Its exactly equivalent to a dictatorship by the head of the CIA, unless the CIA is effectively answerable to some other authority despite not being answerable to the law, and then it is equivalent to a dictatorship by that higher authority.
Yes, and if the hypothetical were that the CIA was effectively outside of control of the law for actions committed in private by CIA personnel in their homes, then the conclusion would be different (even though an agency the scale of the CIA would still have different implications than an individual even then), but that wasn't the hypothetical under discussion, which had much fewer—as in zero—qualifications on the CIA’s lack of accountability.
> if the hypothetical were that the CIA was effectively outside of control of the law for actions committed in private by CIA personnel in their homes
My point is their actions are committed outside the law. They've just been able to avoid punishment by covering it up. What they are not is above the law, at least not in the long run. (There are absolutely short bouts where the CIA acts above the law overseas, and rare cases where it has done so domestically. But the fact that they're covering it up betrays that they're crafty bastards, not invincible ones.)
The CIA ran torture prisons, got caught, then there was a congressional inquiry, and they hacked into the computers of the congresspeople to delete the evidence of torture.
Then they got caught hacking congressional computers to delete evidence.
> CIA ran torture prisons, got caught, then there was a congressional inquiry, and they hacked into the computers of the congresspeople to delete the evidence of torture
One, source?
Two, this above reproach. Not above the law. They deleted the evidence, they didn't just blow the scandal off. (Historically, our IC was popular. Right now, it's the deep state. You're seeing political appointees at the FBI and CIA exert control.)
> agentic capabilities are very much on a roll-your-own-in-elisp basis
I use gptel-agent[1] when I want agentic capabilities. It includes tools and supports sub-agents, but I haven't added support for Claude skills folders yet. Rolling back the chat is trivial (just move up or modify the chat buffer), rolling back changes to files needs some work.
I work on a CAD package for Architects. In C++. It is a native Windows / macOS application.
It's a giant pile of legacy code so a lot of what I do is just C++ generalist stuff, but I have a strong math background so if that's ever called for it's me doing that work (especially because I have English-language skills that don't often come with the strong math background at this pay scale). In particular, I'm the guy wrangling Parasolid (geometry kernel used by SolidWorks, for those familiar) to produce geometry for walls and floors.
Does it have some huge context window? Or is it really good at grep?
The Copilot version of this is just fucking terrible at suggesting anything remotely useful about our codebase.
I've had reasonable success just sticking single giant functions into context and asking Sonnet 4.5 targeted questions (is anything in this function modifying X, does this function appear to be doing Y) as a shortcut for reading through the whole thing or scattershot text search.
When I try to give it a whole file I actually hit single-query token limits.
But that's very "opt-in" on my part, and different from how I understand Cursor to work.
It is really good at grep and will make multiple grep calls in parallel.
And when I open it in the parent directory of a bunch of repos in our codebase, it can very quickly trace data flow through a bunch of different services. It will tell me all the files the data goes through.
It's context window is "only" 200k tokens. When it gets near 200k, it compresses the conversation and starts a new conversation..... which mostly works but sometimes it has a bit of amnesia if you have a really long running conversation on something.
When other models would grep, then read results, then use search, then read results, then read 100 lines from a file, then read results, Composer 1 is trained to grep AND search AND read in one round trip
It may read 15 files, and then make small edits in all 15 files at once
Presumably if it knows it needs to perform multiple searches in order to gather information (e.g. searching for redundant implementations of an algorithm, plus calls to the codebase's canonical implementation) it should be able to run those searches in parallel grep calls.
LLMs are inherently single-threaded in how they ingest and produce info. So, as far as I can gather from the description, either it spawns sub-agents, or it has a tool dedicated for the job.
But LLMs don't really perform well enough on our codebase to allow you to generate things that even appear to work. And I'm the most junior member of my team at 37 years of age, hired in 2019.
I really tried to follow the mandate from on high to use Copilot, but the Agent mode can't even write code that compiles with the tools available to it.
Luckily I hooked it up to gptel so I can at least ask it quick questions about big functions I don't want to read in emacs.
Would be nice to have someone enthusiastic junior to me.
Most of the team is comfortable in their wheelhouse and when new stuff comes down the pipe it's hard to get them mobilized. I had leadership on a big green-field project and felt like we could have really used a junior.
reply