Classic case of not listening to the protests. There are tons of examples now of data centers harming communities. The issues range from overwhelming local power grids to literally poisoning the local water supply. The issue is not the data centers, it's that the government isn't enforcing basic rules already in place for these projects. Until the government or the companies demonstrate they'll protect the adjacent communities the protests will continue.
I think the idea is that the latent thinking space in the LLM will be roughly the same for similar quality results - so the majority of executing well could be stripping back and fine tuning an existing LLM.
Our book club has actually noted that more and more mainstream books are written for low attention span. Repeating details ad nauseum and simplifying the plot significantly. So I don't think books are immune to this automatically.
I'm much more concerned about the odds that this tool is vibe coded and will leak the conversations everywhere other than the LLM agent I've already vetted.
There are simple "ORM"s that just map classes to tables and columns to attributes. Basically focused on serialization instead of query generation. I find those to be a good balance.
I suspect most of the critique even back then was around teaching from static written text, not the writing itself. In my experience that aligns well with modern education theory.
I bet there's an awful lot of servers out there that will happily take CORS requests from any host because someone didn't understand why their second domain couldn't talk to the same API.
Conceptually this is wrapping an agent harness in an LLM call API. I wonder if this format is more digestible than the agent building tools the big labs are rolling out.
A CLI or authenticated web endpoint requires somewhat arbitrary terminal or code access. MCP wraps the functionality in a way that doesn't require nearly the same permissions. Doesn't that enable a whole different class of users?
Interesting how much the post sounds like an AI prompt itself. Are we all going to start talking like that? Think hard, make a plan, and only reply after deep consideration.
There was a post from Github a few weeks ago showing commit volume exploded from linear to exponential growth about 6 months ago. I don't know for sure, but I think they weren't ready for the scale out. Whether that means actual scaling issues or cost cutting because of the scale out, who knows.
Feels like this is missing some of the key points of using generic bucket storage for me:
1. Archive pricing for really large old documents.
2. Cross-provider backups; especially for critical documents.
How much did they end up costing? We do a similar PCB medallion every year for another event and haven't been able to get quite that fancy due to cost. We usually only manage to get some LEDs and a processor in our lower budget range.
We've been doing this with simple mkdocs for ages. My experience is that rendering the markdown to feel like public docs is important for getting humans to review and take it seriously. Otherwise it goes stale as soon as one dev on the project doesn't care.