Why do they care so much? They're a non-profit dedicated to the betterment of humanity via open access to AI. They have nothing to hide. They have no motivation to lie, or lie by omission.
A more interesting question to me (and one where MSFT employees here would have some insight) is to what degree is Windows' recent ABYSMAL fucking quality the result of AI, outsourcing, or bad management? You can also feel the difference in healthy employees vs. unhealthy, when you switch between something like VSCode (polished, fast, intelligent UI, not buggy, consistently improves) and Explorer (paleolithically slow, unstable, buggy, crashy, the worst version is always the latest).
> VSCode (polished, fast, ...) and Explorer (paleolithically slow,
In what world is VSCode fast when its startup time is multiples of Explorer (which had in recent news decided to preload itself to mask that issue) and they are the result of exactly the same fundamental shift from native to web native
VSCode starts up the same as any other IDE, and is responsive and snappy when using it. Explorer starts up about 100 times slower than any other file explorer, and is exceptionally while using it too.
Visual Studio is the one to compare to "any other heavy IDE", vs code is slow as a text editor due to chrome, not its ide features , and there are plenty of code editors with plugins that start much faster, though I won't make up 10000 times, it's only around 10
Now try to follow your own sources, both links for that sentence call it a code editor with MS doing the correct reverence " happy with using a regular code editor like Sublime Text instead of a full IDE like Visual Studio."!
It's also called so on its official page "The open source AI code editor"
So try to make an actual argument/use a proper source for one instead of treating wikis as the ultimate truth
> What IDE with plugins starts faster?
Sublime Text, which is as much an IDE as VSCode
> And I'm not talking about the startup time of VSCode.
You're, you are responding to a point about very slow startup time
>Now try to follow your own sources, both links for that sentence call it a code editor with MS doing the correct reverence " happy with using a regular code editor like Sublime Text instead of a full IDE like Visual Studio."
People call applications apps, and before that they called applications programs. VSCode is called an IDE. My link calls it an IDE. Some people call it a code editor. Some people call emacs and vim IDEs. VSCode does everything an IDE does, including compiling, debugging, and profiling.
>So try to make an actual argument/use a proper source for one instead of treating wikis as the ultimate truth
Just because you don't like it being called an IDE doesn't mean it isn't an actual argument.
Regardless, you haven't and can't attack my central point. The only one I care about, in fact: that the user can feel the vast difference in quality of VSCode vs. Explorer, and can tell that it's made competently vs. incompetently.
I also said nothing about how long it takes to start up--you brought that up. I called it "polished, fast, intelligent UI, not buggy, [and it] consistently improves," all of which are true. All of which you've ignored.
>You're, you are responding to a point about very slow startup time
I'm responding to someone who doesn't understand my argument, and probably doesn't want to.
> People call applications apps, and before that they called applications programs
Those are synonyms, how is it relevant?
> My link calls it an IDE.
Mistakenly because the sources it uses to support that do not call it an IDE
> Just because you don't like it being called an IDE doesn't mean it isn't an actual argument.
The reason isn't likes, but lack of substance. Like when you provide an empty reference to wiki without even reading it yourself to see that the sources don't support the statement
> Regardless, you haven't and can't attack my central point.
I have, your comparison is false, the "paleo slow" app is faster than the "fast" one. Now you claim that "fast" somehow doesn't include the speed of startup!
> All of which you've ignored.
Some of which, and given that you can't see the obvious about objective things like speed, and the common like the difference between an IDE and a code editor, there is no point in addressing more vague points like UI polish.
But also, why do you have the misconception that I should somehow address everything in you list???
Pff, now Microsoft Word probably sends your document to an AI every second and asks it to send back a PNG that highlights every misspelled word, and it overlays that underneath your text.
When systemd came out people said you were a conspiracy theorist if you said it would be anything more than an init system. Now in the year of our Lord 2025 we're discussing "systemd-homed" as if that should ever be a real thing.
This doesn't seem noteworthy. It's called a context window for a reason--because the input is considered context.
You could train an LLM to consider the context potentially adversarial or irrelevant, and this phenomenon would go away, at the expense of the LLM sometimes considering real context to be irrelevant.
To me, this observation sounds as trite as: "randomly pressing a button while inputting a formula on your graphing calculator will occasionally make the graph look crazy." Well, yeah, you're misusing the tool.
It sounds important to me. Humans are where context comes from. Humans do not generally provide 100% relevant context but are generally pretty good at identifying irrelevant context that they've been given.
It seems to me that solving this problem is one approach to removing the need for "prompt engineering" and creating models that can better interpret prompts from people.
Remember that what they're trying to create here isn't a graphing calculator - they want something conversationally indistinguishable from a human.
This should be more of a problem for agents, with less bound context.
But, I would claim it’s a problem for a common use case if LLM of “here’s my all my code, add this feature and fix this”. How much of that code is irrelevant to the problem? Probably most of it.
I agree LLMs are converging on a current representation of reality based on the collective works of humanity. What we need to do is provide AIs with realtime sensory input, simulated hormones each with their own half-lifes based on metabolic conditions and energy usage, a constant thinking loop, and discover a synthetic psilocybin that's capable of causing creative, cross-neural connections similar to human brains. We have the stoned ape theory, we need the stoned AI theory.
I imagine the temperature parameter is the closest thing we have to that right now. Turn it up a little to make things a bit more creative. Turn it up more and eventually it will wildly swing around in its dialogue. Even more and it eventually produces utter gibberish.
Can you describe what kind of stuff you do where it can go wild without supervision? I never managed to get to a state where agents code for more than 10 min without needing my input
Same. I pay for $100 but i generally keep a very short leash on Claude Code. It can generate so much good looking code with a few insane quirks that it ends up costing me more time.
Generally i trust it to do a good job unsupervised if given a very small problem. So lots of small problems and i think it could do okay. However i'm writing software from the ground up and it makes a lot of short term decisions that further confuse it down the road. I don't trust its thinking at all in greenfield.
I'm about a month into the $100 5x plan and i want to pay for the $200 plan, but Opus usage is so limited that going from 5x to 20x (4x increase) feels like it's not going to do much for me. So i sit on the $100 plan with a lot of Sonnet usage.
If you use a single opus instance, you cannot really run out on the 20x plan. When you start running two in parallel, it becomes a lot easier to max out, but even so you need to have them working pretty much nonstop.
That's crazy to me. Maybe i'll give it a try. I find the 5x Opus to be too little to be useful, 4x it seems still insanely small for $100. Wonder if you actually get much more than 4x?
I find I get a _lot_ of Opus with the $200 plan. It's not unlimited, but I rarely cap out (I'm also not a super power user that spins up multiple instances with tons of subagents either, though).
I tend to have two instances going at once often, but i'd be fine with 1x for Opus specifically. Mostly i'm quite limited on how much i can use them because i have to review them pretty hard. Letting several instances go ham for an hour would be far more code than i can review sanely lol.
I’d guess in a sense that it’s on full-auto most of the time with some minimal check-ins? I was wondering how far can you take TDD-based approach to have Claud continuously produce functional code
