9,912 karma · joined September 1, 2011
Given the original mission statement of OpenAI, is that really a conflict of interest?
Having said that, it's clear that the 'Open' in 'OpenAI' is at best a misnomer. OpenAI, today, is a standard commercial entity, with a non-profit vestigial organ that will now be excised.
There is no such thing as a 'moderate amount of storage'. The best you can get is pumped storage (if you have the correct geography for it).
Outside of that, sure, you can add solar/wind to the energy mix - does that invalidate my point? Does that actually support OP's point?
Why? What current problems would changing this fix?
Also also, we've had powerful summarization techniques for decades now (deep-learning, or otherwise) - and they aren't susceptible to the hallucinations of present day Generative LLMs.
You can't retroactively pull an open source license ... whatever the new license is, if it forces you to pay, it no longer meets the definition of 'open source' [1]
I see you had a typical customer service experience with Amazon.
Out of curiosity, how do intellectual property laws handle someone publishing a work under someone else's name? Is that a trademark violation?
Twilio doesn't make billions in profits. Twilio has been losing money for years. I'm not sure if Twilio was ever profitable?
Surely that wasn’t the goal, though it echos the well-intentioned but flawed “everyone should learn to code” initiatives.
You need the right temperament for programming, and some people find it mind-numbing, in the same way I find reading legal documents mind-numbing.
No, it isn't a fact. This assertion is a common internet troll trick that relies on misusing the "product property of square roots" [i.e. sqrt(xy) = sqrt(x)sqrt(y)], with the "trick" being that the "product property" is only applicable for x,y >= 0 and x,y ∈ ℝ, and therefore cannot be used to "prove" 1+1=0.
Why purposely mislead people who may take you at your word?
It's not quite that. It's a semantic issue. Colloquially, people think of "the closest planet" as a planet having an orbit closest to that of earth's, whereas the article proposes a different definition of 'average distance' between the planetary bodies. Both definitions are correct, but I argue that the former definition is actually better as it conveys more information about the structure of the solar system, and is more intuitive.
Obviously wrong how? I saw that Musk was claiming that launch costs for starship will be around $1-2 million at some point - probably after hyperloop launches as a "5th mode of transport" in Neverland.
Or they can't, because there are physical limits to the chemical rocket technology.
>There is room to reduce cost much further. Discarding the second stage is a large cost ... with full reusability costs should fall at least another order of magnitude.
Reusability is not a panacea. Falcon 9 is partially reusable and it is not significantly (if at all) cheaper than the competition. There is no evidence that full reusability will provide an order of magnitude cost decrease in prices. Fully reusable rockets imply that you're cutting into your cargo space (because you have to ship extra fuel for the decent), and maintenance is expensive and time-consuming. Don't get me wrong, there may be costs savings there, but not 10x, more like 10%.
There is downward pressure, we're just hitting physical limits on what can be done with chemical rockets.
Falcon 9 is around $60 million/launch (though when contracted by NASA or the military, launches tend to be 2x-4x the price). This is pretty much inline with what you get from Arianespace, and Roscosmos. There is no path to reduce price by another order of magnitude. We've pretty much hit the limit on what we can do with chemical rockets.
Well … yeah. It’s education at scale. Anything at scale will be inflexible by definition.
> San francisco is rich, people make a lot of money there.
And there’s a lot of funding. The median per student funding in the SF school board is something on the order of 2 or 3 times the median US funding (a quick Google shows ~$22k vs $7k) and when compared to global averages, it’s through the roof. It’s not the level of funding. It’s not even the administration and curriculum either (remember, our bar is very low - we just want kids that can read when they graduate).
The way I figure, optimizing public education is a last-mile problem. There is value in taking kids from a B to an A, or providing a more enriching educational experience. So don’t think I’m arguing against more funding and better management. But this isn’t going to solve the core issue - parental neglect (and it is neglect if a kid graduates high school illiterate). At scale, the state cannot make up for that kind of neglect. In fact, that’s a big reason why you see the divide growing - everything that is tried, just ends up marginally improving or marginally deteriorating educational experience of kids who are already doing reasonably well.
> But while impossible to eliminate, you can try to minimise it and you should.
And my argument is, we’ve largely hit the limit on this kind of optimization. Spending more, and/or changing the curriculum with the same cohort of teachers, isn’t going to give you anything better than trivial (at best) improvements.
That’s the current theory and it fails. The reality is that public eduction is education at (massive) scale. There is no amount of available resources to create, say, a personalized learning environment with dedicated teachers for individual students - when you’re dealing with millions, or tens of millions of students. Parental involvement is so critical because that is the individualized focus that is needed by kids.
There is no way for you to create a public system that overcomes the success that comes out of parents who are involved and invested in their child’s education. No way.
Put another way, you cannot convince me that the parents are not at fault when a kid can’t read or write after 15 years of education (after having sampled many many teachers and teaching styles, including summer school and remedial education, during that time).
It’s not the system. If you have parents that are involved in education and engage in the kind of child-rearing, that is conducive to education - it doesn’t matter the kind of system you’re operating under (keep in mind, our bar for success is very low - we want kids that graduate high-school to be functionally literate with basic arithmetic and basic general knowledge).
Put another way, if a child graduates and they are functionally illiterate, I don’t care how bad California’s public education is, the fault lays with the parents for letting their child be illiterate.
Hundreds, if not thousands.
The 'Madoff Recovery Initiative' was incredibly aggressive in going after all of Madoff's family and acquaintances. SBF's family and acquaintances are going to be fighting lawsuits for years and will lose millions.
Let's set aside AR because I do think there is potential there, and I never had any issues with AR. Granted, I also haven't experience an actually good AR app.
For VR, I've tried a bunch headsets, from the original Oculus, to the crappy Google Cardboard, to the Meta Quest 1/2 headsets, to the PS VR2, and some newish versions of VIVE. To be fair, none of them extensively, but here's what I found:
- if the VR app has you merely looking around, while standing still or allows 'natural' motion (where I physically walk around a space), after about an hour, I'll get a low-level nausea that will make me feel like crap for the next few hours. It's manageable to use VR every once in a while in this way, but I would never use it day-in/day-out because of the after-effects.
- if the VR app requires motion via controller - forget it. I'm not even sure I could handle 10 minutes of that.
Obviously, I haven't done an exhaustive study on headsets and VR apps, so maybe the new high-end headsets mitigate the nausea issues to the extent where it would not cause discomfort - but so far that hasn't been the case.
For gaming or non-gaming? In my circle of acquaintances, I know 2 people that use it for gaming (one individual is an avid "Elite Dangerous" player, which is a perfect use-case for VR). I don't know anyone that uses VR (or AR) for non-gaming purposes. This is what Facebook/Meta was trying to push.
The problem for VR, specifically, is that there is a sizable population of people that experience nausea when using it even for relatively small amounts of time. I'm in that group. So even if I wanted to, I would not use, for example, VR productivity or collaboration apps, just because of the nausea.
It's still a niche product.
It doesn't necessarily imply that that IFF Starlink is providing military assistance to Ukraine, and in response, Russia targets Starlink satellites, that the US will necessarily escalate further. And there is precedence for this. For example, after US assassinated Soleimani, and Iran retaliated against an American base in Iraq (leading to ~100 casualties - though no deaths, and extensive damage to the base itself), America did not further escalate.
>3. Lots of 'magic' Python variables and methods that you have to override, but there's no way to discover them in your IDE through the type system ...
>5. It seems to leak memory.
Welcome to Python. =)
Of course it does. In the real world, anything is permissible (because there is no global federal government) as long as you have the strength to enforce. In the case of GPS satellite constellation, that is infrastructure owned and operated by the US military, so any attack on those satellites brings Russia directly into conflict with the US and NATO.
>My point being, I don't think "legitimate target" has to do anything with Russia not shooting down any satellites; it's either lack of capacity or fear of retaliation, both being rather orthogonal to the targets "legitimacy" IMO.
Right ... To rephrase: is the US military willing to indemnify Starlink to the same degree if it is attacked by Russia?