Transcript: Thinking about excess energy is fascinating because basically the idea is you want to turn something that would have gone into effectively nothing and turn it into useful work. And in the case of this article, the useful work is desalinization. That's great. But I suspect that there's many other useful kinds of work and... Yeah, I guess like which of those is economically viable is the question. I'm also curious how it basically relates to industry and stuff like that. Like, you would think that industry would want to run during those times because it's cheap, but industry already probably does run during those times. Just like a desalinization plant probably also may run during those times. And how do you ramp it up and down unless you are just keeping that capacity on the side? Of course, the Bitcoin miners is another. But you could argue that that's not useful work. It is a kind of work. It's probably better than nothing in some sense. Yeah, basically, what is the useful kind of work that you don't need to run 24-7? And that can be ramped up and down easily. I mean, maybe desal can be ramped up and down easily enough. Of course, we also pump water up hills and stuff like that and basically give it potential energy, which is interesting. But all of this also leads me to the question more fundamentally. I'm looking at energy prices in basically anywhere that I'm looking to live, and I see them going up. And I'm questioning that. Some of it I understand is long-term infrastructure costs and things like that. I'm questioning that. Some of it I understand is long-term infrastructure costs and maintenance, and that is not to be discounted necessarily. I think that's very important. And the cost of energy could be rising partially due to just people needing to be paid and that we have been undergoing inflation. So that is certainly part of the cost. And by energy prices inflating also inflates everything else, because fundamentally it is one of the base units of work that we use to do anything in the world. Which is why Corey's thoughts around energy are quite interesting, because he's just like, if you could measure things down to the joule, you would have in some way some accurate representation of what a thing actually costs. And that is also true, though there's a bunch of other external factors, as we've learned, that may not be encapsulated purely by the joule. With that said, one of the fundamental questions I have is, shouldn't the price of energy be going down? And I suspect that the answer is, that would be nice, but the economics of it maybe don't work out to be that way. That's possible, in the sense that our energy supply keeps growing, but as a result, demand just keeps matching it. We basically have infinite demand for energy. That's what I suspect. I was thinking about this a couple days ago, and I was like, sure. Surely the price of energy must go down. But I guess, like, just assume that we had, like, an extra, like, I don't know, 100 terawatts of power. Like, what would that be funneled into? What would we do with that 100 terawatts of power? And it's very hard to know, taking, like, this abstract notion of power and turning it into, like, literally any form of work that humans could do with power. I mean, certainly in terms of semiconductors, maybe it alleviates the need for certain kinds of efficiencies. Like, we wouldn't need the semiconductor manufacturing process to be more efficient because, like, we could literally just put more power into the system so that the unit economics per chip work out. Because effectively, I would assume, if we just dumped 100 terawatts of power on the market, like, it would cause prices to plummet in the immediate term. But in the long term, I would suspect that the price would kind of go back to where it was. But, yeah, it just, it does make me think, especially when thinking about, you know, effectively power per watt and that whole exploration. And I haven't finished that, partially because I'm just, like, I have so many things going on. And I got my first answers and it seemed to be replicated by someone else. So I kind of, like, okay, like, we know the answer. Though I might want to still do a bigger write-up on that. But maybe also my thinking isn't so clear. Is, yeah, again, this, like, efficiency thing that we're trying to get in compute, I guess, like, basically what happens if the cost of energy was a tenth or a hundredth of what it was? I mean, if the cost of energy was a tenth or a hundredth of what it was, certainly running things directly on your phone would not be what you would do. It just doesn't make sense. Like, anything with a battery, until we have a battery breakthrough, where we 100x the capacity of a battery, it will not be able to match, like, effectively what we can put that power into. And so, yeah, I don't know. I don't know. I don't know. I don't know. It will not be able to match, like, effectively what we can put that power into. So through this lens, it actually very much points to economies of scale. And that, to some degree, efficiency doesn't matter. So yeah, it's strange for me to, like, think about this, because it changes my thinking on, like, if I look through this lens, it changes my thinking on local large-language models. I think it is extremely fascinating that we can have local large-language models. But if this is the case, I mean, what is more likely, the world where we see 100x capacity in batteries or 100x capacity on the power grid? And maybe neither of those is likely, right? But it certainly changes the futures. And it makes me think that again, a lot of the world is effectively up to luck, being in the right place at the right time. Because if you're in a place where there's a 100x jump in grid capacity and you're set up to take advantage of it already, you win in that sense. Similarly, if that's not the case, if you're set up to be in the world where that is not the case, you also win. So maybe the question becomes, what do these two scenarios share? Or how many of these scenarios can we make and what do they all share? One of the questions is, what are you going to do with the compute? Well, what industries scale with power is a different question, but a very interesting question. What are you going to do with that compute, though? And that is kind of the question being answered, that I wish to answer. And a lot of my answer, I think, comes down to building networked brains. And it's hard because I don't even know if the first step is possible. Is it even possible to make a second brain? Like a real second brain? And then is it possible to network that with other people? Or even have multiple people share the same brain? I think that's definitely possible. But then to have them networked is another question. But fundamentally, I think that is a bit of what I'm interested in solving. And I'm curious, your thoughts, John, as well. Just to bring you kind of up to speed of some of my thinking is, I am half debating. I have not, for sure, gone down this road yet. But after Christmas, I'm debating just like driving up or driving down to Silicon Valley and Sand Hill Road and start knocking on VC doors, effectively. I think that I probably am picking the wrong time to do that. Like, during the holiday season seems like a terrible move. But I do partially wonder whether this is worthwhile doing. And if there is enough clear thinking to be able to kind of pull any of this out. Yeah, and part of it is just, I guess, securing, trying to secure funding for, at least, myself. And, I guess, maybe one of the big open questions is, effectively, like, what is the question that we're trying to answer and solve? And, given the futures, which futures can be kind of ignored in some way, because, like, the problem exists in all of them. And, I guess, another framing of the question of, like, what is being solved or what is common across these futures is, like, I wish a computer to understand my context. And, it's actually not necessarily a product in and of itself. And, I think that's something that I've been struggling with and trying to understand a bit more. But, fundamentally, that is the question. And, it's something I've been wrestling with for years. And, I think we see Rewind capturing a bit of that. There's Tab, which is also capturing a bit of that. But, my suspicion is, like, this problem is actually not very easy. Like, it's not as easy as just, like, shoving a bunch of information into a vector database and calling it a day. Or, even one which decays through time. I suspect that's not enough. Because, when I talk about, say, Dave, maybe I know three Daves. How does it know what Dave I'm talking about? I'm not going to use last names all the time. That's not how I speak. Am I going to change my behavior for this thing? I mean, maybe. If it is so good, I guess I will. But... I don't like speaking so precisely all of the time, which is, like, one of the things about being human. I'm not trying to translate myself into a machine. I still want to be human and have all the complexities and emotions that come along with being human. And, like, as I'm thinking about this problem... I just... I am a bit overwhelmed, because I do not think it's very easy. And, when I see demos of things like Tab... Maybe I'm stupid. That's also very possible. Given the knowledge I have. And probably how long that they've been playing with these large English models. Maybe a bit longer than I have. But, when I see Tab, and I see the demos, I think... You know, I think I could get there. But there's still something missing. Or maybe it's because I can't introspect into the thing. That I don't know that it actually works. And maybe that's my own problem. It's like, I want to be able to have a website that is fully... I'm going to be talking about a new app. It's going to be formed by me asking questions. All I have to do is ask questions and give a JSON output, and I can build the website from there. That's all I really want. Because if that can be solved, then you can just build whatever app you want. ... ... Anyhow, John, I'm going to end this here. I recorded this as a voice memo because this is going to go into whatever brain thing I'm building. I'm going to poke at it with questions and stuff. ... ... ... ... ...
90.57% similar
The speaker emphasizes their unique strategy regarding AI, recognizing the existing interest in such pervasive technology and its demonstrated potential. They argue that success in this field isn't solely about attractive designs but also about hiring the best engineering talent to make technological advancements possible. Acknowledging their own limitations, the speaker notes the importance of fundamental technology developed by their friends and the need for substantial technological work, implying that simple technology orchestration is not enough for sustained success. Despite the rambling nature of their thoughts, the speaker seems to aim for a blend of business and consumer offerings, driven by core technological innovation and top engineering expertise.
The writer expresses enthusiasm for the potential of recent technological advancements, specifically with regard to enhancing individual engagement and benefit rather than corporate application. They believe in the potential of mobile devices to run large language models, ultimately changing how individuals interact with computers and information. They draw parallels between early computing and the current focus on corporate-oriented technology, expressing a preference for the democratization of such capabilities. The writer feels optimistic about the direction of technology and its potential for widespread value, despite current perceptions.
89.98% similar
Today has been a challenging day, with the speaker feeling overwhelmed by logistical tasks like taxes and job inquiries. They had a productive conversation with Danny but were left feeling aware of the amount of work ahead. They are also pondering ways to make money and considering the potential of experimenting with data and language models. The speaker is interested in the concept of "brain twin" and is curious about using it in a group setting with others, possibly collaborating with someone named John.
89.70% similar
The speaker is excited about tomorrow but acknowledges that as a developer facing new challenges, the work is not trivial, especially given the lack of extensive documentation and the solitary nature of their current work process. They express a desire to share their learnings, possibly by writing them down, and emphasize the importance of collaboration, suggesting that "if we do this together, it will be a better world." The speaker is tired of creating misleadingly impressive demos and aims to write code and interact with large language models in a more genuine and transparent way. Lastly, they recognize the complexity of building an effective agential system, admitting their current limitations while believing in its importance, and they present open questions about processing and connecting large amounts of data to better understand who we are.
The speaker is pondering whether market efficiency could benefit from restructuring, speculating about specialization and the formation of numerous smaller, highly focused companies. They note an anecdote from Chandler about underutilization at work, suggesting that technological advances have outpaced job functions, leading to wasted potential and the possibility that better time utilization could lead to happier employees and less intense work environments. The speaker questions the need for large companies to conduct all operations internally, proposing that concentrated capital expenditures might be more effective, and uses TSMC as an example while also acknowledging the possibility that their ideas might not be valid. Overall, the speaker is curious about the potential for market disruption through a reimagined corporate structure and is seeking to view the market through this new perspective.