Friendships, Open Source Marketing, and Event Driven Architectures
I made mistakes. Okay? I made mistakes on the Internet, and I'm paying for my mistakes. That's all. Alright.
Speaker 1:What's been going on, Dex? I feel like it's been a year since we spoke last.
Speaker 2:Well, yeah, that's true. A lot has happened. Depends how much you wanna talk about it. It seems like maybe you don't wanna talk too much about the various controversies that you've been embroiled in.
Speaker 1:I would not. I would not like to talk about
Speaker 2:it. Okay.
Speaker 1:I made mistakes. Okay? I made mistakes on the internet and I'm paying for my mistakes. That's all.
Speaker 2:You should have taken the advice that I gave Dev that one time where you never admit a mistake and just always double down.
Speaker 1:It seems like it would work. That would work The out
Speaker 2:second video is even more offensive.
Speaker 1:No. Okay. Yeah. I guess for people who listen to the podcast, there was just a little controversy. I'm not gonna go into it.
Speaker 1:A little controversy over some YouTube videos I made and I might quit YouTube forever, but that's all. No big deal. No. It's But you're man. Fun.
Speaker 1:It is hard. It's so hard to think of you I mean, you do YouTube. It's so hard to think of video ideas. I hate it. I hate that part.
Speaker 2:Yeah. I think the volume of videos you need to put out like, I definitely have a good amount of natural ideas. But I don't have enough for like two videos a week, I think. I have enough for like
Speaker 1:Yeah.
Speaker 2:One or two a month maybe. Yeah.
Speaker 1:I've been telling some like stories. I've been doing more videos that are just like, hey, here's things I learned, you know, in my years at stat muse and just in my career. And I feel like I've only got so many stories, you know. I'm gonna live more life or I'm gonna run out real fast.
Speaker 2:You guys are leaving your house and making new stories. Yeah. Exactly.
Speaker 1:It's so funny. Like, every time I consume anything now, like, we just do so little just this season of life. The kids are young. I feel like our entire life is just making sure they're good. But we just have no time for like watching stuff or consuming any kind of media.
Speaker 1:But whenever I do, like we let the boys watch part of the Lego movie the other day. My three year old still never really like sat through a movie.
Speaker 2:Yeah. You only allowed them to watch part of it or they were only interested in watching part of it?
Speaker 1:My eight year old watched the whole thing later, but the three year old just can't sit there long enough. Like, he just he gets antsy after like thirty minutes. So we let him watch Part the LEGO Movie, and I just watched some of it with him and realized, like, oh, there's so much good humor and stuff like this. And you watch it and you just get ideas for things you could it's like, if you don't consume any media, it's really hard to make interesting content, which is why nothing I make is interesting.
Speaker 2:Yeah. Your backlog of media is from like it's like ten years old, I think,
Speaker 1:Exactly. At this There's no ties to the current culture at all.
Speaker 2:Yeah. That is that is tricky.
Speaker 1:I realized with this controversy over, I had a video that I took down. It was my fault. I wanna just say that on my podcast as well. I shouldn't have singled out one company, one situation. Not fair to anybody involved.
Speaker 1:And I did, and I took the video down. But there is some oh, yeah. Yeah. So what I realized with this whole disagreement that I had with this person, it really was a lot of it, I think, generational. Like, as the person explained where they were coming from, as I thought where I'm coming from, there's just a lot of misconceptions or a lot of things that I thought about the situation wrong because I was thinking from my perspective in my age bracket and the way they view certain things were just very different being younger than me.
Speaker 1:So I realized, like, there really is something to that. We're all interacting on the Internet. We're all, like I just assume everybody's in their thirties. I don't know. But we're all of different backgrounds, different places, different ages, and it makes it hard makes it hard to do stuff together on the Internet.
Speaker 2:Yeah. It is funny though because I feel like I imagine that every person gets to a certain age where they don't really understand what, like, the kids are talking about, quote unquote. But think it, like, delays that. Somehow I like just like being on the internet as much as I am, I pick up on I don't ever use like slang but I like pick up on the words and what it means. And this always comes up with me and Liz because Liz is a much more normal person who doesn't spend all their time on the internet.
Speaker 2:But whenever like new terminology comes up or a new concept comes up, so it's like, what is that? Like, I don't get what that is. And then I I always know and I'm kind of like, is it weird that I know what 18 year olds are saying? Like,
Speaker 1:is that weird? Like, I mean, you think, like, past generations wouldn't have had that. Like, before the Internet, there's no way they were picking up on that stuff. So it's this weird it does kind of blur the lines, I guess, maybe between like the cultures aren't so different. If you're just extremely online, you have a pretty similar understanding of the world.
Speaker 2:Yeah. But I get what you're saying that because it's blurred, you feel like there isn't a difference, but there still definitely clearly is.
Speaker 1:Yeah. There is a difference. Here's a question I'll I'll pose to you. We can kind of talk about some of this stuff that I learned without directly going into specifics. How many friends do you think you have, Dax?
Speaker 1:Like in the world? Online, not online? Maybe. A handful? Okay.
Speaker 2:Yeah. Think about hanging a hand. So I have I have like old friends that I've been friends with since I was a child. So that's like a fairly large number. Not that I see you interact with them that often, but I would just kinda always consider them my friends.
Speaker 2:Yeah. If you remove that and just think about, like, my adult friends or people that I talk to day to day, probably, like, less than five.
Speaker 1:Yeah. That no. That's how I feel. I mean, I I think part part of this whole conversation was realizing that, like, I think younger people I I don't wanna generalize here or put anybody in a box, But I think they view like online relationships are much more like friendships to them than maybe we view them. Does that make sense?
Speaker 2:Yeah. Which which I kind of get because even my real life friends, a lot of how we hung out was just through like playing video games together and like talking online and that's where all of our conversations happened. And we were there was a time where we all lived in the same city so we would always interacting together. But now that we live in different cities, it still feels pretty much the same because there was a huge segment of our relationship that was online but Yeah. The origin of the friendship did begin in person.
Speaker 2:But I do I do kinda understand how you can have a pre trauma relationship online. I mean, our relationship is primarily online. Yeah. But I know other people just don't get it. Like, there was a period of time where Liz just did not understand how that worked, how I just couldn't see my friends for like weeks and months and I still felt like I was close to them.
Speaker 2:So I can see how like you take that a step further and people view relation to the people that they've never met just as strong as anything else.
Speaker 1:And and I guess, like, as I think about the distinction, like, if if we had not spent that week at re:Invent, I don't know that that really changes our friendship that much. I think, like, it's the time we spend together that makes to me friendships. If I haven't really if I've just kinda like existed on Twitter with people, I don't consider them friends. I think some of them I would love to become friends with, but like there's got there's more time involved than just like we're hanging out online in public in a public forum.
Speaker 2:Yeah. Like you need that one on one thing too where like it's totally different when you're hanging out as a group together with everyone and you're all interacting versus like kind of direct Yeah. One on one stuff. Yeah. So
Speaker 1:Yeah. That's it. That's it.
Speaker 2:Yeah. But I guess it maybe is different. These days, with the kids and their internets
Speaker 1:We're those people already. It's so funny how quickly you become the people you would have teased.
Speaker 2:Your kids have friends? Like, aren't they at the age where they have friends? Like, I don't know how that works.
Speaker 1:Yeah. We live on a street that has a bunch of kids. So I mean, neighborhood in generally is mostly young families, but our street, it's like four houses in a row, five houses in a row where everyone has kids around the same age. So they all just play in each other's back yards like Mhmm. Kind of one seamless big backyard and they're always playing out there together.
Speaker 1:So yeah. They I think like my oldest, he's eight. He's got friends just here in the neighborhood. That's basically it though. I mean, he's not been out into like, he's not in school.
Speaker 1:He's homeschooled. Yeah. So that's basically it's just our neighbors. And the three year old, I mean, he's friends with them. I don't know.
Speaker 1:Whatever that means when you're three. They're they're his friends too. He's just the littlest of the group, so it's different.
Speaker 2:Yeah. I mean, that sounds similar to my childhood, like just you just go and hang out in other people's houses, yards all day and you come home. But that also is kinda uncommon, I think, now.
Speaker 1:Oh, yeah. I mean, we this is the first time we've had it. Like, we we've lived a few places since we had our now eight year old and he's never had this before, so it's been really great. We've lived in other neighborhoods where it was more older people or more just not families. It's it's it's probably rare.
Speaker 1:I don't know. I I we love it though.
Speaker 2:Yeah. Because I know I know a lot of people now, like kids nowadays, just have a again, think just from a young age, it's just completely digital, their relationships.
Speaker 1:Yeah. I remember seeing those charts like how how little people like not little people. How teenagers how little they spend together. Like, they spend more of their time hanging out online than past generations. And it makes sense.
Speaker 1:Like, if if we were teenagers growing up right now, I'm sure we'd be in the same camp. Like you said, we did the gaming thing. I mean, that was that was online.
Speaker 2:Yeah. I mean, there's a specific image that I have in my head that I'm kinda wondering if it even exists anymore where it would be the middle of the day, I wouldn't really know what my friends are doing because we didn't have phones or anything. Yeah. And I would go on my bike around my neighborhood till I saw a pile of bikes in front of someone's house. Then I would be like, oh, that's where they are.
Speaker 2:You know, that does that concept even exist anymore?
Speaker 1:I mean, does here on this street. The kids ride their bikes around and they they I mean, it's a lot of like, whose house are they at today, which I I really like that for them. What's been going on with you? I don't wanna talk about my stuff. I just I'm done talking about it.
Speaker 2:Yeah. No. No. Well, what's been going on with me? I guess a few things.
Speaker 2:I think I shared that, that stuff yesterday about some of the progress we've made on the Internet.
Speaker 1:Yeah. SSE is blowing up. Yeah. Get in now or you're gonna get left behind everybody. Just go ahead.
Speaker 2:It was kind of funny because, last week so we basically, every Sunday we have a meeting. Everyone hates whenever I say this. We have a meeting every Sunday at 8PM. Yes. How could
Speaker 1:I get
Speaker 2:you? Get
Speaker 1:angry on On the the weekend, Dex?
Speaker 2:I'm pissed, upset. Everyone's just so mad that I'm doing this, that I hang out with two people who I consider friends on the weekend. Yeah. At thirty minute meetings, people go people go crazy. I would never.
Speaker 2:Anyway, so every week we review certain stats and the stat that we've have the most data for because it's easiest to get is the number of downloads. So what are the number of downloads ST is getting every single week? How do we do week to week? And since January, we've kind of been focused more on marketing and trying to like, get ourselves out there. And that number, the downloads, the week to week number were like, okay, it's continuing to grow, but pretty much the same rate it's always grown.
Speaker 2:So we just it kind of felt this weirdness that, okay, we clearly are being mentioned a lot more, like, every day when I open up Twitter, there's like someone that's like, oh, just try SST, like, Yeah, that's
Speaker 1:I see that.
Speaker 2:That, yeah, that's like not something we had. So anecdotally, we were feeling all these things and then, but it wasn't showing up in the metrics. So I was like, last week I was like, I think it's time for us to find a new metric. I don't like what the numbers are saying. Let's find something else.
Speaker 2:So I just was like, maybe we should look at something. Because because the thing with the download is there's just so much noise in it. There's all sorts of reasons why someone would download SSD. Like if we push out a lot of releases, that causes that to go up just because people are updating more. So there's just a lot of noise there.
Speaker 2:So so then this week, this Sunday, we were like looking at the numbers again and when we did this stupid thing where we had the user, the weekly active users and the new users Yeah. Line on the same chart. And obviously, the weekly active users way higher than new users and the new users chart just always look kind of like flat. Yep. But then we're like, okay, let's put this on its own chart and then we charted it and we're like, oh, shit.
Speaker 2:Like, this is a huge difference. And we started looking at just through the lens of like how many new people are trying out SST and they were just like there, then the data reflected everything we were seeing. As a classic be, Bezos quote where it's like, if anecdotes and data disagree is part of the data that's wrong, like you're looking at the wrong metric. Yeah. So just fine, we're like, we didn't like our numbers, so we looked at different numbers and now we're happy.
Speaker 1:Yeah.
Speaker 2:But that was pretty exciting. It's the thing I was saying that thread is it is so hard to move the new user chart and if you look at the whole history of SSC up until January, going back two years, we pretty much have never really moved that number despite trying all kinds of things. And if you look at a lot of different projects, even projects that like Astro that you're like, you know, you're very excited about, very similar things, right? They just have like a long period of time where it's fairly flat. So moving that number is very hard and it was really good to see validation that the new stuff we're trying is finally moving it and that should be a precursor to, you know, the actual download numbers going up downstream.
Speaker 2:So a big motivator for us this week and it's always important to find those pieces of motivation to keep you going.
Speaker 1:Do you think that the fact that it takes so long is it just a function of time, that graph never moving? Like, if you're just is it that chart where it's like, this is pointless and then it takes off? Like, do you feel like now you've waited long enough?
Speaker 2:Yeah. We'll see how long I mean, it's less about wait because there's a clear point where it shifts and we, like, changed our like, actually did something about you guys
Speaker 1:have been doing, like, just being a jerk on Twitter. That's been working very effectively.
Speaker 2:Yeah, exactly. So it just like
Speaker 1:There's more change of strategy than a function of time.
Speaker 2:Yeah, I think we figured out marketing, like we tried a new because before then, we were just kinda doing what everyone else does. We're like, oh, we have a new feature. We do a release and we make
Speaker 1:Yeah.
Speaker 2:A video showing how to use the new feature and then we just try to like up the quality of that. And every time we did that, we were getting these like spikes. And as we got better that the spikes were bigger, but nothing that was like sustained. It would like spike and then go back to where it was. So we're like, hey, I think we gotta try something that's totally not like what anyone else is doing.
Speaker 2:And that's always a gamble, but it seems to have worked out in this in this case.
Speaker 1:Yeah. I'm happy for you. I did I did wanna point out that you've now talked about having meetings on Sunday and you quoted Bezos in the same episode. So you will be canceled. I'm sorry.
Speaker 1:I don't like the rules.
Speaker 2:Let me get a cute deal quote in there too. When going zero to one, it's important to
Speaker 1:Yeah. Exactly. Make
Speaker 2:our features 10
Speaker 1:times better. Yeah. I've read that book, by the way. You tried to say I haven't read that book.
Speaker 2:I know.
Speaker 1:I read that book.
Speaker 2:I like how it got to you and you have to clarify it twice now.
Speaker 1:Did you offend me? It's funny.
Speaker 2:So that's one thing. So that's good, motivating. But always it feels like, okay, that's good but we have like a million things to do. That's fine. Yeah.
Speaker 2:I don't think I'd have it any other way. The second thing, this is now boring technical stuff. Yesterday, I was like, okay, let me actually dig into the whole thing with SSC and serverless and all this stuff is it's built on the concept of event driven architecture. There's a very basic question that comes up with event driven architecture. When you emit an event and a subscriber and subscriber consumes the event and it fails, how do you deal with that failure?
Speaker 2:And
Speaker 1:saw your Q's tweet.
Speaker 2:Yeah. So I asked like on Twitter, you know, how do people do it? Because I try and this is the thing that we do at SSC. Usually when you're trying to do something like with your own app, how do I deal with failures? You may be to spend an hour on it and then you'll like just pick one approach and you'll go with it.
Speaker 2:Yeah. We get to spend like a week on problems like that and then figure out what the exact best way is. That's what I'm doing right now for that and yeah, there's actually so many different ways to implement it. Like, I thought I had all the ways and we just kept discovering like new ways to do it and based on the replies to that thread, no one really had a slam dunk answer where it was like, oh, obviously, that's a way to do it. So Yeah.
Speaker 2:Yeah. It's kinda interesting because it's such a fundamental thing but it's still largely unsolved or not obvious at least.
Speaker 1:Yeah. I would love for there to be definitive answers on stuff like that because I always feel like there's so many like, where do you stop? Like, there's so many places you could throw a queue in between two things and it's like, I don't know where I'm creating just unnecessary bloat and where I'm really saving myself down the line. It's it's so hard to know early on. Yeah.
Speaker 1:So I'd love to know like a definitive guide on like a flowchart. Like, this is where you're gonna need some durability and you better just put it there.
Speaker 2:Yeah. Because because basically every time I've done that approach of just putting queues everywhere, it's actually been good. It's like something failed where I didn't expect it to fail or I pushed something that broke and it kind of was like a kind of like a cover your ass feature, like worst case ends up in the queue, you fix it, everything retries and goes back to normal. Yeah. So I feel very insecure not having that safety net but there's like seven different ways you can implement it.
Speaker 2:And the way I was doing it, I think based off the conversation I had with the people yesterday, I think I found a new way to do it, maybe better but yeah. Was trying to figure that out. It's just we're doing because we're building a new version of console, it's gonna be very event driven. So I'm like kind of rethinking just for ourselves
Speaker 1:Yeah.
Speaker 2:What's the best way to do it. And we do if we do discover something, we'll bake it in.
Speaker 1:What's the you just tweeted something about something unconventional you solved. Oh, yeah. Or you just told me maybe in a DM. How'd you create the lambda warmer or do you not wanna talk
Speaker 2:about it? No. We can talk about it.
Speaker 1:It's a secret sauce? Okay.
Speaker 2:So this is something and what we're we're gonna do, which I'm trying I'm trying to figure out how exactly to market this. So obviously, so much talk about cold starts. They've gone from mattering to not mattering to mattering again, you know, things if we go through that phase. Yeah. I do think they matter and there's all this like stuff about do we use Cloudflare workers because they don't have cold starts, do we use these lighter weight run times, but then you can't use all this stuff that you're used to using, all these libraries don't work.
Speaker 2:Yeah. I very specifically avoided saying that. I can't believe
Speaker 1:what a
Speaker 2:meme that's become. It came up again yesterday.
Speaker 1:Which Theo I don't think Theo listens to our podcast. But if you're listening, Theo, I don't get the car seat thing. I it just doesn't register with me. The car seat example is I just can't my brain is not smart enough to put together the connection. So just so you know, some feedback.
Speaker 2:I think Theo should give up because yesterday, he got into a situation where Dev had a better point than him. And when you're getting beat up by Dev, you just gotta exit the game.
Speaker 1:I love the way you treat Dev. Oh, it's so funny to me. He's like your little brother and it's the best. Dev is a person by the way. I I don't know if people listen to podcasts are like, what are you talking Development?
Speaker 1:Software development?
Speaker 2:Dave the Dev there. You know, I think when people first discovered Lambda functions, they're like, easy way to solve code source. Let's just have a cron job that pings the function to keep it alive. Which at first feels like it works but actually doesn't work because if that will keep one instance of the functional live, but the moment there's two concurrent requests, like two users on your site, the second user gets a cold start. So there hasn't really been like an easy out of the box way to do Lambda warming in a way that's cost effective.
Speaker 2:AWS did launch like provision concurrency, so you can just say, okay, this function needs to have, you know, 20 instances always up. But again, pretty expensive when you turn that on. It kind of takes away a lot of benefits of of using Lambda functions in the first place. So we're gonna add a feature. We're gonna start just the Next.
Speaker 2:Js site when we'll bring it to Astro, bring it to everything. We can just say like warm, true, maybe some configuration things. And there is a way to, like, send requests to a function to keep it alive and also control the number of instances and have that automatically react. So we basically send like these bursts of requests and spin up, you know, let's say 20 instances of the function. The next time we send that burst to keep them alive, we can actually keep track of, okay, we spun up 20 but only three serve requests, so let's like scale down that so we can only keep three or four alive.
Speaker 2:Okay, we spun up three or four but then like we had a bunch of cold starts, so let's spin that so we can kind of like reactively. Yep. It's kinda like auto scaling for your Lambda function in a way. Point being, you don't have to worry about how it's implemented under the hood, we'll improve that over time. It's more just a flag you turn on.
Speaker 2:Yep. Even if you're using Prisma in a serverless function, which I've been complaining about for years, even your Prisma Lambda function will you effectively won't have cold starts. And the thing with this is it kind of then brings into question, okay, like, do I really care about these Cloudflare workers anymore? Like, if I can just have this and it works good enough. We do the math on the cost, how the cost is compared to just like using provisioned lambda.
Speaker 2:It's like totally different. It's basically a very marginal cost to implement this. So it also raises the question, like, do I even split up my functions? Is it better just to keep one big function I more see.
Speaker 1:This is where I was gonna go with this. It's like, I I'm torn. Like, when Snapstart when they talk when that first came out is it Snapstart?
Speaker 2:Yeah. Snapstart.
Speaker 1:And they had it for Java and they're like, eventually, maybe it'll be Node. Js functions. There'll be no more idea of cold starts on your Lambda functions for some increased cost or something. It was like a mixed emotions thing for me because I feel like I was ready for some of these things that are kind of made for the old way to just die. Mhmm.
Speaker 1:Like, if Prisma is just not good in serverless, then we'll just find other ways. You know? I was kinda hoping for that future. And now it's like, you have to think about things differently. Like, well, I guess, sure.
Speaker 1:Put Express on your Lambda function and just run an Express app. I don't care. It doesn't matter anymore. Does it still matter?
Speaker 2:Sort of. So it it takes it it basically takes away the argument that you need to split up your functions for performance reasons. But there's great observability reasons to still split them up. Like if your infrastructure is aware that these requests are going to do different things, your logs are automatically split, you can set up like alerting in like a nice nice way. So I'm still pro splitting stuff up just to stay organized and let your infrastructure like give hints to your infrastructure on what things how things work.
Speaker 2:But it's just like a nice fail safe for people that do need to do mono functions, which I still do for things like tRPC. So I like like
Speaker 1:Oh, yeah.
Speaker 2:Having the option. But, yeah, it's kind of like yeah. I don't really know where there there's, like, different avenues for the future. There is, these lighter weight run times become more of a thing. If AWS doesn't support them natively, then it's kind of only available in the Cloudflare ecosystem.
Speaker 2:There's a future where Node. Js get continues to get better. Their standard library gets better. It gets better dealing situations. There's a future where you have things like bun come out and, like, some serverless provider offers solutions, like native solutions there.
Speaker 2:So yeah, it's unclear what the future path is but the reality is is there's a huge amount of people that will always use Node and continue to use Node and there's a huge amount of code bases that are on Node. So it's gonna be a long time before things shift. So if the node ecosystem can figure it out to get to a place seem like 80% as good as the other stuff, it's just gonna continue to win. Because the other stuff needs to be 10 times better for, you know, it's a getting adoption. There you go, Peter Thiel quote.
Speaker 2:And right now it is, right, with Cloudflare Workers are like basically no cold starts. That is like a monumental benefit but if it becomes like a marginal benefit, then then yeah, the current setup will continue to dominate.
Speaker 1:Yeah. So that's topic for another episode is your NPM package certification thing you're gonna make. I want you to make that so bad because Node sucks. I hate the ecosystem so much.
Speaker 2:I know. We'll we'll get there. We just gotta once we get a few things done here, I think it's something I'm gonna focus on, try to just improve the bar on. The dependencies that we have available for us, they need to all look the same, they need to all be documented the same, they need to have the same behavior. It's gonna be really controversial.
Speaker 2:I know as soon as I do it, people are gonna be like, one, who are you to do this? Yeah. Two, I disagree with, you know, the 10 things that you say are required. But I'm actually really ready for that. Like, I actually just don't care.
Speaker 1:You you live off that stuff.
Speaker 2:You just I know.
Speaker 1:You drink it for breakfast. Yeah. You just love Papua Mercy.
Speaker 2:Come at you. I'm gonna call it Dax's package. Dax's favorite package.
Speaker 1:It'll catch on eventually. Dax is a pretty cool like, d a x, that could work.
Speaker 2:I mean, there's so many products named that.
Speaker 1:So Oh, yeah. Like, Dynamo. Yeah.
Speaker 2:What's that Oprah Oprah's thing? Oprah's favorite things? You know how Oprah Winfrey, every year, she's got like
Speaker 1:Oh, like a list. Yes. Dax's favorite thing. Yes. Yeah.
Speaker 1:Okay. I like that. I see where you're going with it. Do it in print. I feel like she has, like, a magazine or something.
Speaker 1:You need to have, like, a Dax. It just, like it's like like the Michelin ratings or whatever. Like, it goes in like some big magazine, do like a write up on each package that makes the top 10 or something. That would be amazing. Please do that.
Speaker 2:Just get photography of the the authors and spend some time
Speaker 1:with Exactly. Uh-huh. Alright. Speaking of spending some time, we should probably spend some time not recording this podcast. This has been good though.
Speaker 2:Alright.
Speaker 1:It's been good. Alright. See you, Dax.
Speaker 2:See you.
