Define objectionable? Not ethical is not illegal but maybe if you are okay with it, do it for yourself. Illegal is just dumb, you are still responsible. So at least, if you are doing, make sure you are appropriately compensated.

The parent was talking about giving a diplomatic answer about your attitude / use of AI. I'm not talking about Enron here.

I have a custom build command for a rust project (yarn build:lib) and my experience is 120k for GLM and roughly 200-300k for Opus. After that, they default to cargo build.

My projects have specific build/verify steps as well, and after a certain point Claude forgets to run them. I’m going to try a “No brown M&Ms” hook to halt Claude if it tries to run the default command instead of the instructed commands from CLAUDE.md. Perhaps this will be a good signal that a compacted or fresh session is needed at that point to avoid mistakes.

I mean, that’s basically the magic of the harness. The whole thing that skyrocketted the intelligence is that the harness (cli tool) prevent the LLM from editing the file before reading it.

Can you imagine even a junior making such a mistake?

As a libertarian, they are not even close to being the same. A bribe is generally optional where you can opt out of the transaction. Taxes are a racket.

So if you get stopped by the police in a bribe-affine country and they make clear you can either pay some money or they will take you to jail even though you did nothing wrong, is that a tax or a bribe?

It's blackmail/extortion. The act of paying the money is the bribe. Here is the Wikipedia definition.

> Bribery is the corrupt solicitation, payment, or acceptance of a private favor (a bribe) in exchange for official action.[1][2] The purpose of a bribe is to influence the actions of the recipient, a person in charge of an official duty, to act contrary to their duty and the known rules of honesty and integrity.

Bribes are optional in countries with taxes. In countries without taxes, they are mandatory.

You’re splitting hairs and fail to see the bigger picture.

I'm working on an online diff tool (https://codeinput.com/products/merge-conflicts) and recently added a mergiraf integration. Basically, the tool loads your git merge but uses mergiraf as the resolution driver. Then add these auto-resolved files to the editor instead of auto-resolving directly.

I also tried out weave, but apart from TypeScript, I haven't found any cases where it actually outperforms mergiraf (I run a bot that watches for new merge conflicts on GitHub, so I've got a steady stream of conflicts to test against).

I reached out a couple months ago on Reddit, but I don't think we ever landed on a time to talk. Would be interested to re-connect again.

This is great, and sorry we never landed a time after Reddit, that one's on me. Let's actually fix it, could you email me on rohan@ataraxy-labs.com.

For context on where the gap might be: weave matches at the entity level (functions/classes/methods by name + type + scope) and uses structural hashing for rename/reorder detection, so it should shine on reorders, renames, and large same-file edits, and may be at parity elsewhere. If your stream tags conflicts by language and type, comparing where each driver wins would be useful to both of us.

> 3. It’s literally impossible to catch Pac-Man? You do not move fast enough and the Pac-Man AI is programmed for perfection so it does not make deliberate mistakes for the human player to take advantage of.

It's not impossible. I just did. You just have to corner the man into an impossible situation. But I agree on the AI-slop or lack of quality production.

The best chinese models are deepseek (general purpose) and glm (coding) and they are both open weight and share lots of their tooling.

There are lots of AI companies and it doesn’t seem that they all have the same funding fountain or share monetization goals. I wouldn’t read much into what each one of them is doing.

Had seen weeks back that the top two non-Western models on ArtificialAnalysis were both closed: https://artificialanalysis.ai/#intelligence-category-tabs

How much stock should we put into that graph, though, I'm not sure.

Even if the models by the Chinese labs are open source or open weights even after they get to mythos level intelligence lets say, still inference and the optimization of those models to be accessed at speeds of 1000 tokens/sec in not in the hands of general public as these models have parameters more than a trillion and they can't be run on some publicly available hardware, So even after being open source it does'nt fix the problem as the general public will still pay the company for inference.

I'm pretty sure these large models are run on Nvidia GPUs, not some unobtainable piece of secret kit. You could go down the street and buy from AMD or a number of other vendors to push out FLOPs if you wanted or needed, but you'll need a thick wallet to shell out for a cluster of GPUs to run these models. The reason people don't run the big Chinese models at home is that they can't afford the hardware, not that it isn't publicly available. This tech is essentially a large amount of matrix multiplications afterall.

I think the larger problem is that restricting US AI companies gives the Chinese a leg up because they now have a window open where they can become the source of the most powerful models available due to government restrictions rather than on technical merits. All Anthropic customers just got a downgrade last evening, for example. While the Chinese are able to serve the world or whoever, the US corporations will be limited to the US market, or whatever the powers that be will allow. This restrictiveness could turn out to be disadvantageous to American companies since people will migrate to wherever they can get the most powerful models.

if it's open source there will be many potential providers, though.

If only we had established means of pooling community resources for the public good

You know a statement like this just makes Chinese big tech look bad right?

> The best chinese models are deepseek (general purpose)

DeepSeek is developed by the largest Chinese hedge fund, their models used to make them $ on the share market are very profitable, they've never ever released anything on those models.

Somehow you are claiming that those same group of people are going to totally change their very consistent long term behaviour and start promoting openness when they are in the global leading position in AI?

selling LLMs is much more profitable than trading, and with much less risk

> much more profitable

I think you made this up.

Right now, I don’t believe any LLM company is profitable at all.

Unless you meant “more profitable” to mean “not-as-badly-negative profit”.

The best is GLM (though it's not as cheap as DeepSeek or Kimi) and use it with Claude Code.

> Knowing how to develop one is not a closed secret but getting in secret is impossible without the whole world knowing.

You can get away with a dirty contamination bomb and that detonating in down town Manhattan will scare the shit out of millions of people even the ones in New Jersey. Or, you know, just fly a plane into a really tall building and get the state you are attacking itself to get into a hysteria breakdown.

But yeah I agree with you. There is no point in these restrictions except for government bureaucrats to gain power and control over a domain.

Did you?

Goal? To trigger LLM safety refusals... so that their spyware wouldn't be analyzed by an AI security scanner

The problem is that they affect people unequally. I do benefit from this stuff because I can navigate the UX struggle but many people can’t and these happen to be the not so fortunate people in society and most vulnerable.

It’s definitely why some stuff is regulated (ie: loan interest)