Am I the only one that thinks a test in a technical discipline is poorly designed if more than one person gets a 100%? Exams are supposed to be hard. School is supposed to push you to learn the subject as best as possible.
I just don't think it's that great? It's cool, I just don't think we should all accept it's the greatest. I mean I feel like some spacex stream segments are more awe-inspiring than this. Or some planet earth shots.
For one thing, Apple has tended to focus on privacy at the expense of profit. Apple could certainly be monetizing all of their user data. Now more than ever. It's not just businesses that want your data to sell you stuff, it's the hyperscalers wanting to funnel it into AI training.
Apple is not perfect, by any means. I recently had a conversation with a former Apple employee about how they employ differential privacy internally. This former employee was upset about Apple's interpretation of one parameter ("privacy budget"), but the fact that we're having this conversation at all is a positive. Google, despite being an early adopter of differential privacy, is on the other side of the privacy spectrum: virtually everything they provide is intended to capture what you do on- or off-line.
I will pay a premium for Apple stuff for this, and other reasons. I do wish they were more developer-friendly, however. Enough so that every time I buy a new computer I have to run through the mental calculus of whether I'd rather fight with the cathedral or the bazaar. I recently bought a new computer and the cathedral won the last round.
In this case i am using “good” to mean “not actively hostile towards users”. Yes they are more expensive, but many people are happy to pay a premium to get a premium product. Like going to a fancy restaurant and getting good food. Google’s version is like going to a less-fancy restaurant and getting less-good food but also they sell photos of you eating to TMZ.
I mean I agree with you. But also, it's not that unreasonable of an opinion. As long as it's coupled with optionality, which I think is the actual issue. Well the actual "issue" is that most people don't care or think that much at all about it. HN is a very special crowd.
For people who are just technology consumers they don’t see what could be offered, only what is. This is so frustrating when one understands how railroaded everyone is into maximizing platform ad revenue while holding the reasons people go on the platforms out as a carrot on a stick that gets further and further away. It’s 300 PHD psychologists vs someone just trying to keep up with their family.
Look at the title. Imagine what you'd expect an article whose title is "What causes lightning?" to say.
Now, here's how the article closes:
"These features suggest that even as explanations get more comprehensive, the case of how lightning really works will keep getting reopened. “It just gets more and more bizarre the more we look,” Dwyer said. “Clearly our very simple pictures here are really incomplete.”
So TLDR, we don't know, we know we don't know, and in fact we anticipate not knowing for quite some time. The article explicitly admits it doesn't know the answer to the question it posed in the title - no, the answer doesn't keep getting more interesting, because we don't have the answer yet.
No. The title has a question mark. And the premise that many people think lightning is a dielectric breakdown from high voltage is accurate (if not those words). Posing it as a question is completely valid because it still is an open question.
My half-baked solution is requiring colocation of the "why" for every decision and doc the llm writes, ideally my exact words. And similarly, every so often the llm why it's doing something reveals a mismatch between your intent and its PoV.
I find the intent/why very important. I have a rule in vault related that basically anytime CC writes a WHAT (requirement, concept, RFC) it must colocated a WHY and the WHY should be MY words. It helps resolve a lot of confusion.
reply