators could become less efficient from dust buildup.

    There’s a lot of extra costs associated with making solar panels space worthy. No atmosphere also means no radiation shielding and no cooling. I actually managed to find satellite solar panels for sale: https://www.cubesatshop.com/product/solar-panels/. They have front and back panels but if we assume they didn’t have a back and all panels faced towards the sun it would be ~120W. That gives us $133/W, which means 100kW would be $13.3 million. Unfortunately the mass isn’t listed, so we can’t estimate the launch costs. I don’t have a way to estimate the cost of a fission nuclear reactor on the moon since we don’t know how it would work yet, so this calculation is mostly for fun. That math would change significantly if we are able to manufacture solar cells from lunar regolith.

    In nuclear’s defense, we’ve been sending plutonium-238 into space since 1961. There’s been a few accidents, but the fuel casing has been improved so that the later accidents resulted in no leakage. That was in the early days, so we know a lot more about safety now. Do you think the risks are too high for any nuclear fuels in space, or does uranium pose unique risks?


  • It was a bad demo. I solved every “puzzle” near instantly except the robot, where I didn’t realize the arm had to be on a specific side. The demo was basically the story prologue and intro tutorial, but then it stops without ever giving you a proper puzzle. It didn’t do a good job of showing depth in its mechanics. Each puzzle has a single correct obvious solution. I wanted to optimise, but that single solution doesn’t give anything to work with. Forget single solution though, most levels are a single part. I get that they’re introducing new parts, but you’re allowed to what you’ve already taught even when introducing new mechanics. The lack of interaction between parts is a large part of what makes it seem so shallow. It makes me question if the dev even knows how they interact.

    I didn’t mind the story at first, but it overstayed its welcome. Your “less is more” is an excellent way of putting it.

    Hopefully the full game will be better, but I’m not counting on it.