Episode 270 - Exciting and Terrifying - NYT Podcast Trashing, GPT4, and Bank Runs
Todays news update starts on the lighter side with an article from the New York Times that seeks to denigrate podcasters by branding them as the "podcast bro". Then we get into the release of GPT4 just months after chatGPT shook the world of AI and tech, and the bank runs at Silicon Valley Bank and elsewhere.
Probability distribution of the week: Exponential Distribution
Links
The New York Times: Would You Date a Podcast Bro?
TechCrunch: OpenAI is testing a version of GPT-4 that can ‘remember’ long conversations
YouTube: OpenAI Reveals GPT-4 Demo
YouTube: GPT-4 Developer Livestream
YouTube: Gamer Presidents Debate Halo
YouTube: Trump, Biden, and Obama Play Dungeons and Dragons ft Ben Shapiro
YouTube: US Presidents Make A Browser Tier List (AI)
YouTube: Biden, Trump, and Obama make a Mario tier list
YouTube: Biden, Trump, and Obama finish their Mario tier list
Wikipedia: Exponential distribution
Related Episodes
Episode 157 - Financial Tsunami on the Horizon
Episode 167 - Blockchain Rising: Wordproof and DeFi
Episode 242 - Artificial Art, Reverse Proxies, and Extraterrestrial Dice
Episode 255 - AI's Got Chat: The Rise of chatGPT
Episode 268 - Pascal's Mugging, Doomsday Clocks, and the AGI Debate
Transcript
Max: You're listening to the Local Maximum, Episode 270.
Narration: It's time to expand your perspective. Welcome to the Local Maximum. Now here's your host, Max Sklar.
Max: Welcome everyone, welcome. You have reached another Local Maximum. Oh my God, I am here on the soundboard today. So I am a little bit. I'm a little bit shaky, and we're gonna have a really good time this episode. First of all, some news might not be such great news. But today we do have an announcement.
Aaron: Is it the Ides of March?
Max: It actually is, I guess. But that's not the announcement today. So it's, look.
Aaron: What do we have to be aware of? If it's not the Ides of March?
Max: Well, this is sadly the last episode, not the last episode of the Local Maximum. Don't worry. It's the last episode in this podcast studio.
Aaron: Local Maximum Labs North?
Max: Yeah, no, I have to move for various personal reasons back down to Connecticut. And so that will be happening very shortly.
Aaron: What did you do to get exiled from New Hampshire? Or is there a gag order on that? We can't talk about it publicly?
Max: No. I'll be back. I'll be back in New Hampshire. Frequently.
Aaron: I know you've already got tickets for PorcFest.
Max: Yep. And I started a new job. So that's good. So we'll talk about all of that pretty soon. Yeah, I got tickets for PorcFest. I also, I plan on staying up here for two weeks, actually in June. But I'll also be here before June as well.
So yeah, some people listening who live in New Hampshire who know me in New Hampshire are going to be very disappointed about that. I know. And I guess when I moved up here, it was during COVID. And I really just had to get the heck out of where I was. And I still like being up here.
But for various reasons. It's time to move somewhere else. And I really, I don't know. It's not like what I was feeling when I left New York where it's like gotta get the heck out of here. Now it's more like I'm getting dragged out of this place.
Aaron: You’re chasing opportunity, not running away from madness.
Max: Yeah, yeah, exactly. Exactly. But the next phase of life is going to be very exciting, actually. And I have good reasons to do what I'm doing, which are also very exciting. And so there is a lot of good that comes with the bad but yes. I'm a little bit sad to be leaving this apartment and this state of New Hampshire, this great state, the Free State. And so I'm hopefully going to make a lifelong goal to come back whether it's having a vacation home here, or just moving here outright, which is, which is what I hope to do, at some point in the future.
All right, well, you know what, since it's our last day in the studio, we're gonna have a little bit of fun today. I don't think we really did a lot of in-depth... We, who are we kidding, we can't do a bad job if we tried. We did some research, didn't we? But like, we could just have a little fun today. We've both got a little bit of rum in our coffee over here. And look, we've got matching mugs.
Aaron: I didn't think we were gonna reveal that we were enhancing our performance with illicit substances.
Max: I think that's dragging down our performance. I think that's actually good. But you know what? I was thinking I could just get up and move around the — by the way, maybe the editor can kind of edit out my voice and use the one from the video — but let's look at this studio for a second. I don't know.
Okay, so it's yeah, we've got here is the, here's the full wall that this is actually it's actually lots of wallpaper up top we didn't use this is. This is wallpaper; these are not real bricks. We're giving away all the secrets because here's my nice couch over here.
Aaron: Those of you who are longtime listeners to the Local Maximum will know that in the early days, there was footage of the pre brick wall studio.
Max: Oh yeah, for sure. This is my whiteboard. I'm doing some work on newmap.ai over here.
Aaron: Super secret, top secret data.
Max: Here is- oh, camera is panning too fast. I'm not a cameraman, folks. I don't know what I'm doing. This is my lighting. This is the camera that points to Aaron.
Aaron: This is why we need a producer to prevent you from going off script.
Max: Oh what is this? I am totally off script here. Let's see if we could open the door. Oh, I love this painting. This has actually become one of my favorite paintings. I don't really — what is this, rum? I don't really drink rum. Except today we're drinking rum, aren't we? But I don't smoke cigars. But I got this in Cuba. I really liked this painting. It's just…
Aaron: Don't just don't go much further. You're gonna unplug yourself there.
Max: Oh. Okay, okay. Well, let's see if we can just peek into my kitchen here. All right, here we go. Are you trying to peek into the kitchen? I don't think I'll unplug.
Aaron: Okay, famous last words.
Max: Also this does have battery. All right, let me put this down. And then we'll get back to what we're doing. Oh, I also, since I'm moving I put a bunch of crap on the, on the on the. Look at that instead of Vcr. Yeah, that's a VCR.
Aaron: Been a while since I've broken one of those out.
Max: Yeah. Well, I happen to have- let me get down there. I happen to have a VCR right here. You know what, I don't like the way that I set up the angle. We're gonna try this again. All right.
Aaron: This is thrilling audio folks.
Max: Thrilling audio! All right. We're gonna keep it in there. But you got to keep listening, because this is gonna be a fun episode. You could already tell. Come on. This is really different from the previous episodes we've done.
Aaron: Well, I think this is going to lead into- I may have changed my opinion on our first topic here after that interlude.
Max: Yeah, okay, because the first topic is an article from the New York Times, and we're getting increasingly serious in our topics today. We have three topics.
So first, we're going to have that one, then of course, we're going to talk about GPT-4, and then we're going to talk about Silicon Valley Bank, which everyone's talking about. Okay. This article from the New York Times over the past week, Would You Date a Podcast Bro? And apparently the New York Times is saying you shouldn't date people who have podcasts. Bad, bad idea.
Aaron: Well, maybe the most generous read of that article is rather to say that there are many people who would not. We are not necessarily saying you shouldn't, but you would not be alone if you have chosen to make this an exclusion, a red flag in your selection criteria.
Max: All right. Well, neither of us have had problems dating because of the podcast.
Aaron: I wouldn't use me as a useful data point, because-
Max: You were already married when we started.
Aaron: Yeah. I have not dated since being on the podcast. And I was married before then, so.
Max: Yes, but you're still married. And you have two kids. And so it's worked out rather well.
Aaron: I tried to get my wife to start her own podcast, in pursuit of equality.
Max: Well, she should and you know, if she should come on the Local Maximum to drum up support, we can get started there. If it's like, okay, just try to get as many converts as possible to our podcast religion.
Aaron: There was a brief period where she was workshopping the idea of something I think she was calling it. It was in the early days of COVID. And I think she was calling it Dispatches From the Panic Room. But it didn't, didn't really get off the ground.
Max: All right. Dispatches From the Panic Room. That actually sounds pretty good. I would listen to that. All right. So here's a quote from The New York Times, we can talk about what the New York Times is trying to do for Ms. Robinson or Robertson? Robertson. Okay. And these are college students that they're interviewing, I believe,
Aaron: Mrs. Robinson. Are you trying to get-
Max: No. Ms. Robertson?. See what happens when I give you rum? “It wasn't just the content of the man's podcast, but that he had one at all. That was the problem. It wasn't that his podcast was bad, but that he had one. Like many other women, she associates the form with a certain kind of man, one who is endlessly fascinated by his own opinions, loves the sound of his own voice, and isn't the least bit shy about offering unsolicited opinions on masculinity, sexuality, and women. Many women have taken social media to mock just that kind of programming, and the men who make it.”
Well, that is a pretty big statement. And one that also assumes, I mean there are a lot of different kinds of podcasts. I feel like they're targeting maybe one type of podcast, one type of shitty podcast that kind of exists. But that is not actually that common, or maybe it's more common than I think. I just don't know.
Aaron: Yeah, it's a depressingly broad brush. I mean, if we were to flip this around a little bit and come up with some alternative versions that have the headline, which probably wouldn't be published in the New York Times, but — I don't know, is Maxim even still a thing?
But if they ran with a headline, like, “Would You Date a Woman With Her Own Opinions?” Or if the opening line was, “It wasn't that she had bad opinions — it was that she had her own opinions at all.” You know, or “Is it acceptable for your girlfriend to hang out with other women and talk about things she's interested in when she could instead be making you a sandwich?” Like that is the level.
Max: Yeah, okay, I see it.
Aaron: This is blatant misandry in the way that they're posing and, and it makes me feel dirty to have to say that because that sounds like a line out of some men's rights activist. Which is, which is not my stance.
But this article is so shallow and vapid. And like I said, misandrist. And should I be surprised that the New York Times Style section can be described as vapid and shallow and misandrist? Or are they simply like I texted you earlier today? Are they just sinking to the level of my expectations continually?
Max: Yeah, I don't know. I mean, I do know. Yes. So it ends with another quote from this guy, Pang, who does a science podcast, I think a Science Podcast would be closer to what we do here on the Local Maximum. So that wouldn't be the people that we're talking about.
He was quoted as saying, “It could mean that you work at this American life. Or it could mean that you record a podcast with a bunch of your friends, talk about the latest week of football games or something like that, or worse, or worse, like misogynistic stuff,” he said. It's like, why do you have to throw in? Is that something people do? They like to sit around and- I mean, well, first of all, there's nothing wrong with making a podcast. That's what this medium is for. Anyone can have a podcast. Yes.
So if you want to get together with podcasts discuss the latest week of football games. What on earth is wrong with that?
Aaron: You know, what the best part about podcasts? I don't have to listen to them if I don't want to.
Max: Yeah, I mean, well, actually, that's a great argument from 2003, Aaron. People don't abide by that one.
Aaron: And, well the woman who was the kind of opening vignette in this article, it sounded like for several years of her relationship with the aforementioned podcast, bro, she did not listen to the podcast. And it was fine. You know, it's like, I no longer read my wife's Twitter. And I think that makes me a happier person,
Aaron: Really? I read her Twitter, actually, it's okay.
Aaron: I blame that mostly on the fact that Twitter got rid of their capability to follow somebody's Twitter profile with an RSS feed. And so it became, it required me to actually go to the website. So that hurdle made me just give up and throw up my hand.
Max: It gets you on Twitter. Right, right. And then you have to read everyone else on Twitter. Right. Okay. So, right. So here's the problem. I happen to think there's a deeper kind of hidden agenda of the New York Times. They might disagree with me.
By the way, I did an interview with the head of data science of the New York Times, it's going to be coming out in a Local Maximum in a few weeks. And I really hope that he doesn't ask me to rescind the interview, because now we're crapping all over the New York Times.
Aaron: Well, I would be curious to share an adult beverage with him and hear if he has any inside gossip on the New York Times Style contributors, see if there's some internal conflict there.
Max: I'll ask Chris Wiggins next time I see him. Okay. So I think that the hidden agenda is that podcasting is actually the freest medium, podcasting is the only medium out there that has not seen a significant amount of censorship, and deplatforming.
And I believe the reason is — first of all, just the structure of the market, how it's so fragmented — but also look, I mean Apple podcasts, and Spotify and all these big players. They don't host podcasts, but allow you to play podcasts, like they could if they wanted to, like put warning labels on things or take things off.
And they have, in some extreme circumstances. But I also think it's because the information that you're getting from the podcasts is not right in your face. It's not like everyone can be viral.
Aaron: Interestingly enough, you could see a situation where something like like Apple with with, I guess what used to be iTunes, it is now the Apple Podcast store, or Apple podcasts, that if they were the Goliath in the market, which they probably are, the biggest player in distribution still, but but it's not like on Facebook or Twitter, where the platform can decide who to amplify and who to de-boost or suppress or whatever terminology you want to use.
Max: They can a little bit.
Aaron: They can but there are so many other venues for distribution. And the host doesn't really do that. It's about who gets promoted in the app. And so nobody's talking about that in the same way they talk about “my Tweets were shadow banned,” because if you're interested in something, you can subscribe to it and you get it. And they're not choosing not to deliver episodes to your feed because Joe Schmo’s podcast hasn't paid for promotional credits on Meta or whatever.
Even though there are recommendations, there are a small number of big players in the podcast distribution arena, they haven't wielded that power in a way that pretty much every other social media network has. And maybe it's because it's not a social media. It's some other type of media. Yeah, no, even though there's definitely social and parasocial relationships, at the heart of a vast majority of these podcasts.
Max: Yeah, exactly. And I think, yeah, there's a whole lot of other things that can be said about this. Couple more bullet points I have about the New York Times: they actually do have about a podcast section where they highlight different podcasts. So there. They do have someone with over at Styles.
Aaron: Any of them hosted by bros?
Max: No. Imean, I'm sure some of them are hosted by men, but they're not. They're kind of New York Times approved podcasts. But I really think that they're ignoring the incredible diversity and innovation that has taken place in the podcast space. And no, it's not just a bunch of bros. I know what you're looking at here on a Local Maximum right now today, but it's first of all, is not always true on the Local Maximum. And it's not always true-
Aaron: And for what it's worth, I don't identify as a bro.
Max: Yeah, I no longer identify as a- no, I never identified as bro. But we're not the bro personality types, if there is one. Look, can I just finish my point about I believe that like in the future, there's going to be like a museum of communications technologies.
And podcasting will be there, there’ll be exhibits about podcasting. And it'll be like, this was the medium that people flocked to where they wanted to be free to innovate without all the gatekeepers, and they're gonna have tons of examples.
Maybe they'll even have hey, there's a bunch of like, lesser known examples, like these guys and Local Maximum. I don't know if I can make the museum. Maybe that's a little bit too much to think but you know, I really think that's what this podiverse is doing here.
Aaron: Yeah, there's a little bit of schadenfreude involved in looking at them, watching them attempt to gate keep all the things in a world where being a journalist and that allowing you to automatically get a Twitter blue check no longer has currency. Yeah, it's, yeah, that combined with with their kind of weaponizing of the concept of toxic masculinity. Yeah, that makes for a fun time. Good read there. All right.
Max: We're doing some good social commentaries today. Welcome back. So oh, you also have this quote from Mike from Pennsylvania from the comments section?
Aaron: Yeah, it’s one comment I pulled out of the responses there.
Max: By the way, the New York Times comments section is hilarious.
Aaron: It’s like the worst aspects of a cocktail party. Everybody's trying to show off how smart they are. Which granted, any sort of posting on the internet, there's an aspect of that, but this feels like that cranked up to eleven.
Max: It’s very on brand for the New York Times.
Aaron: Anyway. So Mike from Pennsylvania said, “Having a podcast is something someone does for fun or a side hustle, not as a full time gig. Particularly if it’s just a podcast of one or two people riffing about stuff and asking for support on Patreon.” At that point I just stopped it. I'm like, ooh, this is hitting a little close to home here.
Max: We're not asking for donations on Patreon. We're asking for donations on Local — and by the way, do it, it's very important. Okay, go on.
Aaron: Yeah, “and not creating other types of content like a newsletter or interviews with well known people,” like, phew, we slipped by on that one. “At that point the podcast becomes a company which I don't mind supporting.” So yeah, let's draw some arbitrary lines there on declaring your worth.
Max: Oh, so we should go talk to Mike. Maybe he would support us on maximum.locals.com because we do interviews with people and occasionally a well known person. And he wrote here that becomes a company, which he doesn't mind supporting. It sounds like he's one of our core listeners.
Aaron: Trolling aside you know, Mike from Pennsylvania, we can't call you out as a specific individual. But if you are listening and and you would like to come on and discuss further you know, our email inbox is open.
You can always come on The Locals and have a conversation with us there and any of our listeners who think that we've come a little bit out of line here we're being a little too bro-y, that I didn't put enough rum in my drink, can chime in there.
Max: You know, some guests were former listeners. So we do have people or guests who are listeners. We also occasionally we do like a short call-in show where we have guests who like kind of listened-
Aaron: I hope they remained listeners and did not become former listeners. They started off as listeners.
Max: Yeah. All right. So this is a big news week. Aside from that, we were just having a little fun for the first 20 minutes. Sorry, that was too. That wasn't the introduction. That was part of the meat of it, right.
But you know, so big news week in AI: OpenAI launched its long awaited GPT-4 model this week, now available on ChatGPT as well, which I have used because I am a paying subscriber to, to GPT got to do it for work got to do it for the research it uses an order of magnitude more transformers.
What's a transformer? Well, I think the size of a neural network, it just the bigger a neural network is, the more information it can contain the more interesting functions, the more complex functions that it can hold. So in a sense, it holds more information holds more intelligence.
Aaron: You mentioned you're a paying subscriber. Being the cheapskate that I am, I am not. But is there a single like level of subscription? Or can you be at different tiers depending on the kind of quality of?
Max: No.
Aaron: So it's a binary, you're in or you're out?
Max: It's like $20 a month?
Aaron: Okay, so it's not like you can pay more to get you know, speedier access, more powerful?
Max: More transformers? Well, then these are, it's not like these models are all trained individually. So it's like GPT-4 is one model that was trained, it probably took months and months to train lots of money. And then it's just sitting there as a model that you can use. So it's kind of like, it's sort of like, I mean, I guess, I don't know, maybe they can throttle it for like, certain customers or whatever. But yeah, it would be hard to like, try to make it less intelligent, I suppose.
Aaron: And is unpaid use just limits your interactions?
Max: I think it just slows you down a little bit. And I think, yeah, I think when it's popular when the service is popular, they throttle you a little bit. And also GPT-4 you can use ChatGPT-4. It’s multimodal, as opposed to multi-model, which, when I was at Foursquare was building something that I called multi-model approach, they build a lot of models. But management kept calling it multimodal, because I guess they had heard that word before. And so they didn't know to call it multi model.
But anyway, this is multimodal, that means it can not just do text, but images, code, things that are different from texts. And so it has different modes of operation. And so it can only generate text, it can't generate images, like some of these generative AI systems. But I've noticed that because it can generate text, it can also generate code, and it can generate things like tech, and HTML and CSS just fine.
So even if it can't generate images outright, how long before it can at least illustrate stuff very quickly, by like building a webpage or just rendering an equation or rendering a graph really quickly without having to do any of that fancy image stuff. Feels like that's right around the corner
Aaron: One of the examples I've seen circulating is somebody who, who sketched out I think it was literally on a cocktail napkin with a ballpoint pen, a design of a website, then took a photograph of that submitted the photograph to GPT-4, and it created that website, which is pretty slick.
Max: Yeah, yeah, very slick. A few more headlines of TechCrunch, OpenAI is testing a version of GPT-4 that can remember long conversations, the quote is “OpenAI has built a version of GPT for its latest text generating model that can remember roughly 50 pages of content, thanks to greatly expanded context window.”
Aaron: Is that the 25,000 word I saw that I've seen tossed around before?
Max: I suppose so yeah. And the article goes on. Yeah, 25,000, about 50 pages. The article goes on: “That might not sound significant, but it's five times as much information as the vanilla GPT-4 can hold in its memory and eight times as much as GPT-3.”
Aaron: I see. So this is the open AI…Wait, so there's GPT-4 vanilla and then there's the OpenAI version of GPT-4?
Max: Well, well, no, they're all OpenAI. OpenAI is the company. I am not sure if it's like if they actually trained multiple models, which I just said before that they didn't or if it's the same model that they're adding.
Aaron: Just giving more resources.
Max: Yeah, like if it's a recurrent neural net.
Aaron: I know one of the things that I talked about with GPT-3 was that if you interacted with it long enough, eventually you know stuff eventually dropped out the back of its memory.
Max: Yeah. So that's what's happening when you play tic tac toe, I think and so I that's what I tried and if you remember when GPT-3 came out, or ChatGPT came out, we were talking about in the program only a few months ago this has gone fast. By the way, what episode is that we should probably, we should absolutely link to that episode.
That was probably Episode 255, the Rise of ChatGPT. Alright, so basically, it can't play tic tac toe, GPT-3. And it was like adding X's and O's on the board that shouldn't be there.
Now, I played Tic Tac Toe again, and it didn't do that. But it still declared itself the winner when it didn't actually win. So I feel like it has a slight, it felt like very incremental, like it had a slightly longer window of understanding. But not much.
Look, either eventually, this technology just gets so good that it just remembers and you don't have to worry about the fact that it's all you know, that it's limited, or it hooks into some database that has exact a memory of what's going on. I don't know what the answer ultimately is going to be. But so it failed the tic tac toe test.
But I know we just watched the GPT for developer livestream. And so there are a lot of interesting things you could do with it, particularly in terms of having it generate code really quickly. Python code, website, code, JavaScript, HTML, and that I think is gonna be very useful for me on the job, I think it's gonna make me a lot more. It's gonna make me a lot more productive.
The question is, is it going to replace my job? I don't know. I mean, I assume that for now, it's just going to seem like I'm really productive. But then the companies might be like, well, we don't need to hire that many more engineers, I don't know how that's gonna work.
So it's, it's, I assume that there is going to be a role in the future for data scientists, for machine learning engineers, for software engineers, for coders for infrastructure, and for data and for all of that, but it's just what is the nature of your job? It's probably going to be changing quite a bit over the next few years. And I'd like to actually hear about how people think the nature of the job is going to change.
Aaron: I mean, there was a lot of talk with was it StableDiffusion and some of the graphical AI tools that it's going to make the move for kind of mid tier artists, that that may disappear, and be replaced by AI prompt generators, people whose expertise is not necessarily in Photoshop, but in manipulating the criteria correctly for talking to something like Stable Diffusion to get the results out that you want.
Max: Yeah, Stable Diffusion. That’s what I was talking about.
Aaron: That is indeed a skill set, or at least something that you can learn and get better at.
Max: Yeah. And so two points, I mean, the point that I was making before, but now that I realized the word for Stable Diffusion, because I was fetching for the words. Okay, how long before it could just render HTML on the tech, right off the bat? Probably very quickly. And then how long before it just integrates something like Stable Diffusion, and you can just do that. So it's, I mean, maybe a few years.
Aaron: In the the engineering and project world, there's, there's a lot of talk about defining requirements, which probably will take on a slightly different aspect to it in this context, but but the, the heavy lifting the project is no longer necessarily the implementation, as it is learning how to properly specify what it is you want done. And then the AI can turn the crank on that as long as you clear, you've set up the right requirements,
Max: Think about this, this is an incredibly impactful, and incredibly impressive piece of technology that they have here with GPT. Because you can almost write instructions, like you'd write instructions for a human and even though it really doesn't understand it in the same way a human does. It gives you very good results.
And I've sort of thought of it as, kind of like doing research on the internet or doing research on Wikipedia, where it's a great starting place, and sometimes it gives you the right answer right away. And but you can't, you have to double check things that you see on it, but you don't always have to double check. Because like sometimes in Wikipedia, you kind of have a sense of okay, like, I know, this is right, and I don't, this is not life or death. And I can just go with it.
Or if it's, if you're trying to debug something like how do I change my account on this service? If you ask GPT that, it'll give you a list of steps. Try the list of steps. Usually it works. It's probably better than asking Google at this point where you might get into someone's YouTube video, and then you click the video and they're like, welcome to my video, and why don't you smash that like button and I'm going to explain some good stuff for you. And then you're like, where are you going to?
Aaron: Don't be so rough on the YouTuber bros.
Max: Yeah, okay, sorry.
Aaron: They need to get dates too.
Max: I'm sorry, but how many people out there, a lot of people listening who understand that?
Aaron: Ring that bell!
Max: Yeah. But yeah, but also like to take the websites, with out-of-date information like no, you just ask GPT.
Aaron: I think you made an interesting distinction there. Because it's one thing to say “Can you give me a list of steps for how to do task X,” rather than saying, “ChatGPT, go change my password on this service.” Which I would be a little anxious about letting it run off and go do that, even if it can do that right, nine out of ten times, if it does it wrong, that could get messy.
Max: Especially if it’s attached to your bank account.
Aaron: Move all my crypto from this wallet to this wallet!
Max: Yes. Attach me to an IV and start pumping stuff into my brains, into my veins.
Aaron: Yeah, at least if we're looking at the steps, then hopefully we can catch anything that looks wonky. But you know, we do that successfully long enough. And we're gonna start getting comfortable with kind of skipping that intermediate layer.
Max: Yeah. Yeah. All right. So this is very exciting. And because I'm starting a new job, I'm kind of looking forward to trying it, GPT first. When I first got into software engineering, back in the early 2000s, it was always like, people had to explain to me, like, engineers. It's almost funny now to think about because we're coming out of college, you almost think that using Google is cheating.
And they're like, No, Google is your friend. Use it as much as possible to be as productive as possible. I had to be explained this back in the day. Yeah. And now I think we're jumping to GPT. And it's like, no, this isn't cheating. I'm going to use this to my best advantage to just give myself the biggest advantage I can to move as quickly as I can. And I feel like I don't know. I don't know if I'm any good at it. But I feel like I'm gonna. I feel like not using it is worse than using it.
Aaron: So well. There's definitely the derisive term “script kiddies,” which is, which is kind of a modern equivalent, maybe people who don't really know how to code. They just know how to copy things from Stack Overflow.
Yeah, but it's the idea of taking somebody else's code that you don't understand and running it and that, the old school attack on that would be well, that doesn't make you a hacker, you're just a script kiddie. But if a more modern view might be that if you're getting results, then who cares? It doesn't matter until it does.
Max: Yeah. When thinking about copying things from Stack Overflow, the dirty little secret is that we all do it. Now, look, I have kind of, I have both a practical and an educational background in computer science, the Bachelor's degree in Computer Science from Yale, masters, NYU, all that.
Does that help? Like, because I actually understand how algorithms work? I've taken many algorithms classes, I've studied algorithms very carefully. I studied type systems very carefully. But look, there are other aspects of coding that I haven't studied very carefully. You know, for example some of the infrastructure type stuff, I know, a lot less than, than some of the infrastructure engineers. And so I kind of feel like I'm kind of a bootcamp guy when it comes to that stuff.
So it's like, does the in-depth knowledge, does the theoretical knowledge help you? And I'd like to think the answer is yes. I mean, maybe. It's like sometimes you feel like, not so much, although there are people who, who adamantly say yes, and know that it is helpful. But I think they're probably right.
Aaron: I think in a perfect world, you would understand everything at every level all the way down from soup to nuts, top to bottom. And that would make you better at writing code or whatever implementation it is you're doing.
But there's, there's probably a point of diminishing returns there, where in order to accomplish your objectives, for you to write the type of code that you're doing, it is not necessary for you to know assembly, or to be able to write it in binary.
Now, if you did, you could probably optimize some more and do things in a more efficient way. But is that gain worth the additional investment it would take to have that level of knowledge and expertise? Almost certainly not.
Max: Yeah, I feel like some of the in-depth knowledge gives you the intuition, the the ability to solve problems, maybe that's the value of education.
Aaron: And not everybody will have the capacity to be an expert in all things. So there's a specialization aspect of that too. If you only have a certain number of skill points to invest, you’ve got to pick where to put them.
Max: Yeah. All right. So lots that can be said about GPT-4 I've been listening to kind of these fun videos online where they have the presidents who it's almost fun to imagine, it's usually like George W. Bush and Obama and Biden and Trump are like hanging out. And they're all they're all like, riffing on each other, but it's kind of like, they're, they're still friends for some reason. It's very rare.
So in other words, using AI to generate their voices. And, and like, I posted one where their debate Halo, I also saw one before, maybe I'll try to find it, where, you know those videos where people are putting, like categorizing things into tiers, and there's an A tier, a B tier, a C tier, and there’s an S tier on top, they’re like categorizing all the Mario games, and then they were disagreeing. And like Trump would be like, “No, that's a terrible game. I don't know what you're talking about.” Biden was like “No, I think it's an A,” and then Obama was like “We're just going to put this into C and move on.”
I'm sorry, I do very bad voices, but you know, I maybe got some of that the cadence but not the articulation. Not the timbre, but these are really fun to listen to because I don't have to do the voices, GPT can do it and it makes a lot of like really fun, oh, you put the one on Dungeons and Dragons here. Yeah, it makes a lot of like kind of funny meme material.
Aaron: Yeah, I haven't looked into how it's created whether literally all you have to do is feed a script into it and and it creates the audio for you or what level of kind of tweaking and customizing is being done versus comes out full fully formed but it's at the point where it is perhaps not trivial but extremely feasible for people to create audio deep fakes which are very convincing and I think they're in a safe zone here because they're doing this on things that are that are clearly not the actual president's talking to each Yeah, we're no one's worried that we have misrepresented their views on Minecraft.
Max: Sometimes like the, sometimes they hear glitches in the AI and I hear the guy's voice who is doing it. But another funny thing is like, the the voice for Bush is like less-
Aaron: I don't know if I've seen any Bush ones.
Max: Yeah, it's, the audio is not as good because the training data is on lower quality recordings, older recordings. Yeah. So it's kind of funny in that sense.
Aaron: So we're well beyond the point where we should be worried about audio deep fakes of presidents or presidential candidates or any politician saying something that is potentially controversial, or inflammatory or, or whatever. That has arrived. That's, in fact, been here for a while.
The big question is, when is video going to catch up to the point where it's going to be a real problem? And I think, excuse me, I think I've already said that we are going to see that in the 2024 election, that it's just a matter of when the big, big drop comes for that.
Max: I've seen deep fake videos of Obama, while Obama was president being demoed to me. So this has been around for a long time. And video is probably to get past the point where like, it's not-
Aaron: But the ones that were contemporaneous with the Obama presidency, those took a fair amount of work and expertise, and were kind of in a controlled environment. We're at the point where it's entirely feasible for someone to basically flood the zone with garbage that is of high enough quality that it's difficult to distinguish what's the garbage and what isn't.
Max: I feel like news organizations will be able to tell in 2024, but there might be some good social media hoaxes coming well.
Aaron: Yeah, it's it becomes a question of when when our news organizations going to have to restrict the reporting to things where we had an actual reporter on the location who observed this thing and can vouch for the quality of video, and how long will it be before people are willing to pull their, well this just this one time where, where they're willing to take the deep fake and run it and burn their credibility for some reason on it.
Max: I don't think that I mean, I don't think we need deep fakes to show politicians saying dumb things.
Aaron: Well, the big concern is not so much. I mean, one approach is you use the deep fake to get the politician to say something dumb and you know, impact their reputation and their perception. The other thing you do is that when they actually do say something, that they can just write it off, they will say that was a deep fake, it wasn’t real.
And I've heard it speculated, how long will it be before? You know, there's video evidence of something presented in the courtroom. And the defense presents a deep fake of something that appears to have just happened in the courtroom in which they are currently present, that everyone who was there knows didn't actually just happen, but is indistinguishable from the credible video evidence that is being used against the defendant.
Will that mean the end of video evidence in judicial proceedings?
Max: I think, I think we're a ways off before it's like forensics proof, even if, even if we get to the point where like okay, could fool some people. Which we're not even quite there yet.
But it's like, with video, I think, at long as someone who's paying attention, but it's like I think that. Yeah, I mean, we can get there at some point, I just don't think we're, we're right on the doorstep.
Aaron: It’s coming in our lifetimes, for sure. Whether it's in the next year or two is an open question.
Max: What if someone is elected president? And then you have deep fakes, saying, This is what the President says we should do, that's an order. So but the President said no such thing. You could have a deep fake coup? Yeah. In a country or any organization really.
Aaron: Well, there was the classic incident of was it? Was it Reagan? Who was doing like, basically a mic check. Before going live on television for a speech, from I think it was from the Oval Office? And he says “We have just passed legislation outlawing the existence of the Soviet Union, the missiles launch in five minutes.” And yeah, part of that accidentally was broadcast.
Now imagine if if, instead of that being just a gag in the White House, and everyone is immediately cleaning up that mess, if the Chinese put something out and get it to the right people in time, just right, so that, I don't know, maybe North Korea decides it's time to hit the big red button, or Iran decides that, balls to the wall, it's time to throw everything we've got at Pakistan.
It may not take much injected into just the right moment to nudge things to catastrophe, because, because we've come so close, without deep fakes before.
Max: You might have to get like, cryptographic signatures on, on orders all the time, or on I mean, we had an episode, actually, then this reminds me of an episode that I've long forgotten on a Local Maximum, this was I believe, back in 2021.
I'm trying to figure out what this is, it was definitely something along the lines of. Was it June 2021? It was something along the lines of like, how to- oh, here it is.
Aaron: Was it talking about using LBRY?
Max: No, no, it wasn't LBRY. that was the one with Jeremy Kauffman. Episode 167 on wordproof, where it was somebody who wanted to timestamp everything online. So not only timestamp, but maybe you could also cryptographically sign it.
And that would kind of that would kind of say like, hey, if I'm gonna give a presidential order, I better cryptographically sign it with something you know I own and sign the sign the video as well and sign the video of me signing it, just so that it's like it can't be faked. Because in the future maybe all of that can be faked. So that can be interesting.
That could be an interesting idea if it comes to that. We better hope that quantum computing doesn't break encryption, which I have an episode of quantum computing coming up.
Aaron: Makes me think of the nuclear football on the biscuit but yeah, anyway, we've got a little bit of field here.
Max: Yeah. All right, onto the next topic. Because I almost forgot about this next topic. And we've been 45 minutes into the show. And I literally, it just slipped my mind. Is our banking system falling apart right now?
Aaron: This is old news. It's almost four days old now.
Max: Right. Okay, so apparently, this is the third story, Silicon Valley Bank. Yeah, our banking system almost collapsed over the weekend Silicon Valley Bank.
Aaron: Did it start on Friday?
Max: I think Thursday. I think the FDIC, the government came in and shut down the bank on Friday. Thursday that was the start of the bank run. So yeah, this has happened fast.
Aaron: How fast did the savings and loan collapse back in…now I'm mixing up my banking collapse because the savings and loan collapse was what the 80s? But then there was the collapse that happened.
Max: Lehman brothers?
Aaron: I'm thinking back to the Jimmy Stewart era, because that was way before that. But yeah, how quickly did those happen?
Max: Probably very fast because once it starts happening, then there's a panic. And you don't really. I mean, that's the definition of panic, is you don't really have time if people panic slowly, then it's not really a bank run, is it?
Aaron: I'm thinking less in terms of like, well, that bank fell quickly. But did it spread across the nation quickly? Or? Or in a time when communication wasn't as fast, it contained the contagion?
Max: I don't know. I don't know the answer. I don't know the history, unfortunately, that's a good question. But Silicon Valley Bank, that's the second largest bank in history.
Aaron: Was Lehman the biggest in history?
Max: No, Wells-Fargo. When did Wells-Fargo fall apart? I'm trying to remember when that happened.
Aaron: Wells-Fargo is still around, in some incarnation.
Max: Yeah. The collapse; when did that happen? I want to say 2010. But I felt like it wasn't really a big deal when it collapsed in 2010. Like it wasn't. It wasn't, because that was after the financial crisis, right? And so yeah, I don't remember when that was.
Yeah, well, Wells-Fargo is still around. So or could I be getting the- No! It was Washington Mutual. Okay, ah, that's it. Washington Mutual. Oh my god, I'm gonna start a bank run on these poor people at Wells-Fargo.
Aaron: Wells-Fargo has done enough bad things.
Max: Washington Mutual collapsed in 2008, along with those other companies like Bear Stearns and Lehman Brothers, Lehman Brothers collapsed several times.
Aaron: Wasn't one of the Lehman guys on the board at SVB.
Max: I wouldn't be surprised.
Aaron: I mean, it's a small world. So they, I don't think many of the Lehman Brothers folks left the profession. I think they just shuffled around other places.
Max: Right. So the FDIC stepped in and bailed out depositors, the FDIC is its insurance, they're required to bailout depositors up to $250,000. But because a lot of these were companies and startup companies that have to make payroll, $250,000 is not nearly enough.
Aaron: And I think technically, corporate accounts are not covered by FDIC insurance. But I'm not I'm not positive.
Max: So I think they stepped in and covered everybody. And so there's the question of like, is that a bailout? That's kind of a controversial.
Aaron: Or at least they're saying that everybody is covered. I mean, not everybody has gotten their money out. So we don't know.
Max: I thought they got it out this week. But I could be wrong. We, I guess we didn't do enough research on this one. But if the money was locked up on Monday, I think we would have heard about it, because that would have sent shockwaves throughout the system is what I heard, but I could be wrong about that.
And so yeah, they're not bailing out the owners of the bank. They're not even bailing out the bondholders of the bank, just the depositors. And so I want to do a little bit more research on that and what that means.
But yes, there are other banks in trouble. Credit Suisse. Credit Suisse. What am I saying? Like, what am I, French? Credit Suisse signature, which is Barney Frank's bank, Barney Frank wrote the Dodd-Frank bill that protected the banks. And then there's this whole blame game going on, in terms of what caught what's causing it.
The proximate cause seems to be the interest rate hikes that have gone up very considerably. And for the first time in our working lives, we have a significant interest rate right now. And so that means that when these companies bought bonds that payout a certain amount, well, they pay out money in the future, the value of those bonds have gone down, because now people can get a lot more money.
Aaron: Yeah, I'm no expert on the bond market, but generally, it goes in the opposite direction of interest rates.
Max: Yeah, exactly. Other people have said that it's the spillover from the crypto contagion. I believe that Barney Frank said that, which is in response to Joe Biden, who said it was a banking law from 2018 that seemed to relax some of the provisions from Dodd-Frank. Biden used that to blame Trump for it. And so you may know a little bit more about that.
Aaron: I believe there was a change to the ruling that before the relaxing of those regulations, there was a $50 billion threshold where if the bank had had holdings in excess of $50 billion, they had to meet certain additional reporting requirements or that they had to comply with additional regulations. That threshold was bumped up to 250 billion.
And I believe it's been reported that that Silicon Valley Bank was most likely intentionally, but their holdings were just under that threshold so that they were not required to, to comply with those additional requirements, which sounds when presented that way sketchy.
But it is entirely reasonable that if I can keep my holdings or my income or my earnings under X, I don't have to deal with onerous requirements that have additional costs and overhead required. Whereas if I bump over that, I got to do all this stuff, which and I don't necessarily have additional gains on my end for doing it. That's, that's entirely reasonable.
This gets to one of my pet peeves that what people call a loophole, and I haven't heard anyone talking about a banking or regulatory loophole explicitly yet, but but it's very much the same behavior, what what some people call a loophole, others can can accurately describe as complying with the law.
Max: Yeah. All right. So, yeah, so I think that banking law sounds like that's just a red herring. That's just so that Biden can say like, it was Trump's fault. And then, of course there are people who are blaming wokeness in the banks. The social justice warrior, like Tucker Carlson was talking about how one of the risk, I think the head of like risk management at the bank was spending a lot of her time on political causes and social causes.
I don't see any evidence that that's what caused the bank to collapse. Although I do see evidence that it was probably very annoying for other employees who worked in the bank.
Aaron: Yeah, I think that is equally a red herring. But the way it's being dismissed by a lot of people is a little frustrating, because while this may not be a zero sum game, to some extent, every, every ounce of time and effort that is dedicated to pursuing DEI goals and requirements is not being put towards other objectives.
And if that is an important goal and objective for your organization, then that may be a worthwhile trade off. But to not acknowledge that it is a trade off is to willfully blind yourself to reality.
Max: Yeah I really think and if it's too controversial, you don't have to be a part of this one. But I want to talk about, at some point, how much damage some of this some of this ideology that's been injected into companies has caught I mean, I already have done episodes on it, but I kind of, I haven't feel like I haven't done one that like a, tap it dead-on with my thoughts on it, but maybe I will soon.
Anyway, the significance will play out soon enough. I want to also, if I could say I told you so, call back to a Episode 157 which, ironically, was the last episode before I moved up to this apartment.
Aaron: Before you departed the aisle of Manhattan?
Max: No, I was actually in Connecticut. That was the brief moment when I was broadcasting from my parents basement, like a real podcaster should.
Aaron: Do we have footage from that era?
Max: No, no, no, no. I made sure that was absolutely not televised
Aaron: So I only saw that through the Zoom.
Max: Yes. No, no one will ever see that. So yeah, and now it comes full circle for the last week here. That episode was entitled 157 Financial Tsunami on the Horizon. I quote myself, the changes to our financial and economics and this was in February of 2021.
The changes to our financial and economic system are like a tsunami on the horizon. And we see it coming closer and closer to our shore. And we have some, we're experiencing some big waves first, and we're like, was that the tsunami?
And no, it's like the small tsunami, and the bigger one is still out there. So that's gonna crash in a few years. And we will be along for the ride, if I could use that if that metaphor has landed properly. And then you responded, “Well, now I'm truly terrified. So thank you for that.” It was towards the end of that show.
And yeah, since then, we've had kind of runaway bubbles and inflation, followed by high interest rates, followed by all of these like crypto bank runs, not just the price collapsing, which happens all the time, but all of these crypto companies holding people's hard earned money, just kind of collapsing and locking people's funds. And then now to see that moving into the banking sector itself is kind of scary.
I feel like we are experiencing this tsunami and I don't know if, like a true bank run is going to be allowed to happen, but some domino’s gonna fall somewhere. I feel like the other shoe is gonna drop. I am still terrified.
Aaron: I was not worried until on the drive over. I was listening to another podcast on a completely well, not completely unrelated subject but not focused on this explicitly. But it triggered some thoughts on my end. And I was really considering whether should I maybe go to the bank and withdraw just under $10,000 so that I have a significant amount of cash on hand because, right now I've got like, what, maybe 50 bucks in my wallet.
And if for some reason my bank went down, even if the money wasn't gone, but if for some reason I couldn't access it for a couple of days. What am I going to do?
Max: Yeah, move to crypto, that’s what. All right. So I think that's good enough. Now, let me see if I can get the right animation or not animation. If I can get the right…
Aaron: You got a 50-50 chance. And either way, it's going to be entertaining.
Max: All right, let's go to the segments.
Narration: Now, the probability distribution of the week.
Max: All right, the probability distribution of the week, well chosen wisely. We're just having a crazy old time here. We don't have time for a very complicated distribution today, do we?
So and because we're doing continuous distributions, and we've already done some of the simple ones like uniform, whatever. We're going to talk about the exponential distribution. I think that's pretty straightforward, right? It's just like, like e to the minus a-x, or? Well, I mean, it all has to integrate to one.
But it's basically, the support once again, is positive numbers. So you're picking a positive number. And the probability of, the probability density of any given number as you get higher and higher has an exponential decay on it. And that works. Because if you integrate, if you find the area under the curve off to infinity, that's a finite number. So that works.
I should also point out, it's the basis for a lot of stuff. So we talked about the gamma distribution last time,
Aaron: Was that our most recent? It's been a while since we've done a distribution. I guess you did at least one solo episode.
Max: Episode 268. Yeah. So that's when I did the gamma distribution. So this is actually a special case of the gamma distribution. So it's a special case worth mentioning. And also, it's kind of not a special case, but a building block of the Laplace distribution.
So the Laplace distribution is, I would say, okay, you know that the normal distribution is kind of like a bell curve, right, like a bump, the Laplace distribution takes an exponential distribution, and kind of reflects it over the line. So it kind of copies it. So instead of a bump, it kind of crests up to like a point on the wave. And it could be used as another form of like, a normal distribution it's got like, it has very different properties from the normal distribution, but it has actually somewhat similar properties. They're both short tailed, they both have very unimodal, a very well defined mean, standard deviation, all that. And so that's Laplace distribution.
And we can talk about that another time as well, like what's useful, because it actually has a very big use in machine learning. But, yeah, exponential distribution, I think, is pretty straightforward. It's, the continuous version of the geometric distribution, in discrete, where it's like, each step is half as likely. And you could use it in several ways.
You can use it if you make the exponent, like, it could scale up in several different ways, depending on what that rate of decay is. So if that rate of decay is very high, then this distribution has a very small variance and all of your probability distribution is very close to zero.
So they're all positive numbers, but they're very close to zero. But if you make that decay very, very low, you could actually create distributions that's almost like uniform on the, like, if you make the rate of decay, like for every unit, I don't know, like e to the point. Instead of e to the x, it's like e to the, well, what would be? Well, let's talk about half life here. Maybe that's an easier way.
Aaron: Because you already, you mentioned decay, right? Which of course implies half life.
Max: Right So it's half life of the crowd probability density. So let's say you have some value that you expect to be a certain order of magnitude, like maybe you think it's going to be in the 1000s or whatever, if you set the half-life life at like a billion, you can get something that's almost like a uniform distribution over the positive numbers, except in this case, it will be, it will still integrate to one.
So you'll still be able to use it. So you can, I mean, there's actually a way to consider a uniform distribution over all positive numbers, but that's kind of controversial. That's like an improper distribution. But like, if you don't want to use that you can kind of use the exponential distribution, as like a poor person's uniform-
I was gonna say poor man's and then that's, I felt like that was sexist in some ways. But now I realized that it's also classist, because we're saying poor. I don't want to- maybe I can't use that phrase anymore. Okay.
Aaron: This is the frugal person's.
Max: The frugal person’s! We’re getting very politically correct here.
Aaron: Have I complained to you yet about? We use Microsoft Teams at work for our chat software. So a few months ago, they rolled out a new feature that's called the speech coach.
Okay. So when I'm in a meeting, it basically records and analyzes everything I say. And it'll give feedback on things such as you've got a lot of filler words, you're saying you'll “um” a lot, and “you know” a lot, it'll count how many times each of those
Max: Thanks for not telling me as we use the podcast.
Aaron: But the other thing, one of the things that it calls out, and that made me really raise an eyebrow, the first time I saw it, is it will, it will ping you for non inclusive language. So if you're in a meeting, you say, “Okay, guys, listen up.” It'll take note of that. And it’ll say, oh, you used a gendered phrase four times in that 30 minute call?
Max: Did you see that slack bot? That like, there's a kind of a famous meme going around where somebody right, like writes on Slack, like, guys, the elevator is stuck. I'm stuck in the elevator. And it's like, “Oh no, you said ‘guys’, not gonna send this out.”
Aaron: No, I have not seen that.
Max: It was something like that. So but anyway, what does this have to do with exponential distribution? Why did, why did I get, why am I injecting woke ideology?
Aaron: We were talking about poor man's-
Max: Yeah, I know, I said, and all of a sudden, I was earlier, I was against the so-called woke ideology. And now all of a sudden, I am injecting it into my own podcast.
Aaron: I will say that dignifying it has at least a subconscious if not a conscious impact.
Max: It has. All right. Do you have anything to add about the exponential distribution?
Aaron: I like half lifes? Yeah. I think it's interesting that I can wrap your head around if we go back to not the beginning of the podcast, but the beginning of the pandemic, everyone was talking about exponential growth, this exponential growth that we hear less about exponential decay, which, which is it's lesser known sibling, but, yeah. Important in many venues.
Max: Right, right. Yeah. And so if you pull numbers from this, what does it look like? I mean, they're I guess, all of the exponential distributions kind of look the same. If you just scale them a little differently, they're all kind of the same shape. So you're gonna kind of just get you're just gonna get denser and denser as you go to zero, but there's some starting density and some halflife.
So I don't, you know, I don't know how else to describe it. But I feel like this is a very intuitive one, which at this time of night is a good one to end with?
Aaron: Well, so is there a kind of classic typical example of a case where the exponential distribution is very representative of a particular behavior?
Max: Well, I think it's a very good-
Aaron: I mean, we've already mentioned half life.
Max: But yeah, well, I think like, if you go back to the gamma distribution, when trying to come up with a distribution background or rate of something, a gamma distribution is often a very good prior. And because an exponential distribution is a special case of that, I think that that is also a very good prior for rate. I don't have a good example, other, like if you know, the probability of something kind of decays, as you go further and further into the future, then exponential distribution is probably a good one.
In fact, the decay of particles: probably a great one. Let's say you don't even know what the half life for a particle is. Maybe you even want to use an exponential distribution. You know what the half life is, but you want to kind of kind of see how long that particle is going to last, it might look something like that.
That could end up being a different distribution. But I'm pretty sure there's an exponential distribution in there somewhere I'd have
Aaron: Yeah, I would have to expect that.
Max: Yeah. So I'm doing it off the bit. But I suspect there are a lot of physical quantities that obey this one, for sure. It could also be like the distances between events, possibly. I don't know. All right. So yeah, I wish I had more to say about that. But look, this is a huge episode. One of the most fun in recent memories.
Alright, I think I think it's sad to say that we are coming up towards the end of the show. It's been a little more than an hour. So pretty perfect. Yeah. Do you have any last thoughts about today's episode, Aaron? And what's our overall theme?
Aaron: I'm gonna miss the studio. But you know, it's been a week full of exciting and possibly terrifying news on all fronts, whether you're a podcast bro, or you have money in a bank, or you're worried about the AI apocalypse. So hopefully next week is a little less interesting in the, in the style of the classic Chinese proverb.
Max: I don't know. I mean, if it's less interesting, then we'll have less fun stuff to talk about on the Local Maximum.
Aaron: I’d like occasionally for our current events, focused episodes to be less about panic.
Max: Yeah, I guess you're right.
Aaron: We're approaching this in a calm fashion.
Max: We shouldn't be rooting for the world to fall apart just so we have podcast material. That's, that's probably the wrong approach to life, I would say.
Aaron: Reporting to you live from the edge of the fiery crater.
Max: All right. So appreciate everyone listening. Remember to subscribe, Locals: maximum.locals.com, we are going to send an email soon to the, my entire email list about the fact that we've been doing this for nearly for over five years.
And, in fact, I don't know if we have any, like, if we have any milestones coming up in terms of like number of, like maybe total number of downloads or whatever. We're not really at a big round number.
But I suppose in like a year or so we'll be at half a million, that would be a cool number. We're at 380,000 all the time. But that's a lot if you think about it. This is a small sized city listening to us.
Aaron: And I know Max, that you've got a bunch of exciting episodes already lined up for the next couple of weeks. But we'll intersperse them. If you're listening, and you did not like the topics we covered today, send us an email or post something on the locals and tell us what we should be talking about.
Max: Awesome. No, I was actually, I was surprised by the number of people who really liked when I talked about multi armed bandits. Like this audience really likes when I get into a very nerdy technical topic.
Aaron: You have a type.
Max: Yeah, well I'm very, I was very happy to get that feedback, because sometimes I talk about those things. And I'm like, Well, I don't know if anyone's gonna like this, but apparently people do. So. All right, I guess we'll call it a show and take some pictures.
Aaron: What else would we call it?
Max: Okay, come on, I guess we will, we will adjourn this meeting of the Local Maximum. Until next time, when I guess you'll interview me about a well, okay, next week, actually, I'm going to be interviewing someone who works at a company with a big office in Connecticut called ASML.
You might not have heard of it. But it is a company that does hardware, semiconductor manufacturing. And so they that is kind of a different side to all of Information Technology, up to and including the current AI.
Aaron: Critical components underlying everything.
Max: Yeah, underlying Moore's law. So that is going to be a very interesting episode. And then hopefully the week after, maybe we could do a remote episode where you interview me about all the new things that I'm up to, and where I'm living and all that. So tune in in a couple of weeks. And then I have tons of interviews lined up.
And also I hope you enjoyed Susan Conover last week, who was on health AI. So, we've got a lot of really, really kind of entrepreneurial scientific people coming up on the Local Maximum in the future.
Aaron: I wish my job evaluation was done on the basis of, what was it? Top 3.75%? Like, like dermatologists?
Max: Oh, yes.
Aaron: That'd be a great criteria for my evaluations.
Max: Yeah. No, I hear you. Yeah. So I really liked that one. And I also really liked going to their party when they announced their security. That was actually the last. You know, one of the nice things about living here is you just drive right into Boston. And unlike New York, Boston is a city, you could just kind of drive in and drive out.
I mean, I never thought of Boston that way because when we were growing up, it was all construction, but you actually can do it. It's actually not like if I had to drive into Boston tonight, and then drive home, it'd be like for dinner, like it'd be fine. Whereas-
Aaron: Parking might be a pain in the butt, but it's doable.
Max: It’s not. It’s not that bad. I mean, the worst is it can be a little expensive, but Boston expensive is like 20, 30 bucks.
Aaron: Well, yeah, you've you've anchored on a rougher market.
Max: Yeah. On New York. Exactly. So. All right. Yeah. So bottom line is exciting and terrifying. That's I don't know if we'll name that show that because but, but that's the theme of the show.
Aaron: The Local Maximum: An Exciting and Terrifying Experience.
Max: Yeah. And a lot of exciting ones coming up. All right. I think I think that's it. I don't have any last thoughts. All right. Other than that, have a great week, everyone.
That's the show. To support the Local Maximum sign up for exclusive content and our online community at maximum.locals.com. A Local Maximum is available wherever podcasts are found. If you want to keep up, remember to subscribe on your podcast app. Also, check out the website with show notes and additional materials at localmaxradio.com. If you want to contact me the host, send an email to localmaxradio@gmail.com. Have a great week.
