Readit NewsIt is not irrational to cut millions in wasteful spending. It is irrational for a department of government efficiency to spend all of its energy cutting random million dollar contracts instead of figuring out how to plan the attack for the confrontation above.
DOGE did not fail because its a hard problem. They failed because they thought it was easy. Taking over a company with a thousand employees is categorically different than revamping the spending of the United States.
Here are a few outcomes they were able to achieve: (1) Cutting funding for agencies and organizations that were investigating companies run by Elon Musk. (2) Cutting funding for organizations, like NOAA, that have high economic returns for every dollar the government spent on them. (3) Copying information from multiple government databases.
(1) had immediate benefits to Musk. (2) leaves openings for someone with enough capital to fill in the gaps left behind and make a profit charging for what used to be a government service. (3) provides numerous long term benefits to Musk and anyone else with access to that data.
(1) You'd eliminate the system of advantages and supports that cause employers to offer private insurance, which is where most people get their insurance from.
(2) You'd create a huge adverse selection problem --- the more effective/useful Medicare is, the fewer families will want to spent $24k/yr on private insurance, meaning the families left on private insurance have a reason to want it, meaning the composition of the risk pool would shift dramatically.
Like, if we ever did M4A, we'd probably end up with a widespread system of supplemental insurance; we already have it with Medicare! But that's not the same thing as keeping your existing plan.
If M4A plus supplemental insurance gives me about the same coverage I have now for a reduced total cost that sounds like a win to me. Even if it ends up costing me the same amount the net improvement from everyone having access to basic health care would still be a win.
And it would be exactly the kind of political engineering minmax scheme large corps in the US are great at: petition legislators to cut regulations so you can cut costs and maximize profits, but keep juuuust enough of the right perks in the right places so that a slim majority of people in Wisconsin, Michigan and Georgia oppose shaking things up.
While doing so in an awful manner, the current administration is definitely getting things done.
I primarily blame Democrats for the current situation for they have been doing just an awful job of getting anything done or standing up to opposition, they are ineffective cowards and invited the current situation with their incompetence.
I agree with you that Democrats have been ineffective in opposing Republican policies but I think you've come to the wrong conclusion. When someone gets robbed I don't primarily blame them for being ineffective at securing their home, I blame the person who robbed them. Why wouldn't you primarily blame Republicans for pushing bad policies instead of Democrats for being bad at blocking them?
It cost me $50 and I've taken it with me every time I've moved.
I really don't get dropping thousands on a single piece, I've never felt any work speak that loudly to me.
But, I don’t think the idea of just stopping charging works. For example, I had some of their machine image thingies (AMI) on my account. They charged me less than a dollar a month, totally reasonable. The only reasonable interpretation of “emergency stop on all charges completely” would be to delete those images (as well as shutting down my $500 nodes). This would have been really annoying, I mean putting the images together took a couple hours.
And that’s just for me. With accounts that have multiple users—do you really delete all the disk images on a business’s account, because one of their employees used compute to hit their spend limit? No, I think cloud billing is just inherently complicated.
Like LLMs, the bottleneck is still training data and the training regimen, but there's still a demand for smaller embedding models due to both storage and compute concerns. EmbeddingGemma (https://huggingface.co/google/embeddinggemma-300m), released just yesterday, beats the 4096D Qwen-3 benchmarks at 768D, and using the 128D equivalent via MRL beats many 768D embedding models.
Of course, I can't prove that from scratch in a HN comment. What I can do is point out that in the science studying this, it is an uncontroversial fact.
I didn't substantiate that, which made it less convincing, but here is an Economics textbook saying the same thing: https://pressbooks.oer.hawaii.edu/principlesofmicroeconomics...
I know, you can think of an externality. Trust me, Economists can also think of externalities, far more than you or me. In general, they just add interesting nuance to the supply/demand model. They don't completely invalidate it.
But I can't easily demonstrate that, so I suspect I have not changed your mind.