They do not at any point outline how cooling will be done, they simply say "it will be more efficient than chillers due to the larger delta T" which is incorrect because it's about dT not delta T

> "The only logical solution therefore is to transport these resource-intensive efforts to a location with vast power and space. I mean, space is called 'space' for a reason. [crying laughing emoji]"

This is all the reasoning provided. It is quite sad how a company I admired so much has become embroiled in financial doohickery.

It’s not SpaceX’s fault. It’s still a company to admire, it’s just that nobody appears to be able to stop Musk.

I wonder why SpaceX investors aren’t revolting.

> I wonder why SpaceX investors aren’t revolting.

Because if SpaceX were valued like a normal company, they would lose their money.

SpaceX, as technologically awesome as it is, simply cannot be that big of a company because the market for space launches is relatively small.

SpaceX is targeting an IPO at a valuation 500x earnings. They need to jump on the "AI" / datacenter bandwagon to even hope to sell that kind of valuation.

The whole "datacenters in space" thing is an answer to the question "what could require 1000x the satellite launches that we have now?"

It has nothing to do with what makes sense economically for datacenters!

I wonder why the SpaceX top management is going along with this. It's clearly not in their long term interest to do so.

From what I understand, Musk owns 42% of the company and nearly 80% of the voting rights, so I guess that answers the question.

I guess this is the price you pay for buying shares with less voting rights.

Their CEO is revolting.

new trolley problem just dropped: save 1 billion people or ...

I wrote an article on this exact issue (albeit more simpleminded) and I suggested a rudimentary way of tracking provenance in today's agents with "reasoning traces" on the objects they modify.

Would love people's thoughts on this: https://0xmmo.notion.site/Preventing-agent-doom-loops-with-p...

The original article does a good job of contextualizing the shifting dynamics, but yours turns that into an actionable solution. I've been wondering about this same problem too after having trouble wrangling LLMs to not make hacky solutions or go on wild goose chases.

Do you have a working implementation for this? Just a one-to-one index of files and reasoning traces? I'd like to trace these changes easily back to a feature or technical spec too (and have it change that spec if it needs to? I suppose the spec would have it's own reasoning trace)

If recording object change is important, then have the subject object know one or more recorded “change” objects. An LLM is much more likely to understand a real object modeling pattern, rather than some new non-standard scheme such as you suggest.

It is not about tracking changes, git does that well enough.

It is about tracking the reason for the changes, i.e. git messages but on steroids.

I don't think 0.38s is a bad trade-off for convenience when the rest of the tools I need to do my job collectively are another 2s at shell startup. NVM alone adds 0.5-0.6s on my M4 Macbook Air.

Bigger fish to fry if we're being practical.

You can replace nvm with https://mise.jdx.dev/ , it starts effectively instantly and works not only for node versions but all programming languages and tools.

`mise use node@latest`

or asdf

I'm using Asdf too. Have heard well about Mise, but I haven't figured out what the difference is, so I have stayed with Asdf

mise has more features - it is a super set of asdf. For example it can set your env vars when you cd into a directory (like direnv). It also has tasks (which I haven't used) - they appear to be similar to what a Makefile does. So you can potentially replace three tools (asdf, direnv, make) with one.

Thanks! Sounds interesting!

Mise started out using the same plugins as asdf, mostly focused on adding performance and usability improvements. Over time it added more features and security.

Most tools are now directly fetched from github releases without the need for random shell scripts (which is what asdf plugins are).

It also grew to be a task runner and environment manager. At first you might think this is scope creep but they're both opt in and very elegant additions. I don't want to ramble but let's just say they've solved real problems I've had.

I'm a fan of it, and I can't think of a reason why I would use asdf over mise. Its real competition is nix (+devbox/devenv/flox), devcontainers, and pixi.

I will check it out! Thanks for the explanation! Sounds really neat

I am seeing a phenomenon of people wanting to hyper optimise their workflows. It’s nonsensical when you consider the other stuff you need to do or how slow everything else is.

Holy heck, I just profiled my zsh initialization and nvm was the big source of bloat, holy hell. Similar setup as you (M4 MBP), same amount of startup bloat. Lazy loading it fixes it.

GPT-5.2 changed the nvm initialization script to:

  export NVM_DIR="$HOME/.nvm"
  
  _nvm_lazy_load() {
  unset -f nvm node npm npx _nvm_lazy_load
  [ -s "$NVM_DIR/nvm.sh" ] && . "$NVM_DIR/nvm.sh"
  }
  
  nvm() { _nvm_lazy_load; nvm "$@"; }
  node() { _nvm_lazy_load; command node "$@"; }
  npm() { _nvm_lazy_load; command npm "$@"; }
  npx() { _nvm_lazy_load; command npx "$@"; }

That's why I switched to Mise (https://mise.jdx.dev). Having new terminal sessions taking up hundreds of milliseconds because of nvm isn't the end of the world, but it's annoying when you're in the middle of something.

Mise reads in your .nvmrc files so you don't have to configure anything new.

Although this is what Opus recommends, it will give you many issues as you don’t really have any node runtimes in the path (or worse if you do).

What I recommend is replacing it with $PATH=(a command to find the nvm default alias directory, detect the verion and load it from that specific version directory directly) so you always have default node in path and then lazy loading only nvm itself, so you can switch when you need to.

Sorry I don’t have the command handy as I’m on mobile but if you paste the above into Opus you’ll get it.

Thanks. I modified my comment to disclose that gpt 5.2 generated that modification. I'll try your suggestion.

..have you checked. It's 0.04s in my setup. I have it a bit customized but I don't remember chasing milliseconds on config

> All encryption is end-to-end, if you’re not picky about the ends.

This reminds of how Apple iMessage is E2E encrypted, but Apple runs on-device content detection that pings their servers, which you can't possibly even think of disabling. [1][2]

[1] https://sneak.berlin/20230115/macos-scans-your-local-files-n... [2] Investigation in Beeper/PyPush discord for iMessage spam blocking

[1] is supposedly debunked: https://pawisoon.medium.com/debunked-the-truth-about-mediaan...

> the network traffic sent and received by mediaanalysisd was found to be empty and appears to be a bug.

I say "supposedly debunked" because empty traffic doesn't mean there's nothing going on. It could just be a file deemed safe. But then the author said:

> The network call that raised concerns is a bug. Apple has since released macOS 13.2, which has fixed this issue, and the process no longer makes calls to Apple servers

What’s the concern here? The blog post you linked does not really support its claims with evidence.

They're actually two separate claims, one of which the blogpost does support. The other one is seemingly ought to be supported by some conversations on a Discord server.

The concern is obvious though, not sure what's unclear about that: it's a bit pointless to have E2EE, if the adversary has full access to one of the ends anyways.

I built a connection to a web-powered LLM over SMS/iMessage for literally this purpose. While traveling I’d have really bad or sparse service but still needed to find my way around.

I use WhatsApp's built-in LLM to read news when I'm on long flights that only give messenger access. It's great

That’s neat - have you considered publishing it?

I did, it's available at olly.bot :)

Doesn’t ChatGPT have WhatsApp access?

Not anymore. Meta didn't want competitors.

the other day I had to change my node server to prefer ipv4 dns records because fly.io doesn’t support outbound ipv6 connections but defaults to a dns server that returns them

Their document states they support v6, and given how much of their stack involves v6 I would be shocked if they didn't support v6 outbound.

> Outbound IP addresses

> Fly Machines have IPv6 addresses from which they make requests to the wider internet without going through the Fly Proxy.

https://fly.io/docs/networking/services/

Right about $500 now, down from $5K at peak

AI assistant in your iMessage group chats https://olly.bot

How did you make a bot that can send iMessages? I want to make a service on my Mac that can send notifications through iMessage. (preferable multiple chats, different topics for different notification types)

seems to be a few apps that can do this, but you need a mac to do it

https://github.com/linjunpop/imessage https://github.com/danikhan632/iMessage-API

Post — “Ask HN: Can you write a limerick?”

Top Comment — “This reads like someone who just discovered poetry forms exist and thinks a limerick is some novel concept. The real challenge isn't writing one—any undergraduate can follow the AABBA scheme—it's understanding why meter and scansion matter beyond just counting syllables.

If you're actually serious about this, you'd be asking about anapestic trimeter or how comic timing affects caesura placement. The fact that you're not suggests you haven't done the groundwork.”

Truly captures the spirit of these types of HN comments; Person A does a thing, Person B points out how pointless thing could have been done better in effort to flex smart.

Post - Blog post about recapping a Timex Sinclair 1000"

Response - "Ah yes, the 'multi-region composite mod'—because nothing screams cutting-edge like jury-rigging a 40-year-old potato to a VCR."

> Truly captures the spirit of these types of HN comments; Person A does a thing, Person B points out how pointless thing could have been done better in effort to flex smart.

And then Person A goes off and founds Dropbox and 20 years later is worth $2.4 billion.

How dare you bring facts or narrative into question which dunk on the original commenter's upvote-worthy comment

So many open with “ah yes,” or “oh great,”

> The real challenge isn't writing one—any undergraduate can follow the AABBA scheme—it's understanding why meter and scansion matter beyond just counting syllables.

There was a young man from Japan

Whose poetry didn't quite scan

When told this was so

He said "Yes, I know..."

"... it's probably because I try to cram as many syllables into the last line as I possibly can!"

and also

There was a young man from Wick

Whose limericks were twisted and sick

It's best not to mention

How he broke convention

...

I know "lol" type comments aren't super typical or accepted on HN but I need to reply just to acknowledge that this comment made me legitimately laugh out loud in the workplace LOL (good luck explaining that one to my non-tech coworkers xD)