sosodev

Do you mean agents dating other agents for their own sake or on behalf of their owners?

An excellent quote, but I'm curious, how do you think it applies here?

The knee-jerk reaction reaction to Moltbook is almost certainly "what a waste of compute" or "a security disaster waiting to happen". Both of those thoughts have merit and are worth considering, but we must acknowledge that something deeply fascinating is happening here. These agents are showing the early signs of swarm intelligence. They're communicating, learning, and building systems and tools together. To me, that's mind blowing and not at all something I would have expected to happen this year.

The browser works shockingly well considering it was created in 72 hours. It can render Wikipedia well enough to read and browse articles. With some basic form handling and browser standards (url bar, history, bookmarks, etc) it would be a viable way to consume text based content.

> The security hazards of artisanal hosting are more real than ever

How could this possibly be true? It's not at all rocket science to create a static blog and serve it via a production grade web server (nginx, etc).

> The UX of DNS hasn't improved (it's still impossible for normies)

The UX of DNS sucks but we're talking about a single A record. Is that not within reach of a normie in the age of AI?

> Custom domains are not just vain, they're ephemeral. Certainly more so than, say, the domain of a blogging platform that's managed by a non-profit.

I can't think of a single free blogging platform that has stood up to the test of time. Depending on centralized resources, particularly when you're not paying for them, is the recipe for ephemerality. If you're going to pay for it why can't you afford a domain?

The data does not say that 53% have a conviction or charge. It says that 27% do.

The 26% you miscategorized are people with pending charges. Everyone is innocent until proven guilty.

I'm not saying that pre-trained only models are useless. They've clearly extracted a ton of knowledge from the corpus. The interface may seem strange because it's not what we're accustom to but they still prove valuable. Code completion models, for example, are just LLMs that have pre-trained exclusively on code. They work very well despite their simplicity because... the model has extracted the signal from the noise.

Except it's not an impossible request. If my manager told me "fix this code with no questions asked" I would produce a similar result. If you want it to push back, you can just ask it to do that or at least not forbid it to. Unless you really want a model that doesn't follow instructions?

Are you sure about that? There's a lot of slop on the internet. Imagine I ask you to predict the next token after reading an excerpt from a blog on tortoises. Would you have predicted that it's part of an ad for boner pills? Probably not.

That's not even the worst scenario. There are plenty of websites that are nearly meaningless. Could you predict the next token on a website whose server is returning information that has been encoded incorrectly?

I think you're confused about the training steps for LLMs. What the industry generally calls pre-training is when the LLM learns the job of predicting the most probable next token given a huge volume of data. A large percentage of that data has not been cleaned at all because it just comes directly from web crawling. It's not uncommon to open up a web crawl dataset that is used for pretraining and immediately read something sexual, nonsensical, or both really.

LLMs really do find the signal in this noise because even just pre-training alone reveals incredible language capabilities but that's about it. They don't have any of the other skills you would expect and they most certainly aren't "safe". You can't even really talk to a pre-trained model because they haven't been refined into the chat-like interface that we're so used to.

The hard part after that for AI labs was getting together high quality data that transforms them from raw language machines into conversational agents. That's post-training and it's where the armies of humans have worked tirelessly to generate the refinement for the model. That's still valuable signal, sure, but it's not the signal that's found in the pre-training noise. The model doesn't learn much, if any, of its knowledge during post-training. It just learns how to wield it.

To be fair, some of the pre-training data is more curated. Like collections of math or code.

In theory NPUs are a cheap, efficient alternative to the GPU for getting good speeds out of larger neural nets. In practice they're rarely used because for simple tasks like blurring, speech to text, noise cancellation, etc you can get usually do it on the CPU just fine. For power users doing really hefty stuff they usually have a GPU anyway so that gets used because it's typically much faster. That's exactly what happens with my AMD AI Max 395+ board. I thought maybe the GPU and NPU could work in parallel but memory limitations mean that's often slower than just using the GPU alone. I think I read that their intended use case for the NPU is background tasks when the GPU is already loaded but that seems like a very niche use case.

Nope. Pretraining runs have been moving forward with internet snapshots that include plenty of LLM content.

Hallucinations generally don't matter at scale. Unless you're feeding back 100% synthetic data into your training loop it's just noise like everything else.

Is the average human 100% correct with everything they write on the internet? Of course not. The absurd value of LLMs is that they can somehow manage to extract the signal from that noise.

That idea is called model collapse https://en.wikipedia.org/wiki/Model_collapse

Some studies have shown that direct feedback loops do cause collapse but many researchers argue that it’s not a risk with real world data scales.

In fact, a lot of advancements in the open weight model space recently have been due to training on synthetic data. At least 33% of the data used to train nvidia’s recent nemotron 3 nano model was synthetic. They use it as a way to get high quality agent capabilities without doing tons of manual work.

He asked the models to fix the problem without commentary and then… praised the models that returned commentary. GPT-5 did exactly what he asked. It doesn’t matter if it’s right or not. It’s the essence of garbage in and garbage out.

I think you're overestimating how much people care about quality.

It already exists. Tailwind has had GitHub sponsorships enabled for years but only 5 people have ever given them money that way.

The paid products Adam mentions are the pre-made components and templates, right? It seems like the bigger issue isn't reduced traffic but just that AI largely eliminates the need for such things.

While I understand that this has been difficult for him and his company... hasn't it been obvious that this would be a major issue for years?

I do worry about what this means for the future of open source software. We've long relied on value adds in the form of managed hosting, high-quality collections, and educational content. I think the unfortunate truth is that LLMs are making all of that far less valuable. I think the even more unfortunate truth is that value adds were never a good solution to begin with. The reality is that we need everyone to agree that open source software is valuable and worth supporting monetarily without any value beyond the continued maintenance of the code.

I’ve spent a little bit of time testing Minimax M2. It’s quite good given the small size but it did make some odd mistakes and struggle with precise instructions.

I’ve spent some time thinking about privacy and LLMs. I developed the impression that encryption isn’t meaningful in this space. It seems like end to end encryption only truly works when both ends are outside of the system and can manage their keys independently of it. In this case one end is the system. My message has to be decrypted for processing by the LLM. So is “end to end encryption” in this case any different than HTTPS? It doesn’t seem like it

I'm something of a data scientist for a community college. Most of the problems are social not technical but I am still writing code often enough.

It’s refreshing to a see single optimistic take in this thread

Prosecution for insider trading in 2026? I highly doubt that

I suspect that 2026 will be the year we see a big breakthrough in the use of LLM agent systems. I don’t know what that will look like but I suspect the agents will be doing meaningful research (probably on AI).

You do know this reads the same as every pessimistic commentary on technology ever, right? So many people were convinced that television was going to fry our brains.

Nvidia released Nemotron 3 nano recently and I think it fits your requirements for an OSS model: https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B...

It's extremely fast on good hardware, quite smart, and can support up to 1m context with reasonable accuracy

Did you see this HN submission? https://news.ycombinator.com/item?id=46242838

It seems similar to what you're describing.

Yes, and I'm a little ashamed to admit my morning routine wasn't the same without it.

The claim that a small, fast, and decently accurate model makes a good foundation for agentic workloads seems like a reasonable claim.

However, is cost the biggest limiting factor for agent adoption at this point? I would suspect that the much harder part is just creating an agent that yields meaningful results.

I love how detailed and transparent the data set statistics are on the huggingface pages. https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B...

I've noticed that open models have made huge efficiency gains in the past several months. Some amount of that is explainable as architectural improvements but it seems quite obvious that a huge portion of the gains come from the heavy use of synthetic training data.

In this case roughly 33% of the training tokens are synthetically generated by a mix of other open weight models. I wonder if this trend is sustainable or if it might lead to model collapse as some have predicted. I suspect that the proliferation of synthetic data throughout open weight models has lead to a lot of the ChatGPT writing style replication (many bullet points, em dashes, it's not X but actually Y, etc).