pull down to refresh
0 sats \ 0 replies \ @elvismercury OP 1m \ parent \ on: An Overwhelmingly Negative And Demoralizing Force BooksAndArticles
Agreed. Future Shock, 2025 edition.
Would be a mixed bag, imo, resembling the emergent culture around wine snobs and literary douchebags. Taste is so overwhelmingly a cultural signaling mechanism. But maybe something awesome would arise, too.
It's interesting to note that "taste" was one of the metrics of advancement in the recent AI 2027 writeup. I think it's a good idea to develop it, but nothing uniquely human about it. No lasting buffer there.
37 sats \ 0 replies \ @elvismercury 40m \ parent \ on: What job/s would you happily see gone to AI? AskSN
Same. It has already worked out quite well for me on exactly these axes. My day to day has got generally more fun, but it hasn't been obvious how to make that happen.
Longer term, my expectation is that AI will be so productivity enhancing that people will be able to focus on the parts of their work that they enjoy and find meaning in, even if AI can do it "better" (on some dimension).
I consider myself on the right tail of AI users and what you said here reflects my experience; but I note that it's not the default. Like everything, you can apply intelligence and artfulness to something and become more "human" in the process; or you can do stupid shit and get other results. Given the effort required by the former, I shouldn't be surprised that most perceptions are fixated on the latter.
That said, I'm super sympathetic to the critiques. In addition to becoming an AI cyborg being beyond the intellectual capacity of an increasing number of people (which continues an established trajectory with unsettling moral implications) we're given some complicated philosophical knots to untangle.
For instance, there's a recent burst of people writing blogs / newsletters / comments with AI. Most of these are shitty (see above) but a few aren't, which forces you to get really concrete about what the point is. What does it mean for me to write this, now, wholly out of my own brain, vs as a collaborator with a commodified artificial mind? What do we implicitly expect of each other, when we read thoughts that are ostensibly from another person? What does it mean to have a significant chunk of discussion largely constructed by the same entity?
From the POV of things that I think are interesting having become world-shakingly relevant, it's a great time to be alive. From the POV of wondering whether civilization will survive it, it's less great.
These dogs make me sad -- literally designed by humans to be physiological trainwrecks, limited ability to breathe, etc. Everything shitty about modern civilization instantiated into an animal.
Validation is at the heart of Bitcoin. Any improvements in validation speed will have a direct impact on the scalability of the system, including everything that is built on top of it. For this reason improving validation performance may be one of the most important things we can do.
The generalized observation of SwiftSync is that if we have two sets (in this case the inputs and outputs), and we want to know the difference between these sets (the UTXO set), validating a given answer to this question (the hints) is significantly easier than calculating the answer itself. The novel part here (to my knowledge) is specifically how this can be applied to sets. It seems likely that this observation can also be useful in other blockchain-related contexts, and perhaps beyond that as well.
Regular validation as it occurs today in Bitcoin Core involves traversing the blockchain sequentially (with some more minor context independent checks being done in parallel), adding outputs to the UTXO set, and subsequently looking them up and removing them. Often the UTXO set doesn't fit in memory, slowing things down further with disk access.
SwiftSync changes this completely, without reducing the security assumptions. There are no database lookups, memory usage drops to near-zero, and everything can be done in parallel. This would essentially remove memory, disk IO, and single threaded performance from the list of potential bottlenecks. The remaining bottlenecks are CPU and bandwidth, and the more of it you have, the faster you'll be able to validate.
The statelessness of the protocol may also make it well-suited for memory-constrained scenarios such as validating the blockchain with specialized hardware (FPGAs, ASICs), building zero-knowledge proofs, or simply freeing up more memory for other operations such as batch validation of Schnorr signatures.
It's true. I overheard someone say "... I mean, he's not a millionaire ..." and it took a beat before I realized this was meant as an aspirational statement.
Inflation has really had its way with us.
This could be very good or very bad. You wonder if it's a Chinese Democracy kind of thing, after so many years. Although I guess the post-GR gap was pretty big too, if I remember right.
There's a shitty version of this already, which is welfare paid out per child. This has obvious virtues, but my friend is a section 8 landlord and has many stories over the years of this benefit being exploited as a kind of "career" which I think is not holding with the spirit of the thing.
Not a thing you can say in polite company. The prevalence of it I don't know, though, so maybe it's rare in absolute terms.
Anyway, interesting comparison to Hanson's scheme.
Most interesting data set in the world, I think. Will be v interesting to see how he exploits it going forward. Grok 3 is already the best model to talk to in a bunch of ways.
Each time you attack a bad idea, you are feeding the very monster you are trying to destroy.
This is sort of true, but it's missing such an important thing that it might as well be false.
The missed thing is that most people don't really give a shit about destroying monsters, whatever they may say about the matter. When people dunk on the cowardly and hated enemy, they're not striking a blow for justice, they're getting a crack rush off the moral superiority that envelopes them like god's love.
That joy, combined with the points they earn from their own tribe, overwhelms by two orders of magnitude any instinct toward truth-seeking.
"Hell for the ideas you deplore is silence. Have the discipline to give it to them.”
They will never find themselves in that hell because the ones who could send them there can't bear the separate hell of doing without them.
100 sats \ 1 reply \ @elvismercury 26 Mar \ parent \ on: What's one of your more controversial opinions? AskSN
No, you’re mostly just as retarded as everyone else. Maybe even more because now you think you’re right about way more things where normies would actually be humble enough to admit they don’t know and don‘t care.
This is the best and most succinct expression of what I believe that I've ever encountered.
It's a good point. I solved wallet issues and don't think about it anymore. My pain points are around long-term use and value, which are probably not representative.