The problem with this hype cycle has always been that the hyperscalers are pouring unbelievable amounts of capital into a technology that hasn't proven it can generate the revenues needed to justify that.
Nvidia might have an ok P/E right now, but the question is if the industry can sustain buying over $50B of GPUs every quarter(or that it even needs to).
This exactly. How sustainable are the current spends in the wake of needing ROI against these spends in the not too distant future? And who will be able to afford an upgrade cycle only 2-3 years from now given none of the capex spent will have hit positive ROI 2-3 years out.
Will everyone just accept negative ROI in the name of hype? Will scalers be able to meaningfully increase service prices without eroding customer interest?
These are all unanswered questions that a simple PE statement can't support.
That's the critical difference. You could always find some person who understood a particular piece of a complex puzzle. It's a very new, worrying thing to have pieces that no one understands.
Social media being bad for mental health in childhood is one of the most robust theories I've ever seen for these kind of society-wide problems. You can peruse the After Babel Substack for the evidence if you're not convinced, but Jonathan Haidt has consistently done incredible work here.
All due respect, I do not think the substack of one of the world's leading proponents of the theory that screen time is harmful is a good source for evidence that runs contrary to that narrative.
Here's Nature reviewing his book:
> Hundreds of researchers, myself included, have searched for the kind of large effects suggested by Haidt. Our efforts have produced a mix of no, small and mixed associations. Most data are correlative. When associations over time are found, they suggest not that social-media use predicts or causes depression, but that young people who already have mental-health problems use such platforms more often or in different ways from their healthy peers
> These are not just our data or my opinion. Several meta-analyses and systematic reviews converge on the same message. An analysis done in 72 countries shows no consistent or measurable associations between well-being and the roll-out of social media globally. Moreover, findings from the Adolescent Brain Cognitive Development study, the largest long-term study of adolescent brain development in the United States, has found no evidence of drastic changes associated with digital-technology use. Haidt, a social psychologist at New York University, is a gifted storyteller, but his tale is currently one searching for evidence.
I actually do think that Dr. Haidt is a good source for getting a fair understanding of both sides of the issue. If you've read or listened to him you'll know that it's a huge part of his ethos.
Even the author of your link says "considerable reforms to these platforms are required, given how much time young people spend on them" whilst stopping short of a ban. The problem is these "considerable reforms" will always be half arsed.
There are a lot of problems with the way these platforms treat adults too. I think an age gate is the wrong solution and in many ways it doesn't go far enough.
I’m not sure highlighting studies that seem to agree with his thesis is a particularly strong defense against the charge that the totality of the evidence is mixed and inconclusive. He’s a good writer though.
Why did one study in Spain find an association with the rollout of high speed internet, but a much larger international study specifically looking at Facebook usage did not? Seems like that one should even more directly measure what’s alleged to be occurring.
This hits the nail on the head. There's a marked difference between a JSON parser and a real world feature in a product. Real world features are complex because they have opaque dependencies, or ones that are unknown altogether. Creating a good solution requires building a mental model of the actual complex system you're working with, which an LLM can't do. A JSON parser is effectively a book problem with no dependencies.
You are looking at this wrong. Creating a json parser is trivial. The thing is that my one-shot attempt was 10x slower than my final solution.
Creating a parser for this challenge that is 10x more efficient than a simple approach does require deep understanding of what you are doing. It requires optimizing the hot loop (among other things) that 90-95% of software developers wouldn't know how to do. It requires deep understanding of the AVX2 architecture.
Nvidia might have an ok P/E right now, but the question is if the industry can sustain buying over $50B of GPUs every quarter(or that it even needs to).
reply