CC and Codex are like contemporary televisions. You are the product in this arrangement.
24,550 karma · joined June 23, 2017
You may contact me using my email address, encoded as b64:
cm9iQHplbmlmeS5pbw==
CC and Codex are like contemporary televisions. You are the product in this arrangement.
With custom we don't have to disclose to the caller that the implementation of the tool is another LLM. Each piece of deterministic tool code can firewall the perspective of the model as appropriate.
Think of a sub agent as being isomorphic with a tool call.
Wrap a nested prompt with a tool call that is invoked by the caller prompt. That is the answer to all your suffering. You can repeat this for as many layers of invocation as you desire. You can layer determinism and non-determinism as much as you like in whatever orders you see fit with additional alignment, retry policies, logging, etc.
Agents running concurrently will not go any faster than one well-designed hierarchical agent system.
Imagine the simple task of incrementing a global counter. That would be my #1 benchmark. If you can demonstrate a MAS can increment that counter faster than a single agent can (monotonically without skipping), then I will happily delete 100% of the AI-related code I've authored and start over using your ideas.
I can have my groceries delivered same day via multiple parties. I can have many things from Amazon on my porch the next morning without resorting to aerospace. FedEx can get pretty much anything else to me within 2 days (sometimes involving aerospace).
Where is the demand for this coming from? Does HN feel a burning urge to consume something so essential that it must be delivered now? What the hell is this? A hankering for some snack crackers and a Lego set? Is the demand genuine? I can't help but feel like sometimes this technology is bad on purpose just to antagonize. Like 7000K LEDs being sold to other humans. Evil exists in many formats. Technology companies bring hyperscale to the activity of spreading misery.
I really want to use 6.1 Terra, but the way copilot is set up in visual studio proper, they don't handle reasoning models correctly (or maybe my install is screwed up). Eventually, I will dust off that old winforms project and get my custom harness working with the latest responses API contract.
I prefer (pre) paying for tokens. You can offer me all the proof you want but I'll still never believe that those subsidized tokens are always the same as the ones I get through the API.
Anyone using a swarm of agents in November 2026 to write code is no longer at the frontier of capability.
Running multiple system prompts at once is a perfect example of anthropomorphism. We have this fantasy that a community of diversely aligned agents will somehow outperform a single well-aligned agent. There might be some cases where the random walk takes you where you want to go, but how stable and stressful do you want it to be?
Some of us are using AI as a means to make a living now. If I slip even a little bit on best practices my clients will likely turn into ghosts. Multi agent is an extremely expensive form of entertainment in this context. It's the entire basis for why we have these junkies trolling around HN complaining about how they can never get enough tokens. If I was still playing with these ideas in front of my customers I can guarantee I'd be 100% unemployed right now. They brought me in to make AI into a profit center, not a cost center.
I think linear program tapes are the best substrate if we want to diversify our architectures. Massive GPU farms have definitely won at the neural network game, but they would struggle to execute a highly branchy piece of code faster than a humble single core CPU.
Sequentially following a series of register-based instructions that produces the same effect as Astra would be an unbelievable accomplishment. There is still no conclusive evidence that this type of system cannot be produced. The search space is so awful (on paper) that people rarely try. You have to be a little bit crazy to spend time on this stuff in 2026. However, you will often be rewarded with unexpected results. Never underestimate the power of a good fitness function when combined with 1v1 tournaments and large populations. Even a ridiculously simple ISA can produce amazing results under sufficient evolutionary pressure.
With a Turing machine, you don't need to find the global minimum. That is a strong anti-pattern. You just need to find one of countless really good minima. The search space is not one needle in one very big haystack. A meaningful percentage of the hay is needles.
If it was truly imminent they would claim they already had RSI. That is how the game works. Take a look at IBM and quantum if you require an additional sample. Review the things these companies strongly imply they can do relative to what they can actually do. Anthropic is arguably making less aggressive claims with this RSI narrative.
The lies flourish in the definitions behind these terms. When Anthropic says "RSI" and IBM says "Quantum", they are invoking a highly curated lawyer routine that has a mountain of legal documentation surrounding it.
The mapping between spaces in this case is mostly linear. This isn't like gzip or a Turing machine. I'd argue this is even more permissive (easier) than the LLM's originally intended use case of human language.
What happens if we can come up with a model that can directly emit a frontier LLM given some seed context? I feel like we could train a model to recreate astra verbatim. At some level it's just bytes on a disk that can be bundled up into tokens. As long as the representation is rational and consistent I don't see why this wouldn't work.
Centralization is exactly what you want at this kind of scale. The most marginal participation from the server would make a huge difference.
There is a reason game studios don't use git. It's not just about the large objects.
By overwhelming the supply. How can a consumer keep track of what to play anymore?
As an example, I didn't even hear about PBZ until you mentioned it just now. The fact that Valve might occasionally select lottery winners is separate from other forces involved. I would hesitate to highlight the occasional breakout (artificially-promoted) success as some kind of evidence that there is nothing to see here.
> They just need to make good games.
If only we could just do that.
Perhaps enough indie game studios could align their thinking at the same time to accomplish a similar outcome. In any case, pooling of talent and resources is probably the only way to compete now. The individual developer simply cannot keep up despite being able to do more than they ever were before.
If you seek evidence of Jevon's paradox I think you may find it here. Albeit in a very undesirable form. Even if I can now do 100% more work per unit time, my prospective customer is demanding 300-400% on top of that in terms of the final delivered experience. The water level has gone up everywhere and continues to rise rapidly. Even outside of game development. Deliverables that would have elicited reverence from the business twelve months ago are completely unrecognized now.
> I don’t mind paying a premium
https://www.bhphotovideo.com/c/product/1916990-REG/lg_65ep5q...
Multi-agent is mostly dead. One big agent loop running GPT6.1 absolutely dominates any weird ideas like this now. I still use sub agents for some tasks where I want to compress the action space and context. For example, testing a workflow that requires 100+ tool interactions for each logical test. Without a subagent it's really hard to guarantee a test makes it E2E. I have some very "persuasive" methods for long horizon tests that would break the main loop's dynamics.
We may not feel so compelled to exploit the system if we could figure out how to not be so wasteful. Multi agent schemes can bloat token usage by orders of magnitude. They force us to consider consuming less ideal tokens from contexts and vendors with higher variance.
HN loses its marbles when I suggest it's economically feasible to prepay for API tokens and use these as the exclusive means of performing AI work. I get done with a thousand tokens what my peers cannot do with a million. It doesn't matter how many hammers or workers you have if you don't have the aesthetic preferences and domain expertise to employ them correctly.
With a consumer dehumidifier you might get 5-6 gallons per day. A 5-ton HVAC system will pull over 50 no problem. You might have hundreds of gallons of water soaked throughout the framing, drywall, insulation, etc. It can take a really long time to remediate without a powerful HVAC system.
In the spirit of nuance, the vast majority of the population of Florida is at risk. Something like 10% of the US population in total could be impacted directly. You don't actually have to live on the coast to get rocked by a hurricane. Ike took out power for millions in Ohio.
https://www.facebook.com/100044470387919/posts/ohio-remember...
I am not aware of any economically feasible way to run my 5+ ton HVAC system full-time on batteries after a hurricane just reset all the infrastructure, and you do need to run it full time because of the insane humidity and latent heat that follows a tropical system. Doing the math, I'd need something approaching 800kWh of storage to ride through this kind of event (7+ days). This is approaching half a million dollars in batteries. I can install an extremely nice 1800 RPM generator for ~15% of this cost, which would also include a year's worth of propane buried on the property.
I think a simple generator on an ATS remains undefeated from an engineering perspective. If you have an air cooled unit, the maintenance is not much more complicated than with a lawnmower. You just have one extra spark plug to deal with, and sometimes the oil pressure sensor needs replacing. I can get my stubborn Honda generator to start with all kinds of weird tricks. If my Ecoflow battery starts to act weird, I have very little recourse by comparison.
The FFT indicates this is a periodic signal, not a once a day or week thing. It's happening continuously all day every day. I don't know of any UPS systems that perform tests with this kind of regularity.
In any case, I refuse to believe an external sync signal is causing this. The more likely explanation is that the active elements are pulling each other to a common mode due to being physically coupled on the same grid. This is like how clocks on the same beam will synchronize.
Forcing this kind of safety into the models and harnesses is a mistake. I think the strategy of post hoc moderation is the right one. Trying to prevent evil ideological activities as they transpire is a cursed mission. Let someone who is genuinely doing a naughty thing accumulate a meaningful body of evidence and then send in Mr Holmes (the FBI / police). Attempting to prevent wrong think at every token is too granular. We have systems designed to handle the possibility that crime and punishment may occur at separate moments in time. Most people don't go around murdering because the eventual consequences are pretty severe. Just because an AI gave you advice doesn't magically make those existing consequences vanish.
Using 6.1 Terra for E2E workflow testing feels like cheating. I thought it would have all kinds of problems with the action space on certain steps, but it runs like a deterministic unit test.
I am convinced that OAI has the superior models for most practical scenarios. I've tried Claude recently and it makes me wonder if some of yall aren't actually massive trolls. It's genuinely not in the same room of performance if you place one foot outside junior programming or shiny blog posts about rewriting factorio to run on a ps2.
The demand is what is collapsing. 30 seconds of trashy TikTok music here and there is about all the average consumer demands these days.