Hacker News Reader

Comment by csoups14 | Hacker News Reader

Would imagine it's the simplest answer: they're flying by the seat of their pants, there's 1000 things happening every day that demand attention and there's not enough of it to go around. They toss their LLM at it, give it a cursory glance, and ship it. A quick glance at the Claude Code source code bears the result of this process out. The fundamental question is, if their model is so powerful, why do they keep fucking up such simple things? We're led to believe this is a serious company with a model so powerful they can't release it to the general public.

View on HN

Hermes is one of these OpenClaw clones, so this was certainly intentional, not a model hallucinating something.

I think the problem is clear. Anthropic saw their usage go up much more than their capacity could handle. There are a few tried and true solutions to this, like "increase the price" or "restrict signups so you can guarantee service to what you have already sold".

Then there is the "large scale fraud" option, where you materially change and degrade the service you have already sold. Just because you have obfuscated and mislead in how you describe the product you are selling doesn't mean you get to capture the cash flow of 1 year subscriptions then not honor that contract for the full duration.

> Hermes is one of these OpenClaw clones

So that's what it is. Reading its README I thought it was another harness like Pi [1], but with built-in memory so it remembers what it learns, and gets more capable the longer it runs.

Like Letta [2], Dirac [3][4] and the other "more experimental harnesses that look interesting but I haven't had time to try out".

1. https://pi.dev/

2. https://www.letta.com/

3. https://dirac.run/

4. https://news.ycombinator.com/item?id=47920787

Late in replying to this, but just wanted to say I found this pretty compelling. I generally think people are too quick to assign to malice what could be assigned to incompetence. In this case I'm not convinced of that anymore especially given their public statements about these third-party harnesses. It does seem unavoidable that they'll have to move away from subscription-based pricing and towards token-based, but they're managing this in a really ham-fisted and user hostile way regardless.

Mind pointing out where exactly in the contract you were allowed to use OpenClaw?

Non-Claude client access is not permitted in the terms and conditions, except via API key.

The correct implementation of this condition by Anthropic on the server side would be to block usage by non-Claude apps via Claude's authentication mechanism, and allow it via the per-token API key billing.

Instead of a simple 403 error, which would block usage, they silently redirect to a different billing bucket, which is not ethical behaviour especially since it is based on fuzzy heuristics.

Yeah, at the least it should alert the user that it is happening. Maybe the thinking was alerting it gives people signal on how to get around the restrictions, but having it silently charge from a different bucket isn't the answer either.

I think part of the issue is they were letting people use plan's API for random stuff, so people could do testing or small projects. Then the agents came along and exploded the cost, so they want to restrict those but still let some other usage, which I don't think is tenable.

I'm sure there is some way that they could enforce that all calls are coming from the Claude app or Claude Code. It might be hard to 100% enforce, with stuff running on a user's machine, but they could still could make it quite difficult, where someone has to be intentionally trying beat the system (like stealing encryption keys out of the Claude Code binary or something).

I doubt an AI would be stupid enough to write code like that without being explicitly prompted to do so. It's so... specific.

That specific nature would mean it would get caught by even the most cursory of code reviews.

Even if I was just "scanning my eyeballs over the code" without properly reading it, this would jump out as very odd and make me pause.

They've said publicly that they don't want apps like OpenClaw (Hermes is a variation) being used with a monthly plan vs per-token billing. The problem is this was implemented pretty badly (trying to regex??). And they should put a firm boundary between the two. It shouldn't be trying to switch over to a different billing plan automatically using the same api key.

I think they wanted to try not to totally lock down the monthly plan for non-agent uses, but that makes it all too fuzzy. They should use some specific method like encrypted signatures or something, so that anything sending to the monthly plan that isn't Claude Code or the desktop Claude app just errors out and be done with it.

Reply on news.ycombinator.com