Hacker Newsnew | past | comments | ask | show | jobs | submit | robertclaus's commentslogin

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.

>There are tons of examples now of data centers harming communities.

Are there tons like "there are tons of sand grains on the beach", or are there tons like "tons of people run a marathon every year"?

Because as far as I can tell, there are a small handful of adverse datacenter stories that get repeated over and over. What there definitely are "tons" of, are datacenters that people aren't even aware of being right in their backyard for years.


I think it is worth noting that older data centers were not materially moving the electrical grid. That Utah data center is planning to use as much power as the entire state.

Regardless, it still comes down to: things are bad and unaffordable. The powers that be keep claiming that AI is about to take our jobs and end the world. Why wouldn't people be mad?


What’s the acceptable number of people harmed? We don’t really need to wait until it’s statistically significant to worry about it.

Statistically isn't it something like 1 death per $50k profit?

> There are tons of examples now of data centers harming communities.

But there aren’t. At least not in any of the anti datacenter discussions. There are just vague fears of using up the river water and spewing chemicals back in. Then the majority of actual issues people come up with aren’t even data center specific, they are just construction mishaps.



These are clearly written as fear porn rather than serious academic analyses.

The reason they need to be written like that, it because honest takes have a pretty hard time making data centers look close to being as bad as everyday things like golf courses, average car emissions, AC's set at 70, eating meat, normal industrial power demand, or the overnight shipping economy.


I am a bit confused as to why this comment is getting downvoted. Seems like a reasonable take.

It isn't.

They never even held the guy who pooed in all our salads a few months ago responsible.

People died. I stopped eating salad. Where's the accountability?

The current administration seems simply immune to criticism. Let us simply hope that the next administration can understand two things can be true. I'm hopeful the sort of internecine party conflict that provided us with Trump on the right in 2016 can happen for Democrats in 2028 in terms of tearing down the Jeb Bushes and the played out old geezers among the Democrats. These guys are losers imagine losing to Donald Trump for the better part of a decade.

Trump was not seriously on anyone's radar in 2014. Maybe America will be blessed with someone who can realize that AI can both be a blessing and a curse and is willing to do the hard work of making sure public policy creates a future fair for everyone.

If we don't figure it out by 2028 the United States is probably entering an ugly doom loop. The rest of the world is not going to sit around and wait for us to figure it out. The country is too divided. Without a president in 2028 with a mandate to right the ship and write a new path keeping up with our increasingly capable autocratic competitors will not be possible. There's not much else to hope for in America. Another weak neoliberal in 2028 will just give us whatever Trump 2.0 (2.0) looks like in 2032.


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 am not sure that that's recent. Most books, even before the internet, were kind of low-effort shit.

You've always had terrible borderline-porn romance novels at grocery stores and terrible pulp novels and the like. We just don't read terrible novels from 100 years ago; whatever we read from that era now has survived the test of time. The same cannot be said about anything released in the last couple years.


Also, quantity is a quality in and of itself.

Grocery store softcore porn was still very expensive in time, money, and space. The amount of energy required to move a book around is monumental in comparison to how much energy it takes to move the same number of bits of information contained in the book. The speed at which just the information moves (in modern digital networks) is nearly instantaneous. The space the digital book carries is not measurable by the human senses.

It was nearly impossible for most people to afford to fill every waking moment with a book in front of them, bar a few addicts. Now there is a river of information actively fighting for your eyeballs at every moment. And how the meme-gene works, the most addictive types will rise as the victor.


> It was nearly impossible for most people to afford to fill every waking moment with a book in front of them

I'm not sure I agree with that. Public libraries have been around for a very long time, certainly well before the internet. People could get all the books they wanted to read for free. During summer vacations when I was in grade school, I would read multiple books per month, sometimes multiple per week, because we lived close to a public library.


Could you get another book while you were going to the library and back? Would the library not kick you out if you tried to live there?

The difference between a library and the internet is the difference between obsession and addiction.


I don't know about your library, but my library would let me check out multiple books at once; I think 5 was the limit for children and 10 for adults. It was pretty easy to have a constant feed of text being sent into your bain.

I would argue there is some recency: it used to be the case that editors were working with authors in the sense of building their careers. Not even 10 years ago it could be expected an author might not hit it big until their 3rd or 4th book (George RR Martin was like this) and editors were focused on building the skills of their authors in storytelling. Now editors are also pulled in a billion different directions and work more like project managers on way too many books at a time, and things like typos and whatnot are being introduced into traditionally published books with glossy special edition covers.

I think that may be true, but also a thought terminating cliché.

It can be that most books were low-effort shit, but that the problem has become even worse with time.


I generally do not read anything from this century and definitely bothing mainstream, but just going trough what ibooks and book stores offer, it seems that it really is bunch of romantasia and other porny pulp for mainly women (supposedly because men have stopped reading completely). Sure some of it might be ok or even good, I wouldn’t know, but the sheer amount of it makes me doubt anyone actually reads them in full.

Sure, in the 90’s everyone bought John Grisham’s newest thriller which they started and never finished, but at least it kinda had a plot, characters and maybe it tried to even say something (although I might be pushing a bit too far there). It wasn’t good, but at least you read a story, not just barely disguised fanfiction of werewolves having sex.

Or maybe I am just an old man and have no idea what kids are up to. Time for your meds granpa.


I've started avoiding many things created in 2023 or later - you can be pretty sure writing prior to that wasn't some AI slop someone threw together with a half-arsed prompt.

These days I have this vague paranoia that most of what I'm watching, listening to, interacting with is just the product of some meat proxy somewhere trying to put some more half baked crap in front of my eyeballs.

I wish I had a portal to human curated, absolutely non-AI, non-clickbait BS like the very old internet (which still wasn't perfect). I think that dream is effectively dead now. But old books/movies/music are still good. Now I sound like an old grandpa reminiscing about the old days.


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.

LiteLLM is basically Bifrost in the Python ecosystem.


Yeah, but bifrost seems tighter and claims to use way fewer resources


hey - litellm will migrate its core to Rust soon. please follow this issue if you're interested! https://github.com/BerriAI/litellm/issues/31263


also if you had resource issues with your production deployment, keen to hear more details


I don't have specifics since it's been months, but LiteLLM on a hetzner server, deployed through Coolify was using 1-4 GB (cannot exactly, and dont know why either).

Bifrost feels very light, though.

Will try LiteLLM if it's in Rust.

Would also appreciate a slicker aesthetic. I got used to Bifrost shadcn with 0 border radius.


- We recently migrated to shadcn, which also comes with dark mode support. - You can subscribe to the issue or email me at yujong at berri.ai

I’m happy to provide any support if you’re willing to try out the initial version.


Same


The trouble is that the loudest voices will dominate, so the little measurable wins adding up in most organizations don't get attention.


I think fail-over for a production critical service is an equally important responsibility of the router.


You cant fail over to a different model though. Even failing over to a different provider isn't always as simple as set it and forget it. For example bedrock doesn't support the tool search tool feature from anthropic.


I didn't know that! I thought that all models had the same features no matter the provider


They beta support for it now I believe


Yes, providers have downtime and retry/fallback solves this problem.


Ya, probably started with asking for a buzzy title.


Ooh, this is nice. I loved windirstat back in the day.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: