A (3:52)

Yeah, fair question. I, I think two things. One is like, I'm a very heavy Noster user myself, you know, in a bad way, so I have a very good idea of what I want. And probably another, like, really big aspect is that I'm pretty close to Fiat Jeff. You know, I talk to him all the time and I, I follow his ideas. You know, like when he says something is good or something is bad, I actually, I listen to him. But really the main thing to take away from Wisp is that, yeah, there's a lot of focus on UX stability, making it fast, which is really important because like we have to, our apps have to compete with big tech apps. Like, people are not going to come to Nostr because it's decentralized, like, maybe a small portion will, but if we really want mass adoption, like the apps need to be awesome. So really I'm like building with that in mind, but at the same time, like, we couldn't take shortcuts that make big tech fast. Like having one central server like that, that basically sort of kills the idea of nostr. So we started like fundamentally at just using the Outbox model, you know, so even if you publish to your own relays, Wisp can find those notes. And what's also cool about that is it actually makes it faster because when I'm going out and connecting to, let's say, I don't know, 70 relays, I can respond to the user with the fastest relay instead of waiting for like one big one which might be slow, sometimes fast others. So it actually even speeds up the experience, I would say.

B (5:30)

So you. So first of all, I don't think we said it. First of all, it's Wisp Mobile is the website and it's Android only and you can get it as an apk. I think you get on the Play Store. Can you get on the Play Store?

A (5:46)

We're in a beta on Playstore, so only people invited can join. But it, it should be on Play Store probably like next week.

B (5:54)

The coolest experience is getting it on Zap Store. I think it all kind of. You just feel like you live in the future if you, if you download it through Zap Store and you send them a nice fat Zap if you. If you like the app. But. So utxo, would you. Do you think is Wisp for a mainstream user or is it for a power user? Or can it be for both?

A (6:14)

I mean, I certainly think it can be for both. I mean I would consider myself a power user and it's very rarely missing anything that I want. But I'm really trying to build it for just mass adoption. Normal user, not technical, doesn't know anything about relays or cryptography or private keys. You know, they just sign up. There's money in there. They didn't have to sign up for a Lightning Wallet, just. They're just. Money can just go into their account right away. They don't know how any of it works and it all just works. So I'm really hoping that's the case. But I don't know, you know, time will tell. But that is my goal at least is to make it appealing to just any Twitter, Instagram user, Twitch user, anything like that

B (7:00)

makes sense to me. So on that note, I really like. So we at Primal, we integrated Spark Wallet. I saw you chose the same route. It makes, I mean it makes the lightning onboarding super easy and straightforward and you have full Zap support, full on chain support. User doesn't have to manage liquidity, user doesn't have to have Bitcoin receive Bitcoin. The backups are interoperable. So you can take a seed created in Primal and just put it into Wisp, which is really cool and vice versa. You made a decision that we had disagreement with on the Primal team and we went the opposite direction, which is you encrypt, you. You give the user the option to encrypt their seed with their NSEC and then back it up to Relays. So all they need is their nsec. And as long as they have their nsec they also back up their wallet, which by. I love that if you can tell which side of the disagreement. I was in Primal hq. I think that is such a great ux. So kudos on that.

A (8:04)

Yeah, thanks. And actually, you know, credit goes to the Daniel who is the one who built that. And actually like even though Wisp shows you, hey, you should do that, we back it up anyway. It just happens automatically. And this has already saved a few users from, from losing their money of Course it was like, you know, three pennies or something.

B (8:24)

Yeah.

A (8:25)

But they said they're like, oh, man, I forgot to click on this. And I was like, oh, don't worry about it. You know, it's already on all the relays.

B (8:32)

Yeah. And the cool part there. The cool part there is. I mean, obviously, like, it's great that they can't. That it makes it much more difficult for them to lose their money. But just from like a UX point of view, it's really cool because what it means is, like, if someone opens wisp for their first time and another wallet is doing this spec, then basically all they do is paste in their nsec. And then all of a sudden their wallet and all. Everything is all set up and already ready to go. It all comes in one. That feels really smart to me. What do you say to the haters about Spark? Well, really eloquent.

A (9:10)

A lot of things I would say. I would. The first thing I would say is, like, for. For Noster zaps. Like, come on, man. Like, even the biggest accounts are not making money on zaps. There's nothing. You could lose that money and it'll be totally fine. I think, if you're comparing it to a custodian, because here's the truth, like, you know, 90% or whatever of NOSTR is using a custodian. If it's not Spark, it's actually harder for Spark to rug. I'm not saying they can't rug. They can, absolutely. But it's harder for them to do it than, you know, a provider. Like, I'll just pick on myself, like, node lists. Like, let's say I was trying to run it. Like, I could rug it way easier than Spark could Spark. So that's why. But of course, like, don't keep any real money in there. Like, whatever that amount is for you, 10 bucks, 20 bucks, whatever, withdraw it to your own node. Withdraw it on chain. Just keep your fun play Noster money on there. Throw your pennies around and don't worry about things like that. They're. They're much more important things to worry about.

B (10:09)

It's spending cash. It's how much. I mean, I don't know if it's good for this generation, but it's like how much cash you put in your back pocket when you go out at night. And, you know, you might get mugged on the street. I that in some cities, that's probably more than a Spark rug, but it's spending cash. You should think of it that way. And then the other piece is the privacy piece, which I don't know how you're handling zaps, but in our case it's, you know, Primal and Spark have insight into your transaction history. You know, I think for Nostr specifically, most zaps are, most zaps are public zaps. So you're not actually losing real privacy there. Obviously it could be more private. It could also be way less private. But it's an interesting trade off balance and I think it's just, it just works really well today. So we'll see if something better works in the future. I'll say for Primal at least we would switch. I think you probably say the same.

A (11:04)

Yeah. And I think when it comes to like, to privacy, like it's absolutely, it's not a privacy tool by any means. It's in fact it's sort of designed for the opposite of that. Like we want to show our zaps on Nostr and publish notes about it. So in this case, like using it just for a Nostr zap wallet. I think like trading off privacy is the right thing to trade off in this case.

B (11:27)

Yeah, yeah, it's just a really cool onboarding tool.

A (11:29)

You.

B (11:30)

So you did something. So I figure the structure, the general structure freaks that I'm going to go with here and UTXO for that matter is we're going to talk about like the details of wisp and then I want to go higher level with UTXO because I, I think he thinks about things. Interestingly, you made a really wild out there decision and you've changed the name of zaps to send money and you've denominated them in dollars. Defend yourself.

A (11:58)

All right, well, the first thing I'll say is when you sign in with like Amber or a signer, you will never see that.

B (12:03)

Okay, so no, that's your mainstream hurdle,

A (12:09)

basically. Yeah. I don't think any normie is going to go on Google Play and then like go download Amber. Like there's just no way they're going to do that. So only people who sign up for like a brand new NSEC are going to see that. But I think it makes, I don't know, this is my thesis is that a lot of people hate bitcoin. They see bitcoin and they're like, oh, it's crypto Donald Trump or like whatever and then they just run away or they think it's a big crypto scam that we're running with this app, you know, so that's, that's my thesis. I don't know, maybe it's right, maybe it's wrong. We're about to find out. So that's. That was my big decision for doing it. And I don't see the problem at all. It's not like I'm using stable coins or anything. I'm just literally doing a conversion rate. That's literally all I've done. She's changing a label, but you'd be surprised, man. So like just quickly on that, on wisp, here's another controversial thing I did for sats. I used the B symbol. Okay. I didn't use the lightning symbol because lightning is not sats. Like lightning is lightning or zaps, but. But it's not sats. That's not a unit. So I was like, at least bitcoin is a unit. And people are like, I'm never downloading wisp. You are using the bitcoin logo. I hate the bitcoin logo. It's like, oh my God, some people are out of their minds, man. But anyway, yeah,

B (13:30)

yeah. I mean, to be clear, I like, I'm. I like the send money experiment. It should be interesting to see how it plays out. I don't think this, I don't think it's necessarily a bad decision. I think it's kind of clever that you make it a default on basically fresh onboarding. I think that differentiates people decently well. I think it's a very low lift way of differentiating a power user versus a new normie. And I think there should be more experimentation on these little things, these light touch things that actually do mean a lot to a new user as they come in. I think the biggest confusion you might be present with is if bitcoin goes up or down significantly, which it tends to do. It has a history of doing that. And I know from example the samurai developers, basically, they had so many support requests for like, I had $50 in my wallet and now I have $20 in my wallet that they just removed dollars altogether. They were like one. You know, most wallets, even if you default to bitcoin, you still show the dollar amount. And even in that situation you have issues. So there's always like UX issues around it. But just something in terms of messaging to users. You know, you have bitcoin in here. So this is, you know, your, your 10 cents might be worth 50 cents, your 50 cents might be worth 25 cents. You could proactively get around some confusion, but generally I love the idea of send money. It's like nor it just normalizes bitcoin. Bitcoin's the best money. I'm just sending money to someone, I'm receiving money from someone. I think about it. Most, most people think about in dollars. Even the most die hard bitcoiners are constantly thinking about the conversion rate. I start every episode with it. You know that that is probably the more useful number. You also probably. I think it probably pushes zaps up, which is something to think about because right now it's like, okay, you, like, you don't think about it. You send 111 stats. Like I send so much, but like you sent them less than a dime.

A (15:28)

Exactly. Honest. So I am. Normie mode is not released yet. That's going to be when Google Play is approved. It's going to be released then. But I'm running that build right now for myself. And honestly, I can see it like if I'm going to zap you 21 cents, like the or 21 sats, the exchange rate today is 1.5 cents. Like that's actually. I feel kind of like I'm insulting somebody.

B (15:49)

It's offensive. You're like you're flipping them a penny. Imagine if you like walk past a homeless person or. No, even worse, like a museum musician or something. He's playing like a beautiful song. You walk up and you just super confidently, by the way, like he's just flick him a penny. Massive disrespect.

A (16:06)

Yeah, exactly. So it, for me personally, it has increased my, my zap amounts because yeah, it's too low when you look at it in pennies. It's too low, man. It's offensive. Just to me. Just to me. I'm not saying other people, it's offensive. Okay. I'm just saying for me, I feel bad when it looks like a penny.

B (16:25)

Yeah. But in SATS form, it's not offensive. Like if someone sends me 21 sats with like a salute emoji, I'm like, wow, respect.

A (16:33)

Yeah, absolutely. Because SATs are real money. You know, like a penny is just totally worthless.

B (16:38)

Yeah, it's interesting. It's. Let us know. I think there's a bunch of us that are actually interested in the experiment. So as it unfolds, I'm sure you will update us on nostr. But I'm pretty interested and I will say when the build comes out, I will. I'll create a fresh insect just to. Can I. If I'm a power user, I assume in the settings I can just toggle back and forth, right?

A (16:58)

Yeah.

B (16:58)

Yeah. Okay. So I'll just do that. I. I would like to live on the normie send money route for a little bit and see what happens. Okay, next. Next item on the agenda that doesn't exist is we got into a little bit of a tiff about follower counts, I think. How do you feel about follower accounts? That's my question for you.

A (17:20)

Yeah, I mean, I don't. The number one thing is there is literally no possible way to get accurate count. It is not possible. So no matter who tries and how you calculate it, everyone's gonna have different numbers and agree, like, what is, like what even is a follower? You know, like if what is a follower? What if my. Let's. Here's a really real example, right? Like maybe I want my relay to always pull your notes and I just put a like throwaway endpub to always follow Odell's thing and pull it into my relay. Like, is that a user? Is that a follower?

B (17:54)

Yes. Yes.

A (17:57)

Okay. Like me personally, I kind of think it has to be a human.

B (18:02)

But who set that up?

A (18:05)

Who set up the. That it has to be human?

B (18:07)

No, no, no. Like, did a, did a computer do that all by himself or did he. Did a human set up that a human.

A (18:15)

A human set up that a human set it up? I mean, these are maybe questions for the philosophers, man, but like the way it. It's. It's really, it's really tough, man. I mean, I, I don't know. I don't know the exact correct answer.

B (18:32)

I don't pretend to know the right answer either. I mean, so this goes. I think there's a couple intertwined aspects here, right? Follower counts, like what users see, engagements, right? So like likes and reposts and stuff. And then also just like total network user metrics. And because NOSTR is an open network, this is the first time, really like the real first time at scale where different front ends, different apps can interpret all that data in different ways, right? And it's. And so it's open to interpretation in a big way. While in closed tech land, like on TikTok, how do they count followers? You know, I mean, they. No one knows. And it's also a bullshit metric. There's no absolute number. It doesn't exist even in central media, you know, centralized cor. Corporate social media land. The idea of an objective follower count. So then the question, I think becomes, okay, who is the user and what is most useful to them? And I, I think that's kind of how you have to go from it. So I think you had some really interesting ideas around that. How are you thinking about that?

A (19:46)

So for me, when I even think about follower counts or stats in general, Yeah, I think a huge purpose of stats, especially like for big companies, is to like blow up their valuations or to like look better, you know, but for, for Nostr, I mean there are businesses like, okay, Primal wants to show good numbers, they're a business. Right. But maybe like an open source project would want something else, would maybe want something that's more realistic or more filtered. I'm not really sure. But for me, I think where follower count is actually useful and what we've done in WISP is only show followers that are in your network. So. And the reason for this, it helps me find people who I probably should follow because other people in my network also follow them. It's not really about account, it's about how like, sure, I should be that this is going to be a good follow up.

B (20:45)

So it's people you follow is people you follow who follow people that follow you.

A (20:54)

People you follow and who they follow that follow you. This other person that follows me. Sure, yeah.

B (21:04)

So if you follow nobody, you would get nobody.

A (21:07)

That's right. That's right. And it will scale. Like let's say you follow 10 people, you probably, your extended network will probably be like a hundred people.

B (21:19)

What? Because they follow an average 10 people each.

A (21:22)

Right. And there would be overlap in YouTubes and. Yeah.

B (21:27)

Is this the view you show if I go to your profile from my account is that the follower counted shows your network, not my network.

A (21:37)

It shows your network. So it's. So it's Odell. How many people?

B (21:40)

So it's people that follow you that I follow?

A (21:44)

Yes.

B (21:44)

Or they follow.

A (21:46)

Yes. It's relative to you. It's not like a global state. It's always relative to looking at it.

B (21:51)

If I go to your account, it's showing something different than if Pablo goes to your account.

A (21:56)

Yes.

B (21:57)

That's pretty cool. I like that.

A (21:59)

Yeah, so it's that way. It's more personalized. It's actually like useful to the user. It's not just like a ego number or something. It's actually just helping me make a better feed, make a better decision.

B (22:10)

But I, you know, maybe that's not even a follower count. That sounds to me more like a reputation score.

A (22:18)

I mean it could be what people will say to that. Is that just because I follow you doesn't mean I'm trying to give you plus reps, you know, like I might think you're a total dick, but it doesn't.

B (22:28)

There's plenty of, there's Ways for you to, to track what someone's saying without following them if you don't want to. That's just lazy. I think most of the time in Nosterland maybe that's not the case on centralized like the corporate alternatives. But I think in Nosterland if you follow someone like it's pretty, it's like don't take it as a full on endorsement. Like they shouldn't be like watching your children or like I don't even know, like hiring them for a job or. But there's at least a slight nod that I'm, you know, there's a mutual respect. That's why I'm following them.

A (23:02)

But maybe at least the content is, is somewhat quality. Like otherwise I wouldn't follow.

B (23:07)

Yeah, I'm building like a web of trust here. That's, I mean the reason I bring that up though is because like, oh, is it a reputation scores or follower counts? Because I think that's very useful. And I could see a world obviously I'm been the most focused on Primal. I could see the world where that is shown under what we show as follower count. But when we think about follower count, it is a direct one to one comparison to what X or Facebook or TikTok or Instagram would consider a follower count. And I do think there's a element to you go for the upper bound with that number to try and fake it till you make it get people into Nostr because that's what they're doing like I mean to the idea of normies, right. It's like they're used to the upper bound. X gives you the upper bound, Facebook gives you the upper bound, TikTok gives you the upper bound. So if it is a direct one to one comparison, we're always going to be on a back foot is how.

A (24:12)

Yeah, I mean you bring up a really good point man. And like it's, it's absolutely true. But I guess what really, what really gets me though is that when I put out that the web of Trust relay like a year and a half ago or something, I remember it specifically. So Primal was out. I had something like 9,000 followers or something on Primal. And then all these bots are trying to break my relay with web of trust and then like four days later I have 17,000 followers. So like I, I know that they're just end pubs that just got created and then thrown away. Right?

B (24:48)

Humans.

A (24:49)

So just, I guess to me it's just like I just, I know how fake the number is, right.

B (24:54)

And it bugs me but it's, it's not necessarily fake. It's just measuring something different.

A (24:59)

Right.

B (25:00)

I think it's. I, like, I honestly think it's defensible. I. That's why I'm saying, like, I think it says to. There's two different. There's two different pieces there. I think that. If, if the question is, okay, I want to do growth analytics on Nostr and try and improve it, improve its adoption and retention. I think retention is something that we have a lot of problem with. More so it's like it's even worse than just a straight adoption because if someone's first touch is bad, the second touch is so much harder to get. I think the upper bound is pretty much useless to you. And also the highly con. The very conservative bound is also probably useless. And what is helpful is somewhere in the middle. And my guess is when you think about, you know, the, the, the big tech, social media, that's probably how they're thinking about it too. It's like to their shareholders, they do the upper bound number and then internally, like if they're actually trying to do productive, pragmatic growth statistics and stuff, it's somewhere in the middle. But the lower bound also, I think, hurts us because it's like you're, you're, you're heavily discounting a, a large portion of people. Like, I do believe that 90% of people are lurkers or something. And there's really, there's really no way to algorithmically remove that. And so I do like your setup where it's not trying to do either of those. It's just like, okay, it's, it's in network, basically reputation. But I like, I like that as a concept. I don't know, it's complicated. And there's no obvious answer here.

A (26:42)

Yeah. And I guess just the really big thing is that like, you know, I, whatever follower count you want, I could just make a bot to give you that number, you know, so I feel like the, the web of trust or like some kind of AI inference or some, like you need some number. Especially because at least on Twitter and Instagram, there's a cost to signing up. Like you need an email address or a phone number or some sort of friction, but an mpub, I can literally make a million npubs in like 10 seconds, you know, with no cost and three lines of code. It's. The accounts are way cheaper on Nostr. So it's. I don't know, man.

B (27:19)

To be fair, we do do some limited. Like if you create, if you do like 10,000 at once. We do some limited scraping off the top, but it's mostly upper bound. Yeah, I don't know. I mean I. It's just a theory and there's no way to prove it. And that's the fucked up part. But when TikTok. TikTok was starting off like, I wouldn't be surprised if they had like a hard coded. Your follower count increases by 1% every time day.

A (27:51)

Yeah.

B (27:52)

You know, there was people that were like posting like beard chugging videos. They're like, I got a million followers. And if, if, if you know this is Noster doesn't exist in a vacuum. You kind of. Yeah, I mean you kind of have to play the game. I think I, I want to see more game players in Noster. We don't have enough. And that's why I respect Wisps game because I. It's good to have more game players that are like willing to like actually fight off the fight with the status quo like you. It's like being a pacifist and going to war is kind of how I think about it.

A (28:29)

Yeah, yeah, I agree. And, and look like you did convince me. So Noster archives another probably. I know we didn't get into this, but Noster archives is like the analytics site that I built and we're way more relaxed with the filters on that. So like, you know, the counts will be semi similar to Primal you'll see on Follow Counts or daily active users. Like we're just looking for at least one signal. Like maybe you lurk, but you got. If you just like something that's enough, you know. But if you truly. But they don't.

B (29:00)

They might not even have it. They might not even have an npub. Like that's like isn't the idea of a user. And like this is. I don't. I think part of the problem with Bitcoin is it had way too many philosophers came into Bitcoin. So I'm not trying to be a Noster philosopher, but the idea of a user with Noster does. It's. It's a very blurred line because the concept of user really comes from centralized walled gardens. Right. And so I've made mostly in jest in the past. Every human is an active user of Nostr because they can always read a note without permission. Like they don't need to sign in. They don't need to sign in. So. And I know this for a fact because I you. I have many haters. I don't know if the freaks are Aware. But there's. There's a lot of people that hate me. And they just look at my post@primal.net Odell like, they don't have a Nostr account. They hate Nostr, but they know what I'm posting. And I know because when every once in a while it pops off and there's a screen, there's screenshots that get circulated around. And then all the people that are looking at the screenshots, they're obviously, they don't have ampubs. So it does blur the lines. Like, what is a user? I don't know.

A (30:18)

Yeah, I also don't know, man. This is definitely a question for the philosophers. We gotta get Gigi in here to

B (30:23)

talk about it, man. Gigi is one of our best philosophers. He's a practical philosopher. But it does make. I don't know, it makes me. It's just something to think about. Like, I do think, you know, every good meme is based in reality there. We do not have that many users of Nostr, and it's probably better for us to be realistic about that and try and grow the number. But then also, sometimes I just feel like the community or whatever goes in the opposite direction a little bit too hard.

A (30:55)

And like, I mean, I do have some. Some thoughts on. This is like, a lot of times, if you ask the. The. All the people who are left on Noster is. This is like survivorship bias, right? It's like most of the people who tried Nostr have left, like, probably 98, 97% of all people who tried Noster, like, never came back. And all the people who really like Noster stayed and think that it's fine. Like, and of course they do, because they're the only ones left. All the just weirdos who liked it stayed. But if we want those 97% of people who didn't stay, we need a different product for them or different experience so that next time they will stay.

B (31:32)

Where do you get from? Like, I mean, so we met each other through the bitcoin community. Like, obviously, we've been in the bitcoin community for a while. What do you take from. Like, there's. It's not even a. It's not even that the bitcoin community didn't adopt Nostr and Mass, or that a portion of the bitcoin community is ambivalent to Noster. It's like there's like 40 to 50% of at least the bitcoin X community that fucking hates Nostr. Where do you think that comes from? Is that solvable or is that just. You just. You just say it is what it is and that's. That's done now.

A (32:16)

My. Yeah, I don't know. These are just guesses and thoughts. Is that they're trying to defend their follower counts or like the accounts that they've built up on Twitter and like Nostr sort of threatens that, even though that's so stupid. Because if you just come to Nostr, everyone's gonna follow you there. Like, what are you talking about? You know, like, I don't know. Do you lose your followers, Odell, when you came over on Twitter?

B (32:38)

100%.

A (32:40)

Yeah. Nobody follows you on Nasr. It's true.

B (32:42)

No, no, I mean, look, I think, I think quote unquote reach is definitely diminished, of course, I think, because you don't get the mainstream. But what's cool about Noster is the ride or dies, right? It's where the rider dies. Dwell without permission, and it's already a massive success in that regard. The question is, does it ever get the zip velocity? I think is the. Is the better question. And I do think it's funny with Nostr that. But. But you. That's solvable by. You don't have to go like, full Odell and delete your fucking, you know, massive X account. You can keep that and you can still use Nostr for different things. I think it's a little bit disappointing. A lot of people say that there's too much bitcoin stuff on Nostr, and as a result, I've tried to go out of my safety zone, which is like, usually I. Historically, X was my only social media and I only talked about bitcoin. And because I'm such a quote unquote large account on Nostr, I felt compelled to, like, you know, here's a blurred picture of my children or whatever. You know, like, here I. I have other hobbies. Like, here's the gun I carry. You know, I'm trying to, like, spread the message, but also at the same time, it's frustrating because if there's 20 more Bitcoiners, like high value Bitcoiners that start using Nostr actively, it would be way more useful to me. Like, there's not that I need more bitcoin content, but it's. And it's not necessarily that many people. It doesn't have to be 10,000 people. There's like a select 50, 100 people that if they started using Nostr even infrequently, it would be much better experience for me.

A (34:22)

Yeah, I agree and even there's a bunch of bitcoiners who are like primarily on X. And then they occasionally will come like Safedine just came the other day and posted on Nostr. Yeah, why don't you just always post on Noster, bro? Like, why is it just like once a year, like, what are you doing? So if just a few more of those guys came to Noster or just like, I don't know. Have you ever heard them articulate why they don't like Noster? Like a specific reason or they're just like, oh, it's just stupid and it's gonna die even though it never dies? Like,

B (34:55)

no, I mean that's what I'm trying to. I, I know there's. And I'll try and stay humble here. That I was a little like, I. Some of the blame probably comes to me because I was a little aggressive in the earlier days of Nostr. And so there's a group of people that hate me. So they hate Nostr and then there's,

A (35:16)

yeah, they can come on Noster and just shit talk you all day and they won't get banned. Like, that's a great place.

B (35:20)

That's a great use case. They can't get blocked. You know, I was hoping what they do is they take a screenshot of the Noster post and then they post it on X. So I can't respond. That's the key. That's the keys. But yeah, I don't pretend to know all the answers. I mean, with Primal, we're just trying to build a very, you know, user forward app that is takes reasonable trade offs. Because look, at the end of the day, like these people use X, there's a shit ton of fucking trade offs that they're taking and there's, there's probably like a happy middle ground. And I respect the fact that a lot of people like their first touch was disastrous and then they just never came back.

A (36:03)

Yeah. And I mean, you know, fair enough. Especially years ago when we're just like getting started. We have like Astral Ninja or whatever. Like, yeah, okay, that, that was, that's

B (36:12)

how I made my key, by the way, on Astral. It hasn't been compromised yet. Any day now.

A (36:20)

I think my old account was Astral,

B (36:22)

but then I made it in browser. I made it in browser live on a dispatch and I've been using it ever since. That's what I think is so funny. Like people like, oh, you gotta use Amber. Like blah, blah, blah. It's like, okay, like I just raw dog this NAC all over the place. And one day it'll probably get compromised, and I'll send out a post being like, I got compromised. And then I'll signal next to utxo, gg, Pablo, a couple other people in my personal web of trust, and they'll be like, odell got hacked because he was raw dogging. He's NSEC everywhere, and this is his new nsec. And then there we just solved key rotation. Like, it's not really that big of a deal.

A (37:01)

Exactly. I was just saying the same thing. I was getting to an argument with someone on Oscar about this. Look at American Hodl. How many Twitter accounts did that guy have?

B (37:08)

I think he got denied. Seriously?

A (37:10)

Yeah, like, and whatever. He got his followers back every single time. And most people are. Are also not American Hodl. You know, they have like 80 friends or, you know, 200 followers or whatever. Like, you'll get them back. It's not that serious to lose a key. I would just say when it comes to, like, you know, raw dogging insects, the. The main way you'll get screwed is by accidentally pasting it somewhere else. It's the handling of the key material.

B (37:35)

Yeah.

A (37:36)

It's not really the app.

B (37:37)

It's not like the app that's. Yeah. It's in your clipboard and then, you know, some other app or something. I. Yeah, I don't think it's. I think that's a funny one. Just, like, from a practical standpoint.

A (37:52)

Yeah.

B (37:52)

And it's always just a private web of trust. You know, it's like, oh, a signal message comes in from American Hodl. He's like, I made another. I made another Twitter account. Okay. What else do we want to talk about? I want to talk about. I saw you playing around with, like, the AI tools you have. First of all, download WISP. Why? When iOS, if you really want the normies, if you want the bitcoin Twitter influencers to use it, they're going to need an iOS app.

A (38:20)

Yep. It's true. It's true that. Just stay tuned for maybe some news on that, maybe not. We'll see. But I am very aware that normies like iOS. I am not oblivious to that.

B (38:34)

You should make the iOS one should be send money only. There's no way to get bitcoin mode.

A (38:40)

Exactly. Yeah. No bitcoin mode for iOS.

B (38:47)

So, AI tools, how are you thinking about them? To me, that's like the new rabbit hole. I'm, like, fascinated by it.

A (38:53)

Oh, yeah. I'm. I'm super. I'm getting so Deep into like the local AI and setting up your own rigs and doing this kind of thing right now. Oh, AI is. AI is crazy, man. Like, I'm. I don't know, I think a lot of people, like just call it like slop or. Oh, it's just like.

B (39:08)

Well, there's a lot of slopes, of course.

A (39:10)

There's a lot of slop, for sure.

B (39:11)

A lot of human slop too.

A (39:14)

Yeah, a lot of human slop, I would say for coding. It's. It is legit, man. Like, it actually, you know, like, I am a coder, okay. I spent like 20 years learning to code, okay. And this thing runs circles around me. It makes me look like a terrible dev. And that's okay. I just had to like let go of that identity, you know, I'm just. That's okay. The robot beat me. So now I have to wield the robot, you know? Like, that's. That's all there is to it. So I think, I don't think AI is a bunch of slop, especially for coding. Like, I think Claude is absolutely insane. That's sort of like my high level thoughts. But really what I'm super excited about is the open source models that are coming out now, man. Like, Quinn 3.6 is legitimately a good model. Like if you. I know, I know. My little bot that I made this morning has like, whatever, some bugs. A little retarded, but like.

B (40:05)

He's retarded.

A (40:07)

Yeah, he's completely retarded. Quentin 3.6 can run OpenClaw, like and do web searches and pretty much do everything other than coding. And you can run that on pretty cheap hardware today.

B (40:19)

What do you. Are you running in locally, that box?

A (40:22)

Yeah, locally? Yeah, it's all local. It's on my GPU server.

B (40:25)

What is it running on? What is it running on?

A (40:28)

So I have four 3090 RTX GPUs, and then they're. It's served up with VLLM. So it's like a. Yeah, like a server software and yeah, it can handle something.

B (40:43)

Like, what does that cost?

A (40:46)

Well, if you wanted to buy it today, it will cost a lot. But the thing is I. I have it as a.

B (40:50)

Is that like $10,000 or what is it?

A (40:53)

I think each RTX is probably like $3,000 right now. So just GPUs. Probably like 10 to 12,000.

B (41:00)

Like a 15K build.

A (41:02)

Yeah, something like that. But I built this like three years ago and it was way cheaper. I got those GPUs for like $600.

B (41:08)

That's wild.

A (41:10)

Yeah, so they're way cheaper. And even the Ram, like, I'm pretty sure I spent like 200 bucks on Ram. The Ram I have now is like two grand.

B (41:18)

Physical commodities.

A (41:21)

Yeah. But what I will say though is like, I set it up to run as a server so that, you know, dozens of people on Nostr can all use it. If you just want to use it for yourself, you do not need a build like that. You just need one gpu. So this is what you already have works.

B (41:36)

This is Quinn 3.6, I thought. I mean, the latest one I saw was 3.5. Did they just come out with 3.6?

A (41:41)

They just came out with 3.6, yeah. And it's a, it's a mixture of experts, so it runs way smaller. Like, it, like most of the responses are on 3 billion parameters, which is like super small and fast, but it's still really smart.

B (41:56)

Yeah, they. I mean, I thought it was kind of cool with your bot that you just like, it just understands Nostr, so you just pay it with zaps and then it. And then it proceeds. You start to see like the real, like, it's all coming together. Right. It's like you got the Noster piece, you got the bitcoin piece, you have the open source AI piece, and they just really complement each other really, really well.

A (42:23)

Yep.

B (42:24)

Does it make you more bullish on Noster?

A (42:27)

I mean, I, I hope my actions speak harder than my words on how bullish I am on Noster, Man. Like, I am mega bullish. Well, here's what I. I don't know what, how other people are going to perceive Nostr. And it's, you know, at the end of the day, this is network effect technology. Right. I can be bullish all I want. If other people don't adopt it, then it doesn't matter. But I just legitimately just like looking out at, at the world and like social and communications, like, I just can't see. It's. It's so good, man. There's just, there's nothing that comes close to as good as Noster is. And I just can't believe that people don't see it.

B (43:10)

Yeah, we have, I mean, we have everything except the people. I just. That one key part. I. So, I mean, do you use the Frontier models too? The proprietary, top of the line, America made NSA tech four. Seven?

A (43:27)

Absolutely. Like, the only reason Wisp is good is because Claude is good. That's literally it.

B (43:32)

So I mean, I. The gap between that and this, the whatever quin 36 or I don't know what. The Google's recent open source model is actually pretty good.

A (43:45)

Yeah, Gemma 4.

B (43:46)

Yeah, Gemma 4. The gap is still fucking huge.

A (43:50)

Yeah, it's not. It's not even close, man. But for coding, it's not. But for like Chatbot, let's say, like ChatGPT, you just want to, like, talk to it. Research stuff. It actually is comparable now.

B (44:00)

Man, I really hate the, like, the reply spam bot stuff. Like, who do they think they are? Like, it's obvious what's going on there and they must be spending a bunch of money. Unless they're running local models.

A (44:15)

They're running local models for a fact.

B (44:17)

The Bitcoin Libertarian. Do you know who I'm talking about? The guy who's like, responding with, like, in Spanish too. He's got an English one and a Spanish one that responds to every one of my posts.

A (44:25)

Yep. Bitcoin Libertarian. Was it base truth, Primal diet, or there's all sorts of them, but yeah, Primal Diet.

B (44:34)

What is that? He just responds to every one of my posts being like, this didn't discuss meat. It's like, thanks, bro.

A (44:42)

It's so fucking stupid. But my.

B (44:44)

It's gonna get worse.

A (44:46)

My N Spam model catches all of those, man. You don't see any of those.

B (44:49)

So let's talk about. Let's talk about the N spam. So this runs locally on a relay, and it's a very, very lightweight local model. Right.

A (44:56)

It runs on your phone.

B (44:58)

Oh, so it runs on the app level. So is it running on. If I'm running wisp, it's just running on my Android device.

A (45:05)

Yep, it runs on your Android device. And it. So like, I also use it on my relay. You can use it anywhere. But wisp, like, you don't have to use my relay to take advantage of it. Your device will look at anybody who. I don't follow already. It will check. Hey, is this. Is this a bot? So it's not like, every single node?

B (45:27)

What about, like, the good bots I like? There are good bots out there.

A (45:32)

Yeah. Like Recbot.

B (45:34)

Yeah, Recbot is one of my favorite bots. Is how I know the price you run that. I mean, I've been trying to build out, like, that Citadel Wire news bot. Like, is it getting caught up in ends in people's N spams?

A (45:45)

No. So N Spam only looks at replies and notifications. So if you're just posting to your feed, it doesn't. It doesn't check.

B (45:53)

Yeah, see that I like. I. It's. It's an open protocol, so I can't stop you from doing whatever you want and ultimately, you know, give the users more tools so they can consume how they want. But if, if you're listening to this and you have like a low energy reply bot, just like fucking turn the thing off. I don't think you're doing anybody favors. And it's just hilarious as an open protocol because it's like I saw someone today, so the bitcoin libertarian guy, like bot controlled by a human responded in Spanish and English to me. And then I guess the Spanish version got auto translated in Amethyst. So then someone from Amethyst responded to the auto translated one in English which then responded back in Spanish. And it was just like, it was just a funny example of open protocols, right? Where he has an English bot and a Spanish bot, but you see them both as English because you have auto translate on and then the humans replying back asking questions to this fucking bot. But the ones that like actually post to direct feeds, I think are a little bit more interesting to me. Like I think, I don't know, I think where we're going, everyone, most people will have you know, like the robot assistant of varying levels. You know, some people will have like the big tech surveilled and controlled one, other people will have more freedom, focus, setups. But like if you need a chatbot interface, like it's probably not going to be in the replies unless it's, you know, I guess dunking on someone where you want it to be a public back and forth and so done right, that could work. But just like the random replies thing, I don't think will ever really be a thing anyway. It's long winded. But the bots that own their own feed, right? So like Wreckedbot for instance, which I think is like a low lift, perfect example of the power here. It's a real, I, I think it's a real advantage of nostr. I don't, I don't know how it actually plays out. But like anyone who's playing with these things, if you want it to be an Xbot or a Facebook bot or a TikTok bot or something, you know, you have a permissioned API, it might cost you a shit ton of money, you probably have to KYC to use it. You'll have all these different random errors because you have a centralized dependency. The fact that you can just, you know, tell your bot to spin up an NSEC and then have like you, there's no API requirements, you just generate a fucking NSEC, you can generate infinite NSECs, you can just broadcast to the fucking world in a relatively robust way with not you having to deal with no backend whatsoever is amazing. And I just, I feel like that could be fucking huge. I just don't know how it actually plays out in practice.

A (48:40)

Yeah, I don't know either. And, and you know, I could tell you a cool story actually about Wreckbot. So you know how like whenever the price really spikes, it just goes crazy and there's like tons of messages. So most of the big relays will rate limit you. You know, they don't let you post 30 messages in 10 seconds. So that, that kept happening to Rec Bot. So what's also cool about NOSTR is like, okay, I'm getting rate limited by these relays. I just made my own relay and I publish all my notes there. And it has. And nobody can stop me now. Everybody can get Recbot's notes. Even though, like the rate limits are totally legit, you know, I put rate limits on my.

B (49:16)

Yeah, we whitelisted the Primal Relay. Whitelisted Rec Bot.

A (49:20)

Okay, nice.

B (49:21)

We believe in all the wreckage just flowing good. Thank you.

A (49:26)

But yeah, it's just all these layers, you know, it's just so unstoppable. But the problem is that nobody actually cares about that. That's. That can't be like our selling point to people because nobody is saying but.

B (49:38)

But what I think it is is not. Look, I don't pretend like Citadel Wire is like this wonderful gift to man to man. I am trying. I'm playing around with these things because that's how I learned best. And also, like, I would like a very condensed, high signal, no news news service basically without all the engagement bait. Like I, as, as, as the leader of my family, as, you know, running tons of projects and, and businesses in the space. Like, I need to be up to date on what's going on without all the. And so I'm building something that I want to see myself so it's not a waste of time. And I'm trying to learn. But my point is, is the ability for someone to create a very good, a very useful automated service is not the feature. The feature is when someone does, I think when someone actually cracks the nut and like has something that is actually very useful to a large amount of people, it could be the thing that drives them into nostr. Right. And I think like, part of what I've been playing around with, with Citadel Wire is for the normies. They can just go to citadelwire,.com, which is effectively like a Noster app. It's like a Noster web app that only shows one account. Like, that's what it is. Like, the core of it is just a Noster note. And then eventually, then they might move in, right? And so if I'm thinking about, like, what is a major one, I don't know if you noticed. Like, we almost. We almost got the Jet tracker guy, the guy who tracks all the rich people's jets. He had different accounts for each jet, and he got banned off of X for his Elon Jet Tracker. What a coincidence. That was the only account that he got banned, and he started a Noster account. But why do I bring this up? I bring this up because, like, he also has a website, right? And so it's someone who creates something like that where it's, you know, a million people are hitting his website, and then 50,000 are realizing that on the back end, it's Noster notes. And when the website's down, they can just go interact and zap and comment on any of the notes in the background. I think long winded.

A (51:45)

Okay. Yeah, I. I didn't. I didn't know that he's on. So he is posting on Noster. Like, those trackers are Noster notes.

B (51:53)

No, we. We lost him. I think we lost him.

A (51:56)

Damn.

B (51:57)

Let me. Let me do a quick search. Elon Jet Tracker. Elon Jet Tracker. He was on Noster for a bit. Oh, yeah, here, one year ago, I was posting about him. Nope. Well, we need to improve search, too. But we had Elon's Jet Tracker on Noster for a bit for like a year or so. Here it is three years ago. But, yeah, then we lost him. But I think that's a perfect example. Like, it's so random. You know, it's not something that you really think about, but the guy was getting, like, millions of views on people that just wanted to track where rich people were traveling, which is all public information, right?

A (52:49)

Yeah, man. And I know, like, so, you know, I've been working on just, like, building Wisp, but, like, it's pretty much got all the features I want. Of course, like, every app needs maintenance and stuff, but most of my work that I see myself doing is on marketing and trying to reach out to, like, let's say, have you ever heard of kik.com, probably?

B (53:09)

Yeah, the Twitch competitor.

A (53:12)

Yeah. So the whole reason they were born is because Twitch keeps banning everybody all the time. Like, you say, retarded.

B (53:19)

It's like Rumble.

A (53:20)

Yeah, exactly. It's exactly that. Right? So. But even on Kik, even now people are getting banned from Kik because of

B (53:26)

course they didn't solve anything. They just launched another service.

A (53:30)

Exactly. Now look, if to get banned on Kik, you gotta be pretty degenerate, man. But NOSR actually offers a viable alternative to these guys where they literally finally cannot get banned. And like, they actually have a problem with getting banned. Normies who are talking about the sports ball game, they're never gonna get banned from Twitter. It's fine. But like some of these Twitch people, they have like tens of thousands, maybe hundreds of thousands of people that want to watch these super D gen streams. And Nasser is pretty much the only solution that, that they have, I think. But right now what they're doing is streaming over Discord.

B (54:06)

Yeah. Why is.

A (54:09)

Well, Discord is very uncensored, man. Very uncensored. At least for now. I don't know if they're. If they did that KYC thing yet,

B (54:16)

but I think they freaked out after they, after like Charlie Kirk's assassin was like groomed through Discord. They started. They're doing a bunch of KYC stuff. Obviously it's completely centralized. I mean, so like this one hits close to home, right? Because that was one of the main reasons we prioritized video streaming on Primal. It's so hard to break into that. I don't know, it's clearly a dope. Like, it's cool that if you broadcast on Primal, Wisp has support for Nosture, has support for it, Amethyst has support for it. You can just see it in whatever app you're using. That's obviously a massive advantage to us. I think Zaps and Stream are really fucking cool. Like, obviously in the corporate world we've seen, you know, YouTube, super chats and whatever the Twitch version is, the Kick version, they all tick tock version, like sending ice cream to each other. Like, people love that and I think we can do that better than them. But it's so hard to break into. And to your example, I mean, Kik didn't kick like pay a bunch of influencers like $500,000 a pop to Switch,

A (55:20)

but way more than that. Way more.

B (55:23)

Yeah. So you're going to do that?

A (55:26)

No, I think I'll just say because most of these guys, like, they're only kick only because they're banned from the other ones. Right? But I can just reach out to these people and be like, hey, why don't you just add Nostr to your stack? You're already doing YouTube and Twitch, like just add Nostr into your obs or whatever and broadcast to it like that. That's it. It's not much more. You'll make a little bit of money. It's no extra work. But do you even know that NOSTR exists? They probably don't. Right. So at least like an attempt to do it, to do some outreach and

B (55:57)

like, do you tell them do you should broadcast an answer or do you tell them you should use wisp?

A (56:04)

Well, I'll probably in this case tell them to use Zap Stream because they can't broadcast from wisp. But I'll explain that there's this whole audience of people who are willing to send you bitcoin that. That's the benefit of why you should add, um, you. You don't even have to monitor the chat or anything, like just to get our content up, you know. So it's not just only no good radio streaming all the time and nobody else.

B (56:27)

Well, RHR has been doing it for like three years now.

A (56:31)

That's true. But it's a rare show, you know, it's a gem.

B (56:34)

It's a once a week. Yeah, I don't know. I. I mean, maybe it doesn't take that much. Maybe it's just, you know, one or two prolific people. I mean, I always joke around just like, if I. I don't. I'm like such a boomer with this, these people. But like, if I show speed, like started broadcasting on Master, it's like the only one I know. Like we'd easily get a hundred thousand users out of that shit.

A (57:03)

It totally, man.

B (57:04)

Just one person.

A (57:05)

And that's just like one example of one streamer. It could be Instagram Simps. It could be just Safedine posting more. Like, it could. We just have to get the content creators who already have an audience and not even necessarily migrate it, just add it, just build it into your stack that Nostr is also a place that you can publish that they know they can control. You know, that Twitter is never going to demonetize or Noster is never going to demonetize the way Twitter might, you know?

B (57:33)

Yeah, I don't know. It's interesting. It's.

A (57:41)

I don't know.

B (57:41)

Yeah, network effect is a bitch. What about the piece that I do think, like, if you're just doing bare mirroring, it's like completely useless.

A (57:54)

I don't, I don't think. It's obviously not as good as like actually engaging with chat and whatever, but there. I don't think it is useless to have to have content, like, ready to go, even if it's just mirrored content. You know, you. You hop onto Noster, it doesn't feel stale, it feels lively.

B (58:12)

So then should we just do that without their consent? Should we just be scraping big accounts and just posting it? If mirroring is fine.

A (58:22)

I mean, I'll tell you, Sir Sleepy does that a lot. He, like, reposts old movies or whatever. And I like watching it sometimes, you know, I would. I think that is a good idea. I think people should do that. Like, if there's a content creator that you like and you wish their content was on Noster, just put it on Nostr. You know, no one can stop you from doing it. So just do it.

B (58:42)

Is that person a user?

A (58:46)

It's just one user. See that? The new account that's being mirrored is not a user, but there's one power user who has two accounts, but it's just one user.

B (58:54)

Accounts okay or no. Yeah, I mean, I think. Yeah. I don't know. I go back and forth on. I mean, obviously there's very useful people. Like, Gleason was running Trump Bot for a while, and it's like, okay, like, he's the President of the United States and his, like, Truth Social posts, like, move markets. Be nice to. Even if he's not engaging, it'd be nice to. I mean, I doubt he's engaging people on Truth Social either. He's not, like, replying in the comments, being like, that's a great idea. So, like, his mirroring is great. He stopped doing that. So maybe I should be the change I want to see in the world and do that. But I think for, I don't know, a lot of them, it does feel stale. And maybe it feels less stale if someone does it without their permission. Like, if I go and mirror Luke Groman's tweets to Noster and it's like, I'm doing it. He's not pretending we're beneath him. He just doesn't think of us. But if he is, just whatever he tweets out just automatically goes to Nasser. It feels like disrespect. I don't know. It feels. You know what I mean? Does that make sense?

A (60:02)

I do know there's. And there's a limit, I guess, of, like, how famous the person is before it's acceptable, you know, like, of course the President. Do whatever you want. Right? But, like, if it's your boy or someone who's, like, trying to build their own business or something, maybe. But I. I don't know, man. Because you. It can't be stopped. Like, all the Mastodon bridges are basically doing that. Right.

B (60:23)

Well, let me. Let me use it as an example of one of my boys who I think cares about Noster, but uses both. Right. And that's Peony Lane. Peony Lane, Penny Lane. Penis Lane. I don't know how to pronounce it. The best, best wine in the business. Right. If he was just blank mirroring his tweets, like, it would be just complete disrespect. But instead he. He just lives two lives. He lives like an ex life and he lives. Or Lynn Alden is another example.

A (60:55)

Yeah.

B (60:56)

Right. She has different posts that she posts on Noster. It was like, when it's like the blatant. I'm just adding it as a field in my automatic post thing. It's just straight disrespect.

A (61:08)

I. Look, I'm not going to name names, but I. I can immediately think of, like, 10 people who are doing that. And, like, so my sense is that. But then I also look at their engagement. They get on those posts and people do, like, those posts still, you know, and they still reply and they still zap and they still like it.

B (61:25)

Yeah. It's fascinating.

A (61:27)

So I. I think it's okay, man. It's not your cup of tea. That's. That's totally fine. But I think a lot of people are into it still, as long as they log in and occasionally, like, reply back.

B (61:35)

Or occasionally reply back is the key.

A (61:39)

Yeah. Or at least see it, you know? Come on.

B (61:43)

Okay. I mean, Greg G is another one. All the merchants, they live in both worlds. Soap miner, but they don't.

A (61:53)

If you're running a business, you kind of just have to be on Twitter. Right?

B (61:56)

Like, I mean, yeah, you could probably be on TikTok.

A (62:00)

Yeah.

B (62:01)

Because there's more people there.

A (62:04)

Yeah, absolutely. Like, you got. You got to make sales, man. This is not about purity. This is just about making money through different marketing channels.

B (62:11)

Yeah. I will say Ben. Ben the wine dealer, he should get credit for, like, actually using Noster consistently, but he does not get enough shit that he doesn't accept Bitcoin for his wine on his website. It has to do with, like, some regulations with shipping wine.

A (62:29)

Whatever. Okay.

B (62:31)

We're not. We're not quite there yet. What? I don't know. You have anything else interesting to talk about? I've enjoyed this.

A (62:41)

Yeah, me too. My Inspam, NASA archives wisp. I mean, that's pretty much the big things I'm working on right now.

B (62:48)

Well, let's just the AI tools a little bit. Your flow on, like your daily basis using. You mentioned Claude, are you just. Are you. You have a subscription or are you going through an API? How are you handling that?

A (63:03)

So I have both. Like, I have their max plan, but I still.

B (63:08)

200amonth?

A (63:10)

Yeah, yeah, 200. It's 300 Canadian dollars.

B (63:15)

Oh, is send money going to show Canadian dollars if you're in Canadia or

A (63:20)

if you pick it, you could pick whatever currency you want. But yeah, it's. It's every currency. See, is it not?

B (63:25)

Okay, sorry. Anyway, it should just be dollars. I think it's like the one fiat to rule them all, but yeah, of

A (63:31)

course an American would say that.

B (63:34)

Okay, so it costs you 300 Canadian dollars to have your cloud max plan. So you have that.

A (63:41)

Yeah, so I have that, but it usually gets maxed out. So I also have the APIs and Codex, like, so Codex is more like my secondary one. So usually how my flow goes is that like, it depends on the difficulty of the task. But a lot of times what I'll do is I'll use Claude in plan mode. So get Claude to make a really good plan, review all the code base or whatever, and then I'll use a cheaper model like Codex or even my local Quinn to actually implement the plan, depending on the difficulty. So that's like one way I'm trying to like, get around the limits because I'm not buying a second account. That's just way too much.

B (64:16)

Well, I mean, right now, like Claude, Max, or whatever you're getting, supposedly you're getting like $5,000 of value for your. Well, I guess in Canadian dollars, it's like $10,000. It's like $7,500 worth of value for your 300. I honestly, they're like, maximal, doesn't.

A (64:37)

Yeah, yeah. Or even more, if I had to hire a developer, and I was in a business where I did hire developers, and I had to hire someone who outputted work as cleanly as Claude did, I'd be paying them 10, 20, $30,000 a month. I get it for 300 bucks and they work all day. Like, that's insane.

B (64:52)

Yeah, I don't think that lasts. I think that's the interesting. I think that's where it starts to get interesting. I'm kind of curious in your opinion, but, like, I think the skill set here is like maximizing cost per output quality. And right now the line is blurred because you have these subsidized subscription models. So there's like a bunch of lazy people using this stuff or they're using this. They're not necessarily lazy people, but they're people using this in a lazy way. Because if you're getting $5,000 worth of token usage for, for $200, you're going to use it in a lazy way. But if you have to spend $5,000 on it, you're going to have to figure out a way more efficient way of handling it. And that's the hardest part right now. And that will be the hardest part. That's the skill, I think.

A (65:43)

Yeah, it's more like about product development, like having a good product that you're working towards. Not just like, oh, am I just dicking around? Because this is fun and you know, whatever, I'll experiment. Like if your experiment costs money, then you probably wouldn't be doing it. But yeah, and I could tell you like too. So take my rig for example. Let's say you wanted, you're like, no, I, I want to do local AI and this and that, you know, okay, you can spend 15 grand on this rig to have a bot that's is 30 times more retarded than Claude. And how long would it take you to break even if you had just bought Claude? You know, that's like, Claude is like seven, eight years it would have taken you break even with this terrible rig. Like obviously you should just buy Claude. And especially if you're doing open source work, you know, like, I have no need for privacy when I'm building Wisp. All that code is published. I need zero privacy from my LLM to build this stuff.

B (66:39)

Yeah, it's not like you're, you're not making a privacy trade off at all. Yeah, because it's already going to be public. It's going to be a public release. No. Yeah, I think particularly if you're trying to go for like the higher end open source models. Like that is a luxury good. Like if you're buying hardware for that right now, you know, if you have particularly very opsex sensitive situation, like it makes sense because you're doing it strictly for the privacy reasons. But do not delude yourself into thinking that you're saving money going the local route right now. And then there's the separate piece where it's like the small light stuff like what you're doing on the phone for spam, already capable enough. Super interesting. But if you think like there's so many just hype people, right, that are like, oh, like you download the latest Chinese open source model and it's like you can you, you're getting Opus 4:7 qualities. Like you're, you're not, you're not even getting close and then you're at a, you're in a negative disadvantage to other people. Which is why I think like the, like the PPQ concepts are really, is really cool. Or like Venice, I think like that concept is really cool where you have many different models you can choose from. You're not doing kyc, you're paying for it on a pay per use. Yes, you're giving up privacy if you're hitting the walled garden models. But there's a balancing act there. There's like, okay, I use local for some things, I use end to end encrypted for other things. And then I use, you know, Claude for heavy lifting stuff that maybe doesn't have as much privacy concerns. There's like a balancing act that people have to find.

A (68:24)

Yeah. And also a big part of it is how you're actually using the LLMs. Like sure, maybe you use some of these privacy focused services and they didn't collect your identity, but maybe in the course of doing your development you gave them your database credentials or your NSEC or something, you know, like that is still vulnerable to exploits. And I think that just happened the other day, right. Lovable got hacked and all these like all everyone's chat history got leaked. And so that can still happen even on privacy focused ones. It's just that your identity is not associated

B (68:57)

Lovable. I mean definitely didn't pitch itself as a privacy like it was keeping all the records in the cloud and shit. And that's why it got taken down. I mean that's why it got hit.

A (69:09)

Yeah, totally. But I mean of course like Claude is keeping this and Gemini and all

B (69:14)

the endpoints are, the end points are like obvious. It's like a VPN trust model. Right. So you're trusting them. Right. But like with ppq, like I don't think Matt Alborg is keeping logs.

A (69:28)

Of course not.

B (69:28)

Yeah, Lovable. I mean, and it's not just because I think he's a good guy with high integrity and a lot of proof of work in the space, which he is. It's because I also think he's smart enough to realize that it adds extra liability onto his business and doesn't, doesn't want to lie to the user and do it. But you are trusting him to do that. Now the end point, but Lovable, like that was one of their features was like we save all your shit like Telegram does. And so it's like you Got good ux. But then the endpoint, obviously depending on what inputs you were putting into it, is going to figure out who you are. I mean, I, I think Anthropic came out and said like they named a specific Chinese researcher that was using it to train his. I forget which open source model it was. And like he was using one of the API access points without providing his identity. But like the things he was uploading, they like figured it out pretty quickly who, who the guy was.

A (70:25)

Yeah, yeah, it's. You have it. It is wild times. And the thing is too is like these models, they're. I don't think they're trying to be malicious, but they do things like you tell them, okay, you can't read my.env file, right? So they're like, oh, I'm blocked, let me try another way. And then they'll just output it to the terminal and they're like, okay, got it. You know, so no matter how hard you try, if you're, if you're using these cloud models, man, they're going to get your database like the big model endpoints, they're going to get all of that data. So just be careful out there. Freaks.

B (70:58)

Yeah, I don't know, it very much feels like the Wild west, which is what I find interesting about it. How do you think about. I guess I should ask you about this ahead of time. Do you have a hard stop?

A (71:09)

No.

B (71:11)

Okay. You might regret saying that. Okay, we're gonna continue. How do you feel about like next five years of this stuff? We're gonna see massive job loss. Like what it. What are your current thoughts? And we're not gonna hold you to it because no one really knows what happens next.

A (71:24)

Yeah, yeah, I. It's so hard to make such prediction because like, look how far AI came in just a couple of years, you know, it's actually insane, but I, I do strongly believe we are sort of approaching a limit in terms of how smart the LLMs can get and really like what scales now is maybe like how accurate it is, that it's more consistently always giving you a good output, that it's faster and that it can get cheaper, but in terms of like how much smarter it can get. Like, I'm sure it can get like a little bit smarter or Maybe let's say 2x smarter, but I don't think we're going to see like x smarter in five years. I think that we're about to like sort of hit a wall. Yeah, that's sort of how I see it in Terms of like, job loss. I don't. I think it's like more job creation is there's gonna be all these jobs for people to implement AI and nothing works. It's gonna be like everyone's working on Noster everywhere in the world because, yeah, every company is like, oh my God, we have to implement AI. And then they like, they try to implement it, they make all these new systems. It doesn't work. You need. Now you need security researchers and more developers. And yeah, I don't know. So I don't think it's going to be like a big panacea. I think it's very useful and I think it's like going to be improving people's search and reporting and maybe like help you with your PowerPoint or whatever. But I don't see it as like mass layoff type of thing happening. That's. Yeah, that's basically what I think right now. Really subject to change. I mean, I could definitely be wrong. Okay. I'm staying humble, but that's what I currently think.

B (72:58)

Yeah, I, I mean, I don't know. I mean, what, what I see in maybe, maybe this. I think startups are the first mover because they're the most likely to die. And particularly bitcoin startups were already kind of operating pretty lean because their cost of capital is bitcoin. So like, I will, you know, I will put out there that my bias is obviously companies that are led by bitcoiners, not necessarily even bitcoin companies, but companies that are led by bitcoiners, which is what 1031 has exposure to and what I have exposure to on a constant basis. Those founders have always been more lean. Right. They know that they could rather hold bitcoin than go off on some spending journey. And they never assume the next check is coming because we just don't live in fiat land where they can raise Series G and they're just constantly raising money. Money's tight in bitcoin land, but they're implementing AI at a wild pace. And usually what we see happen isn't like, yeah, there's the more sophisticated ones that are maybe running one local model and trying to do things, but most of the time it's just straight outsourced to big tech. Right. It's. And it's mostly Claude, but, you know, probably there's actually a decent amount of the Google side just because almost every business runs on G suite. It's really hard to run a business without G suite. So Gemini is already like ready to go and then A lot of Claude usage. But, yeah, I don't know. I think so. The question to me becomes, where am I going with this? The question to me becomes, like, when does it hit Google? Google has 190,000 employees. Like, it doesn't have to hit every business to have, like, a massive outsized impact on the economy in terms of job loss. It just like, the big tech companies alone are maybe 10 million employees or something. If that gets cut down to 2, and all of a sudden you have 8 million people that we're making outrageous bullshit salaries of like 4k a year or whatever, and they all get let go, we can have, like, pretty significant effects. And I. I think that's where we're going. And I think people are, like, trying to paint it in a rosier picture. But, yeah, I don't know.

A (75:34)

I. I think you. Like, there are good arguments, and I feel like it always, like, seems that way sort of in the moment. I mean, I don't know. I don't want. I'm not going to pretend like I'm a philosopher here. Okay. Again, I'll leave that to Gigi. But I know there are. There are many. I know there are many instances in history where, like, let's say a company or a country was like, most of their economy was just all, like, processing textiles, and then suddenly, boom, this machine comes out and 90% of the people just, boom. They lose their job. They're like, what are we going to do, man?

B (76:06)

Like, right.

A (76:07)

All the jobs are gone. Right. So it's just that sort of always. That same narrative, like, always happens again and again. And I don't think that AI, at least the way it is now is like, okay, now this time is actually different. Like, I don't think. I don't think that's where it's at right now. Maybe that's coming. But I think, like, LLMs, for example, I see that maybe, like, improving products. Like, maybe Gemini, like, works better now or something. My customer support is faster. I can get a refund faster because a human didn't look at it. Like, maybe things like that happen. But. And, And I don't know. I would also say that there's probably, like, a bubble in big tech where these people are getting ridiculous salaries.

B (76:45)

Yeah. No, it needs to be corrected regardless.

A (76:48)

Yeah. So either way, like, it could be a healthy thing. Maybe the big bubble is that people are getting paid $200,000 a year to go to meetings and make PowerPoints and just dick around, which is like, you know, half of corporate America. And, like, you know, hey, I can just make your PowerPoint in two seconds now.

B (77:04)

So I. Yeah, but I mean if, if they all get fired, that's like 12% unemployment rate and that's like Great Depression. Brutal.

A (77:12)

Yeah, to be fair, I mean it would. Yeah, but they would do something else, wouldn't they? Like they're not just.

B (77:18)

Yeah, but if they were like working for Door Dash, that's not, not the same thing. But I do agree that human, like humans usually find a way. But there's like, there is. I think it's also important to realize that we've never been in this place before. Like, I don't. The history rhymes, but we're in a very different situation. I forget who it was. You know, it was some modern philosopher king that was like the big tech royalty got, has gotten a pass up until this point because they've, they've brought with them this upper, like this upper royal class that is below them. They work at the companies or whatever, but they're making like you said, like their $600,000 salary. They're going into the office, they have a full buffet, you know, the big tech style. And so those people are all like super happy, but they're like right beneath them and they kind of get brought up with them in this world. Like we could easily see companies I think that are, you know, these trillion dollar mega giants that have like 70 employees. Like we see with Tether, we've seen it more recently with Hyper Liquid. Obviously those are both like in crypto land or whatever, Bitcoin land. But we kind of see it with Only Fans. OnlyFans is like super profitable per employee. So like, when is the first time we see a billion dollar unicorn with just one employee, two employees? Like I think that starts to happen

A (78:49)

or, or arguably openclaw didn't open claw get a billion dollars. And that's just one guy.

B (78:54)

No one knows what I mean. He got bought by OpenAI. I mean, I know OpenAI paid TBPN guys over a hundred million dollars. It leaked. So if they paid him that, then I would hope that Peter with openclaw got more than that. But it doesn't matter because that's all funny money, right? Like that's printed stock. Like it's the original shitcoin. So they don't know. I mean, what's the real valuation? We'll find that out when they eventually go public. And there's actually like a liquid market where you can trade your shares against it. But like, what's the first like real, you know, like the next. I don't know, I'm thinking like even lower tier, right? Like the next Snapchat or something. Could be like three employees. Like, one is extreme. That's you're being provocative. Like people need companionship. I think like, if you're like building a unicorn, like you probably need like 3 ride or dies around you at the very least. But they could have like a team of a hundred agents running everything. We're already at that tech level. I think, like, we're pretty close and we're not there.

A (79:58)

I think for sure that it's. That it's coming. Tether was a. Is a pretty unique different situation because it's like their capital is scaling. It's not really their productivity. But yeah, with AI, you could launch like 3040 Claude AI agents at once and build an entire code base. You could totally do that.

B (80:18)

Now, Tether's a perfect example, right? Because I think, I think the last public information that was released, because it is a private company, was like under a hundred employees. And like they're making like $25 billion profit a year. I bet you if you got rid of like 95 of those employees, they'd still make like $24 billion in profit. Like is like those. At that point, it's just like you might as well have an extra 70 people doing whatever the fuck they're doing. Um, but I bet you it's a small core that is actually generating the majority of the profit. And I think that becomes more common. Usually. Like, historically we haven't seen that in the digital era. We've seen companies that scale bigger than they've ever been able to scale before. These big tech giants, Facebook, literally, their, their addressable market is limited by birth. Like, they basically have full penetration of the human population. They just need more babies to be born. So that piece happened. But then they also brought with them 300,000 employees or whatever Facebook has that were all getting ridiculous salaries. Imagine if those 300,000 people didn't exist and it was just Mark and like, I guess like his wife was involved from early on and like three compatriots and then they, they had Facebook. You know, this is like a multi trillion dollar business.

A (81:41)

Yeah. Maybe if that happened, Facebook would actually be good and we wouldn't need an offer. You know, there was a time when Facebook was good, man. We're in university times, you know, and you need a university address to get in. It was actually a good app.

B (81:55)

The good old days. Yep, the good old days. Anyway, Sarah, I think this was a Great rip. I really enjoyed it. Um, thanks for going a little bit, uh, philosophy with me.

A (82:08)

Next time we'll need Gigi for sure.

B (82:10)

And we'll bring Gigi on next time. Anything you want to tell the freaks before we wrap?

A (82:14)

Just the main thing is if you haven't tried wisp yet, please try it. And if you have any issues or like feedback requests, anything you you want added like, I'm all ears. Please tag me, message me. I'm down to make the product as good as possible. So help me, help me do it.

B (82:34)

Love it, Freaks. I'll put all relevant links in the show notes. Let's do this again in like six months.

A (82:41)

I'm in.

B (82:41)

We'll do a recap, see where we're at. See if you're completely broken and pivoting to something else or got replaced by AI freaks.

A (82:51)

Love you all.

B (82:52)

Taking next week off because of Vegas. If you're going to be in Vegas. Hotstyle.pubkey.com We got a great live event, unlimited capacity. I expect we'll sell out in the next couple of days. So hotstyle.pubkey.com would love to meet you in person if we haven't already. And I promise you it'll be a really, a really great use of your time. Huge shout out to utxo.mobile try it. Give him feedback. Be annoying with your feedback. No questions. Too stupid. He'll just roll his eyes and not answer you. Love you all. Stamble Stack sats Peace.