Yeah, there are many problems which are np-hard in theory, but then realistic cases give you way more constraints that make them solvable. So many hard graph problems become way simpler when applied to real maps, because you know that if you start getting away from something, your minimum remaining distance grows. But on an abstract graph there's no real mapping to our dimensions.
There's no "information you wanted" in the plain np-hard version of traveling salesman for example. There's only cost. My point was that things get easier if you have the extra information and aren't solving the plain version anymore.
It doesn't have to be a paradise in absolute terms to be a significantly better place. EE to the US was definitely an improvement for a very long time. Even if many other places already didn't see the US as good guys.
There are other options if you want POSIX. But nushell just does things differently (and in a much more sane way with pipes).
Why would you want `test -e "$foo"` when you have `$foo | path exists`? Why `test -n "$foo"` when you have an actual == comparison and pattern matching expressions? And you don't have to worry about quoting things just right in any of the situations.
Yes, you need to learn a little bit of a new thing. You can even implement a test function itself if you really want to. But sometimes it's nice to just dump the legacy baggage.
Transformers are nice. You can train a very minimal network that can output reasonable sequences very easily. They won't be high quality, or too pretty, but they will "make sense" way more than randomness and (usually) change keys in a coherent way.
Hey viraptor! I haven't even thought about using transformers but that sounds like a great idea. The current generator is just a standard random walk across the major/minor intervals and could definitely use some TLC!
> OpenAI, Anthropic, etc are in a race to the bottom
There's a bit of nuance hiding in the "etc". Openai and anthropic are still in a race for the top results. Minimax and GLM are in the race to the bottom while chasing good results - M2.1 is 10x cheaper than Sonnet for example, but practically fairly close in capabilities.
> There's a bit of nuance hiding in the "etc". Openai and anthropic are still in a race for the top results.
That's not what is usually meant by "race to the bottom", is it?
To clarify, in this context I mean that they are all in a race to be the lowest margin provider.
They re at the bottom of the value chain - they sell tokens.
It's like being an electricity provider: if you buy $100 or electricity and produce 100 widgets, which you sell for $1k each, that margin isn't captured by the provider.
That's what being at the bottom of the value chain means.
I get what it means, but it doesn't look to me like they're trying that yet. They don't even care that people buy multiple highest level plans to rotate them every week, because they don't provide a high enough tier for the existing customers. I don't see any price war happening. We don't know what their real margins are, but I don't see the race there. What signs do you see that Anthropic and Openai are in the race to the bottom?
> I don't see any price war happening. What signs do you see that Anthropic and Openai are in the race to the bottom?
There doesn't need to be signs of a race (or a price-war),only signs of commodification; all you need is a lack of differentiation between providers for something to turn into a commodity.
When you're buying a commodity, there's no big difference between getting your commodity delivered by $PROVIDER_1 and getting your commodity delivered by $PROVIDER_2.
The models are all converging quality-wise. Right now the number of people who swear by OpenAI models are about the same as the number of people who swear by Anthropic models, which are about the same as the number of people who swear by Google's models, etc.
When you're selling a commodity, the only differentiation is in the customer experience.
Right now, sure, there's no price war, but right now almost everyone who is interested are playing with multiple models anyway. IOW, the target consumers are already treating LLMs as a commodity.
There are different types of contrary claims though, which may be an issue here.
One example: "agents are not doing well with code in languages/frameworks which have many recent large and incompatible changes like SwiftUI" - me: that's a valid issue that can be slightly controlled for with project setup, but still largely unsolved, we could discuss the details.
Another example: "coding agents can't think and just hallucinate code" - me: lol, my shipped production code doesn't care, bring some real examples of how you use agents if they don't work for you.
Yeah but there's also a lot of "lol, my shipped production code doesn't care" type comments with zero info about the type of code you're talking about, the scale, and longer term effects on quality, maintainability, and lack of expertise that using agentic tools can have.
That's also far from helpful or particularly meaningful.
There's a lot of "here's how agents work for me" content out there already. From popular examples from simonw and longer videos from Theo, to thousands of posts and comments from random engineers. There's really not much that's worth adding anymore. (Unless you discover something actually new) It works for use cases which many have already described.
The area is still relatively fresh. Those two media personalities do actual work though and provide a summary for today's state. You can wait for an academic research on what happened 6 months ago or a consulting industry keynote/advertisement about what they implemented a year ago... but I'm not sure you'll be better informed.
But since there’s grey in my beard, I’ve seen it several times: in every technological move forward there are obnoxious hype merchants, reactionary status quo defenders, and then the rest of us doing our best to muddle through,
Because some opinions are lazy. You can get all the summaries you want by searching "how I use agentic coding / Claude code" on the web or similar queries on YouTube, explaining in lots of details what's good and bad. If someone says "it's just hallucinations", it means they aren't actually interested and just want to complain.
Heads up that in most cases I bet this'll result in worse performance as you're then but getting the benefit of the Anthropic tuned system prompts that they use in Claude code, which makes for materially different performance of the agent.
Good question. Self-play / self-evolution is usually where these systems either shine or collapse. Curious if you saw convergence or mode collapse when evolving agents on their own generated tasks.
reply