Post

Maybe there is a nuanced discussion to have about using LLMs to write Haskell, but it cannot happen before LLM crawlers stop DDOSing the Haskell Gitlab instance.

I realize this thread is really about the LLM crawlers and wanting them to stop, but: I personally think it would be great for that nuanced discussion to proceed somewhere, and I hope the Haskell foundation leads the way in making it happen.

Here's the starting point: The imaginary friends requires a constant DDoS of an infrastructure maintained by overworked community members, and the owners of the imaginary friends want to sell back intelligence as a service to a class of de-skilled tech workers and/or the managerial class

Let's question the political conditions leading to the emergence of this technology.

But also please send us money if you wish to keep your LLM trained on the Haskell infrastructure, because right now we mostly return 500s to humans when the AI bros learn how to circumvent our mitigation measures

Just yesterday Anthropic announced they were funding Blender, so maybe the LLM companies sending the Haskell foundation money isn't all that far-fetched. Would that actually be a good outcome? Speaking bluntly, if money is all it takes to solve the problem, it's an easy problem.

Remember that in Haskell every "if" needs an "else". ;)

My inference is that the 'else' is that there's probably people who believe that that solution wouldn't be acceptable and would find some other reason it's bad beyond the DDOS. In which case that's the real discussion, not the money or the DDOS.