China has roughly .4 AC units per person while the USA has roughly 1 AC unit per person. You are simultaneously arguing everyone should have an AC, and that China should stop expanding their usage of AC.

I'd argue everyone should have an AC if they need one (probably China needs more than they have.) But we shouldn't build any more fossil fuel extraction, people who need AC should figure out how to do it with batteries and renewable energy. (Nuclear is fine, if it makes sense economically.) We don't need population control, we just need to add sufficiently large taxes on things we want less of. AC isn't a thing we want less of, it's carbon emissions.

You're confusing bullshit with jargon, which is something they talk about in the paper. The word synergize has a bad reputation, but its mere presence in a sentence is merely a signal, it doesn't mean the sentence is bullshit.

"We will actualize a renewed level of cradle-to-grave credentialing" is an example from the article - you can't actualize a level, you can't renew a level either. And "cradle-to-grave credentialing" is at best a bad way to describe some real concept. It's word-salad from start to finish. It's not coded language, it's bullshit.

Children have no frame of reference to understand when AI is totally making things up. 1:1 instruction is more valuable than ever to teach children to be critical and verify misinformation that AIs subtly interleave.

> If you wanna get your thing to rewrite curl or something, that's again really weird but fine, but just don't share it or try to make money off of it.

The whole point of the GPL is to encourage sharing! Making money off of GPL code is not encouraged by the text of the license, but it is encouraged by the people who wrote the licenses. Saying "don't share it" is antithetical to the goals of the free software movement.

I feel like everyone is getting distracted by protecting copyright, when in fact the point of the GPL is that we should all share and share alike. The GPL is a negotiation tactic, it is not an end unto itself. And curl, I might note, is permissively licensed so there's no need for a clean room reimplementation. If someone's rewriting it I'm very interested to hear why and I hope they share their work. I'm mostly indifferent to how they license it.

Orion is actually pointless, I don't understand why the mission goals are valuable. Partial success would be meaningless. Success is meaningless.

Starship in contrast has a variety of meaningful objectives. Even if Starship only achieves proving that cryogenic fuel transfer in LEO is possible that's a valuable mission goal in and of itself.

If you really think "the whole moon thing is pointless" NASA is pointless.

> If you really think "the whole moon thing is pointless" NASA is pointless.

There's more to NASA than Artemis! NASA's robotic spaceflight programs generate extremely high science return at relatively low cost. Missions like Psyche, Europa Clipper, and Dragonfly are humanity's real explorers.

And their aeronautics work is valuable as well. Low-boom, etc.

NASA does not seem to be constituted to be able to engage in a coherent manned space program of actual value. It's a long standing systemic issue.

They are great at pretending to deliver value, but there's no "there" there.

Orion doesn't seem operationally or financially capable of launching more than once a year. It's not that they don't want to do test flights, it's that they can barely do anything.

Which goes back to the Pork-on-a-stick requirement that everything be about keeping the workers still employed.

So far, Isaacman's competence has mostly consisted of (rightfully) throwing is predecessors under the bus. The real test will be if there are problems on his watch, but also it seems likely the result of having backbone will not be good for Isaacman and sycophants will end up running the agency again.

> Isaacman's competence has mostly consisted of (rightfully) throwing is predecessors under the bus.

That's irresponsbility and incompentence in any position, especially at that level of management.

How? He essentially said that the program would not work as designed and would probably kill people. That is both true and necessary to say in order to fix it--these are exactly the lessons NASA (allegedly) learned from Challenger.

The GGP said he threw people under the bus. That's different than making changes to a program.

> true

I don't believe you can know that. Saying it with assurance - by Internet randos or by the NASA administrator - is more a signal of a lack of analysis. Other people aren't idiots and complex technology issues aren't that certain - those are self-serving fairy tales.

Running a docker container having side effects on the host seems bad. You've just convinced me a little bit I want podman, and not docker.

Not me, docker is the standard, works great for me, if it didn't, I'd look at alternatives.

Podman also works great but one has to stop trying to use it as if it were docker.

It's a single XML file. Zip sounds like the worst of both worlds. You would need a new schema that had individual files at some level (probably at the "row level.") The article mentions SQLCipher which allows encrypting individual values separately with different keys. Using different keys for different parts of a kdbx sounds ridiculous, but I could totally imagine each row being encrypted with a compound key - a database-level key and a row-level key, or using PKI with a hardware token so that you don't need to decrypt the whole row to read a single field, and a passive observer with access to the machine's memory can't gain access to secrets the user didn't explicitly request.

ZIP files can have block-like relatives to the SQLite page. It could still be a single XML file and have piecewise encryption in a way that change saving doesn't require an entire file rewrite, just the blocks that changed and the updated "File Directory" at the end of the ZIP file.

Though there would be opportunity to use more of the ZIP "folder structure" especially for binary attachments and icons, it wouldn't necessarily be "required", especially not for a first pass.

(That said there are security benefits to whole file encryption over piecewise encryption and it should probably be an option whether or not you want in-place saves with piecewise encryption or whole file replacement with whole file encryption.)

My impression is there's a definite shortage of GPUs, and if OpenAI is more reliable it's because they have fewer customers relative to the number of GPUs they have. I don't think Google is handing out 429s because they are worried about overspending; I think it's because they literally cannot serve the requests.

This sounds very plausible. OpenAI has hoarded 40% of world's RAM supply, which they likely have no use for other than to starve competition. They (or other competitors) could be utilizing the same strategy for other hardware.

Which is worrying, because if this continues, and if Google, who has GCP is struggling to serve requests, there's no telling what's going to happen with services like Hetzner etc.

> OpenAI has hoarded 40% of world's RAM supply

I believe OpenAI's purchasing is somewhat overstated, it definitely has no effect on Google's current ability to serve Gemini requests, but it is obvious that there's a shortage of most components, and it's also obvious that even internally Google is having to make hard choices about who to let use GPUs when.

I definitely think OpenAI likely has less use for GPUs than Google. Google has $300B in annual revenue vs. $20B for OpenAI. Even if you assume 100% of OpenAI's revenue is going to renting GPUs and they are taking a 50% loss there's still a lot of room for Google to be profitable and spending more money on GPUs, and not have enough GPUs. Google also just has a wider variety of models to train and run, from Waymo to Search to whatever advertising models.

https://docs.hetzner.com/general/infrastructure-and-availabi... price increases

OpenAI is dependent on same hyperscalers (most specifically Microsoft/Azure) as everyone else, and even have access to preferential pricing due to their partnership.

A better explanation is to point out that ChatGPT is still far and away the most popular AI product on the planet right now. When ChatGPT has so many more users, multi-tenant economic effects are stronger, taking advantage of a larger number of GPUs. Think of S3: requests for a million files may load them from literally a million drives due to the sheer size of S3, and even a huge spike from a single customer fades into relatively stable overall load due to the sheer number of tenants. OpenAI likely has similar hardware efficiencies at play and thus can afford to be more generous with spikes from individual customers using OpenCode etc.

I would guess the biggest AI product on the planet is Google's Search AI. Although even that might not be the case, unless your definition of AI is just "LLMs" and not any sort of ML that requires a GPU.