Conflicted. Obviously open access is great, but it's never been that difficult to find most papers either on arxiv or the author's website. And I despise the idea of paying to publish, especially since unlike other fields the "processing" required for CS papers is minimal (e.g., we handle our own formatting). FWIW, USENIX conference papers are both open access and free to publish.

My understanding is that this is at least to some degree in response to the surge of AI generated/assisted papers.

> e.g., we handle our own formatting

I used to work for a small publisher some years ago, and while this is true to some degree, we spent a lot of effort doing additional formatting or correcting formatting mistakes. For a typical journal publication, this process alone takes weeks if you're aiming at a high-quality publication.

On top of that, there are a lot of small things that you typically don't get if a paper is just put on the author's website, such as e.g. long-term archiving, a DOI, integration with services like dblp, metadata curation, etc.

Now, to what degree these features are an added value to you personally varies from person to person. Some people or even workshops are totally fine with simply publishing the PDFs written by the authors on a website, and there's nothing wrong with that, ymmv.

The Digital Library contains a lot of older material which predates the Web and has often never been put online anywhere else: old Joint Computer Conference papers and so on.

> My understanding is that this is at least to some degree in response to the surge of AI generated/assisted papers.

ACM started this open access effort back in 2020, I don't think that LLM generated papers were on their mind when they started it.