Yeah it really got the Batman atmosphere well. I wonder if it's because Batman was such a cultural icon that even if you "didn't read comics" you till knew who Batman was and what he was about?

I've never encountered a trans person who said "I and only I have ever been treated unfairly." I'm sure someone's said it (the Internet is a big place and full of trolls), but it's a pretty ridiculous thing to say. I recommend that if you ever see it in the future you just shrug at the person who is clearly not speaking in rational good-faith, rather than taking offense.

It's interesting how the English language makes a phrase like "trans women are women" ambiguous about the number of women that each trans woman is... I always assumed "one" but you raise an interesting question.

It's fucked up that the Greens actually went into business as Russian stooges instead of just being an opposition party.

I guess the "pro sanctions" argument would be that countries have very few ways to influence each other: political pressure (asking nicely), economic pressure (sanctions to make the country poor), and military pressure (bombs). And that hey at least it's not bombs (until it is).

But just like with bombs, sanctions will always hurt a ton of people who were just trying to live their lives and did not want to be part of the conflict.

I hope your friend gets by OK.

Cars in general are the problem and even if they all went electric they'd be bad. (But cities would be much quieter and they are hella fun to drive.)

If you're able to use a bicycle for some of your trips instead of a car, that's a good change. (And if you're not then you might not even be able to use an EV car if you could afford it. It takes way longer to charge a battery than to fill a gas tank.)

The "agents" and "agentic" stuff works by wrapping the core innovation (the LLM) in layers of simple code and other LLMs. Let's try to imaging building a system that can handle a request like "find where I can buy a video card today. Make a table of the sites, the available cards, their prices, and how they compare on a benchmark." We could solve this if we had some code like

search_prompt = llm(f"make a list of google web search terms that will help answer this user's question. present the result in a json list with one item per search. <request>{user_prompt}</request>")
results_index = []
for s in json.parse(search_prompt):
  results_index.extend(google_search(s))
results = [fetch_url(url) for url in results_index]
summarized_results = [llm(f"summarize this webpage, fetching info on card prices and benchmark comparisons <page>{r}</page>") for r in results]

return llm(f"answer the user's original prompt using the following context: <context>{summarized_results}</context> <request>{user_prompt}</request>")

It's pretty simple code, and LLMs can write that, so we can even have our LLM write the code that will tell the system what to do! (I've omitted all the work to try to make things sane in terms of sandboxing and dealing with output from the various internal LLMs).

The important thing we've done here is instead of one LLM that gets too much context and stops working well, we're making a bunch of discrete LLM calls where each one has a limited context. That's the innovation of all the "agent" stuff. There's an old Computer Science truism that any problem can be solved by adding another layer of indirection and this is yet another instance of that.

Trying to define a "limit" for this is not something I have a good grasp on. I guess I'd say that the limit here is the same: max tokens in the context. It's just that we can use sub-tasks to help manage context, because everything that happens inside a sub-task doesn't impact the calling context. To trivialize things: imagine that the max context is 1 paragraph. We could try to summarize my post by summarizing each paragraph into one sentence and then summarizing the paragraph made out of those sentences. It won't be as good as if we could stick everything into the context, but it will be much better than if we tried to stick the whole post into a window that was too small and truncated it.

Some tasks will work impressively well with this framework: web pages tend to be a TON of tokens but maybe we're looking for very limited info in that stack, so spawning a sub-LLM to find the needle and bring it back is extremely effective. OTOH tasks that actually need a ton of context (maybe writing a book/movie/play) will perform poorly because the sub-agent for chapter 1 may describe a loaded gun but not include it in its output summary for the next agent. (But maybe there are more ways of slicing up the task that would allow this to work.)

java naturally produces verbose stack traces

I always think of Java as the absolute gold standard of stack traces. Sure, in any given debugging session I don't care about most of the stack. But across all sessions, I've used all parts of the trace and I wouldn't want anything elided.

JS is my least-favorite because it provides a stack-trace so I get tricked into thinking it'll be useful. But since it doesn't cross callbacks it provides no depth.

Hell yeah! When you click over from "this game is impossible wth" to kicking Rodney's ass so much that you'd expect him to just give up...

Nice work cracking a tough game!

Goats sure are neat

enough that voted Trump

Fuck those assholes. The ones that break my heart are the ones who didn't vote at all because "the Democrats are terrible too". Yeah, they are. But voting for the least-bad politician is the minimum-effort thing and everyone I know who skipped voting is not spending their time community-organizing or engaged in violent revolution.

Consent in a situation like this is difficult to establish, to the point of it being pointless. Your comment implies to me that you think if the person said "OK" to a search request then whatever happened next is their own fault.

Consider just the situation where you're in the immigration line and two uniformed officers walk up to you and say, "please come with us." If you go with them, is that voluntary? If you say "yes" I just think "voluntary" doesn't hold much meaning. What happens if you don't volunteer to go with them? Surely, they say, "come with us now or you'll be arrested." And if you don't volunteer at that point, they'll physically restrain you and take you away.

Since most people are able to understand the subtext of the situation, they're able to tell that, "please come with us" actually means "you are required to come with us now. You may either walk of your own accord, or we will take you captive and punish you beyond whatever we initially intended." So, there's not any consent happening. Just deciding whether being beaten and dragged away in public would be helpful to you, and in many cases it is not.

You might be confusing US law around unlawful search and seizure with US law around border crossings. While the ACLU's position is that the 4th amendment trumps CBP, CBP's position is that it does not and that you cannot stop them.