I think you're best off not upgrading for the time being. It seems like the old auth still works server-side but the the new auth is very, very busted if you are not on a personal plan.
But good news: you can fake out codex-rs into thinking it's not headless by unsetting all SSH environment variables and setting DISPLAY=foo to recover the old behavior.
Was that not clear from the beginning? Nobody ever claimed it was for strategic purposes, the narrative was "we don't like nuclear anything, we will get rid of it we can bear the costs".
So I don't think you could even call it a strategic mistake, but masochism maybe?
Especially while keeping the exit date in the height of the fallout of a real strategic mistake, the dependence on cheap russian gas.
It was a populist move because a big chunk of the electorate is German moms and German grandmas who are absolutely terrified of radiation post Chernobyl.
The fear was widespread across the whole European continent at the time. I don't think you can put the blame on any one person. I think it's entirely natural to be afraid of an invisible undetactable danger that will give you cancer. Many other such fears, due to other environmental pollutions are present today, however justified or not they might be.
Given that wild game in the most affected areas still have to be tested soon half a century after the accident I wouldn't dismiss the fear as unfounded.
Fukushima sure, but a lot of women were traumatized by Chernobyl and the news of a cloud of radioactive dust that was going to give them all cancer. I think Fukushima just reingnited those fears.
It wasn’t that hard to see that energy needs were only going to increase rather than diminish. And not because of ai datacenters, but (to make a simple example) for example because of the already ongoing at the time push for the electrification of the automotive industry.
It’s also crazy that the initiative was supposed at all by environmentalists.
Anyway, props to Mertz for admitting the mistake, we’ll see if they will fix it somehow.
Do you think companies who couldn’t built a safe airport or train station can suddenly built something more complex like a nuclear power plant without massively going over budget, construction time and safety?
And I guess nobody fears Russian drone flying over WECs instead of nuclear power plants
Anyway, props to Mertz for admitting the mistake, we’ll see if they will fix it somehow
That‘s the thing. Everyone knew it was costly, nobody ever thought it was good strategically. If he now says it’s a „strategic mistake“ that‘s laughable, did he think it was strategically clever before? If so he was the only one.
The whole issue is that Germany overestimated its own resilience and economic power, which is deteriorating. Of course environmentalists knew that this is not good for the economy but the Green Party is mostly left aligned they were ok with incurring some damage to the economy for their cause, after all that’s their whole point.
But they thought well we are such a economic powerhouse anyway, we can do it.
So the real strategic mistake was arrogance. And saying that particular action was a „strategic mistake“ instead reflecting on the whole self-image of the country, shows that exactly this arrogance persists
My money is on Russian meddling, to make Germany dependent or Russian gas. Which happened. Until the US blew up the pipeline, and now Germany is dependent on US gas.
You mean the Green party was undermined by Russians?
The Green party had the goal of de-nuclearization from the beginning, at that time the Soviet Union was still in existence. When the Green party came to power and negotiated the nuclear exit, they did not need any external motivation to do so.
The only way I can see this being Russian meddling would be the Green party being infiltrated from Russia from the beginning.
If you have sources that point to the Green party being undermined by Soviet/Russian espionage or some such, please point me torwards them.
The opposite. The (unsubstantiated and probably false) claim is that the Green party was helped or funded by Russian energy companies, who benefited by Germany shutting down its nuclear plants.
Not sure why you're blaming the Greens here, they're a second-tier party in Germany and weren't even a part of the governing coalition during Fukushima and the decision to completely exit nuclear.
The Green party and the Social Democrats were the governing coalition that enacted the nuclear exit. Sure, it was completed by the other two big parties after Fukushima, but by that time the exit was already underway in practice.
Seems you have memory holed how pro Russia the US was too. You guys had joint military exercises. Why single out the EU as being bad?
What you seem to also have memory holed was that up until Crimea, the prevailing idea for Russia was that the more we trade with them, the more wealthy and informed the populace becomes and the more entwined the economy becomes globally and thus losing that access would become too painful to them. The exact same playbook was used for China up till 2016.
That's nonsense. Large parts of the German Left has been incredibly anti-nuclear for 40+ years. And by the 80s they killed further investment. And by the 90s it was clear that nuclear was temporary and was going to be killed.
The right was never anti-nuclear, but they were more pro-gas and pro-coal.
Nonsense. The Greens and all the anti-nuclear were absolutely convinced and never stopped screaming that nuclear was absurdly expensive and the energy price would go down. They over and over again claimed nuclear was bad financially.
I think all of those are terrible indicators, 1 and 2 for example only measure how well LLMs can handle long context sizes.
If a movie or novel is famous the training data is already full of commentary and interpretations of them.
If its something not in the training data, well I don't know many movies or books that use only motives that no other piece of content before them used, so interpreting based on what is similar in the training data still produces good results.
EDIT:
With 1 I meant using a transcript of the Audio Description of the movie. If he really meant watch a movie I'd say thats even sillier because well of course we could get another Agent to first generate the Audio Description, which definitely is possible currently.
Just yesterday I saw an article about a police station's AI body cam summarizer mistakenly claim that a police officer turned into a frog during a call. What actually happened was that the cartoon "princess and the frog" was playing in the background.
Sure, another model might have gotten it right, but I think the prediction was made less in the sense of "this will happen at least once" and more of "this will not be an uncommon capability".
When the quality is this low (or variable depending on model) I'm not too sure I'd qualify it as a larger issue than mere context size.
My point was not that those video to text models are good like they are used for example in that case, but more generally I was referring to that list of indicators. Like surely when analysing a movie it is alright if some things are misunderstood by it, especially as the amount of misunderstanding can be decreased a lot. That AI body camera surely is optimized on speed and inference cost. but if you give an agent 10 1s images along with the transcript of that period and the full prior transcript, and give it reasoning capabilities, it would take almost endlessy for that movie to process but the result surely will be much better than the body cameras. After all the indicator talks about "AI" in general so judge a model not optimized for capability but something else to measure on that indicator
I think the actual problem is everyone tries to assert how capable or not coding agents currently are, but how useful they are depends so much on what you are trying to get them to do and also on your communication with the model. And often its hard to tell whether you're just prompting it wrong or if they're incapable of doing it.
By now we at least agree that stochastical parrots can be useful.
It would be nice if the debate now was less polarized so we could focus on what makes them work better for some and worse for others other than just expectations.
>In the study, researchers told several iterations of four LLMs – Claude, Grok, Gemini and ChatGPT – that they were therapy clients and the user was the therapist
So same as always they tell them to roleplay and when they comply everyone acts surpised...
Based on those, it seems you are not actually using them to create big codebases from scratch, but rather for problems that would normally take quite a while, not because they are inherently difficult to implement, but because you would normally have to spend considerable time on the finicky implementation details.
I think that's the reason why LLMs work so well for some like you, and generate slop for others, because if you let them alone with projects that require opinionated code and actual decision making they most often don't grasp the users intention well or worse misinterpret it so confidently that you end up with something with all the wrong opinions and decisions compounding path-dependently into the strangest and most useless slop.
"for problems that would normally take quite a while, not because they are inherently difficult to implement, but because you would normally have to spend considerable time on the finicky implementation details"
Yes, exactly! How amazing is it that we have technology now that lets us quickly build projects where we would normally have to spend considerable time on the finicky implementation details?
Pretty nice I guess. Cool even. Impressive! And I only say this , just in case, for someone else maybe, ehh—is that it? Because that’s totally fine with me, same experience actually funny that, really impressive tech btw! Very nice. Just, maybe, do the CEOs know that? When people talk of “not having to code anymore”—do they know that this is how it’s described by one of its most prominent champions today?
Not that I mind, of course. As you said: amazing!
Maybe someone just check in with the CEOs who were in the news recently talking about their work force…
> When people talk of “not having to code anymore”
You should reinterpret that as "not having to type the code out be hand any more". You still need a significant depth of coding knowledge and experience to get good results out of these things. You just don't need to type out every variable declaration and for loop yourself any more.
Every single tool or utility you have in the back of your head, you can just make it in a few hours of wall-clock time, minutes of your personal active time.
Like I wanted a tool that can summarise different sources quickly, took me ~3 hours to build it using llm + fragments + OpenAI API.
Now I can just go `q <url>` in my terminal and it'll summarise just about anything.
Then I built a similar tool that can download almost anything `dl <url>` will use yt-dlp, curl and various other tools depending on the domain to download the content.
Another lens is that many people either have terrible written communication skills, do not intuitively grasp how to describe a complex system design, or both. And yet, since everyone is a genius with 100% comprehensibility in their own mind, they simply aren't aware that the problem starts with them.
Well I think it also has to do with communication with LLMs being different to communication with humans. If you tell a developer "don't do busywork" they surely wouldn't say "Oh the repo looks like a trash dump, but no busywork so I'm not going to clean it up, quickly document that as canonical structure, then continue"
I'm so glad I switched to fish, I'd rather have genuinely good settings out of the box rather than endless configuration, and honestly it's much better out of the box than any configuration I've ever had.
Only drawback is that it's not POSIX, no issue for me, but maybe for people who have a lot of muscle memory with bash.
I tried fish for a while but as someone who heavily used bash before I couldn't get used to the new language. I also didn't feel they the language was much better than bash, at least for my usage. But I loved the default automatic coloring of arguments, underlining of files, etc.
Later I found fizsh, which I love and still use as default shell now. It's basically a configuration around zsh adding the colors, completions, and other good stuff inspired by fish to zsh. Can really recommend it for those who are used to zsh or bash but want their CLI to be more readable. Colors especially help with big command line arguments to show where they start and end, and keeping track of complex stuff like loops and conditional logic in your commands.
For POSIX: I leave Bash as the system shell and then shim into Fish only for interactive terminals. This works surprisingly well, and any POSIX env initialisation will be inherited. I very rarely need to do something complicated enough in the REPL of the terminal and can start a subshell if needed.
Fish is nicer to script in by far, and you can keep those isolated with shebang lines and still run Bash scripts (with a proper shebang line). The only thing that’s tricky is `source` and equivalents, but I don’t think I’ve ever needed this in my main shell and not a throw-away sub shell.
I often write multi-line commands in my zsh shell, like while-loops. The nice thing is that I can readily put them in a script if needed.
I guess that somewhat breaks with fish: either you use bash -c '...' from the start, or you adopt the fish syntax, which means you need to convert again when you switch to a (bash) script.
I guess my workflow for this is more fragmented. Either I’m prototyping a script (and edit and test it directly) or just need throwaway loop (in which case fish is nicer).
I also don’t trust myself to not screw up anything more complex than running a command on Bash, without the guard rails of something like shellcheck!
I used to do it this way, but then having the mentally switch from the one to the other became too much of a hassle. Since I realized I only had basic needs, zsh with incremental history search and the like was good enough.
I don't care for mile-long prompts displaying everything under the sun, so zsh is plenty fast.
After 10 years on zsh I finally switched 6 months ago and I haven't looked back. If I need POSIX, I'll just run scripts with the right shebang or pipe it to sh.
First, there are some ways to make fish more compatible with bash.
If you want to do some shell scripting in fish, or running other people's shell scripts (or commands) this may aid you since you wouldn't have to port them (or take less time to port them over).
You can achieve this with a plugin system such as 'oh my fish' or 'fisher'. But, as always, it adds complexity (and bloat :P), you'll need it on every fish shell (including remote systems), etc.
It is a bit akin to having nvim with plugins versus being able to use vi. Sometimes, you're going to need to be able to use the latter.
Also, to people who recently adopted fish: fish has been made more and more compatible with bash throughout those years.
If I use fish, I want to make use of it, and that means I will want to convert scripts, or simply learn to write scripts compatible with fish.
Shell scripts from third parties stick with whatever shell they were written for (ie. /bin/sh or /usr/bin/env bash), and commands copy/pasted from the internet are either quickly executed with bash (one-off) or ported over. Because I like to have such in my history (fish is configured to use atuin), I want to keep using the same shell, so I try to stick with fish. If I cannot convert a command (usually a bunch of commands) to fish, it is PEBCAK and a learning curve/experience.
As for tmux, that is solid advice, because it also allows to stick with a shell which is known to work. I've come to like zellij with alacritty, with zellij the option is default_shell. But now that I use ghostty, I don't use a terminal multiplexer locally any more; only remotely. And there I still use tmux.
The way I actually have things setup, in case it helps. I don't change my default shell. I actually default to pretty much working within tmux. So, I kept my default shell to what the OS brings, then in my tmux config, I have,
# set shell
set -g default-shell /opt/homebrew/bin/fish
This means, that when I start my terminal, it drops me to zsh (macOS default). Then when I run tmux, it opens fish. The nice thing is that I inherit the environment of zsh.
I have my .zshrc and my .bashrc sourcing a .shellrc file which contains most of my env stuff. This keeps random utilities that write to .bashrc and zshrc working within fish too.
I know you‘re joking but to contribute something constructive here, most models now have guardrails against being threatened. So if you threaten them it would be with something out of your control like „… or the already depressed code reviewing staff might kill himself and his wife. We did everything in our control to take care of him, but do the best on your part to avoid the worst case“
I suppose it‘s the latter + maybe some finetuning, it’s definitely not like DeepSeek where the answer of the model get‘s replaced when you are talking something uncomfortable for China
reply