The Death of Stack Overflow

[Sponsor] Linear helps teams track issues, manage projects, collaborate on docs, and ship products - and the crazy thing is their software is actually fast and a joy to use. Learn more: https://trm.sh/linear [Description] Is Stack Overflow officially dead? The crew dives into the decline of developer Q&A, the impact of AI training deals, DNS security, and vintage coding memories. Also Teej has made a Stack Overflow simulator game for us to play. [Sources] Stackoverflow Simulator: https://duplicate.trm.sh/ [Hosts/Guests] Prime: https://thestanduppod.com/contributor/theprimeagen Teej: https://thestanduppod.com/contributor/teej-dv Trash: https://thestanduppod.com/contributor/trash Casey: https://thestanduppod.com/contributor/cmuratori [Partners] Linear: https://trm.sh/linear Sentry: https://trm.sh/sentry [Socials] x.com/thestanduppod instagram.com/thestanduppod tiktok.com/@thestanduppod facebook.com/profile.php?id=61573286173150 [Other Places to find us] Spotify: https://open.spotify.com/show/01A062kejnXFkJE01bjN5J RSS: thestanduppod.com/feed.xml Clips: @TheStandupPodClips Website: thestanduppod.com

• 0:00Speaker 00

  I'm actually ready now.

• 0:05Speaker 00

  Uh, anyways, I'm sorry. So I did want to hate, by the way, Trash. I don't know if you've ever tried these bad boys.

• 0:10Speaker 01

  It's upside down. Quaker or something?

• 0:12Speaker 00

  Yeah, a little Quaker. These are little Quaker apple cinnamon crisps.

• 0:16Speaker 02

  Okay.

• 0:17Speaker 01

  Ooh.

• 0:18Speaker 02

  Trash. You ever tried these bad boys? I know you've tried these bad boys.

• 0:22Speaker 01

  I got a bunch of those from Halloween. We actually have those, something like apple chips like that too. Different brand though. Are they kind of like, is it texture like the popcorn chips, kind of? Like the pop?

• 0:32Speaker 00

  It's like, yeah, poofed oats or something. Poofed rice.

• 0:35Speaker 01

  Dude, I will destroy those.

• 0:36Speaker 00

  Rice. Yeah. Oh, yeah. Poofed rice.

• 0:38Speaker 02

  Isn't it at all on the schedule for us to talk about, like, was it crimes? A scheme of like buying candy from kids at, at below market rates?

• 0:48Speaker 00

  Yeah.

• 0:48Speaker 02

  And like, yeah, like basically.

• 0:50Speaker 00

  Wait, what? Actually, it wasn't great. It was not great.

• 0:52Speaker 02

  Candy arbitrage, basically, he was buying, like, he was buying candy for 25 cents. It probably cost more than 25 cents to buy at the store.

• 1:00Speaker 00

  No, my kids bamboozled the crap out of me.

• 1:02Speaker 02

  I had a feeling.

• 1:04Speaker 00

  You know what my kids did?

• 1:05Speaker 02

  Okay.

• 1:06Speaker 00

  He went out and he did a bunch of, uh, work to go get candy, then went to all of his friends and be like, "I'll trade you this candy bar for 15 small pieces."

• 1:14Speaker 02

  Oh.

• 1:14Speaker 01

  Oh my goodness.

• 1:16Speaker 00

  Came back. Guess how much money I had to pay him?

• 1:19Speaker 02

  You got rocked, dude. A hundred dollars.

• 1:24Speaker 00

  Uh, 50 bucks. I had to pay him 50 bucks.

• 1:27Speaker 02

  You got rocked.

• 1:29Speaker 00

  I didn't know. I mean, and then the, the best part is, is that I, there's like nothing I could say. I was just like, you know what? Honestly, that was really smart of you.

• 1:35Speaker 02

  Yeah, I'd be like, I'm proud of you, boy.

• 1:37Speaker 00

  I'm probably, yeah, I like, hey, you got me.

• 1:39Speaker 02

  This is, this is how you're gonna get ahead in life.

• 1:42Speaker 01

  Yeah.

• 1:42Speaker 02

  Sadly, but that's what, that's the truth of the world.

• 1:45Speaker 01

  Very successful.

• 1:46Speaker 00

  He was clever.

• 1:47Speaker 02

  Yeah.

• 1:48Speaker 00

  You know, he figured out a system and then used it to its maximum potential.

• 1:53Speaker 01

  It's going places, kid.

• 1:54Speaker 00

  I know.

• 1:55Speaker 02

  He's paper clipping.

• 1:56Speaker 00

  And right afterwards, he's opening a daycare in Minnesota. It's crazy.

• 2:03Speaker 00

  We actually have multiple blockers. Uh, which one should we do? Should we do an instant buy? Cause this is the oldest.

• 2:09Speaker 02

  Sure. But what is, I, I can't even read it from my.

• 2:12Speaker 00

  I don't even know what it is. It says, "Another instant buy for Trash."

• 2:18Speaker 02

  Oh, we're seeing this in real time.

• 2:21Speaker 00

  Oh, yeah.

• 2:23Speaker 02

  Great, great, great point. Trash wears the glove. Did you buy it yet?

• 2:27Speaker 00

  Oh, my goodness.

• 2:28Speaker 01

  I don't, I don't know. I hope it's real. But yeah, this was, this was cause they released the, uh, they, they released the, uh, the touchscreen for the, uh, for the MacBook. And I was like, hmm. I bet you they're gonna have something. Someone posted that. I don't know if it's real or not. But when I saw it, I was just like, holy shit.

• 2:44Speaker 00

  I love that you said, "I hope it's real."

• 2:47Speaker 02

  Yeah, I hope it's.

• 2:48Speaker 00

  He needs it.

• 2:49Speaker 01

  It has to be real. If you're talking about that little satchel thing, what was that, what was that shoestring thing?

• 2:54Speaker 02

  Yeah, yeah, yeah.

• 2:54Speaker 00

  Nobody knows.

• 2:55Speaker 02

  Trash, don't bring that up, okay? It wasn't popular and I was wrong, okay? I think it was called the Apple Strap-On, if I'm not mistaken. That was exactly what they said, yeah.

• 3:04Speaker 00

  Yeah. Word for word.

• 3:06Speaker 02

  Uh, question. Is it, is it, uh, is it hashtag too soon to get a Michael Jackson branded one of these? Cause it feels like one glove that you put on by now, I think it's fine.

• 3:18Speaker 00

  Apparently he's coming back. Apparently people are like, actually really had autism and they actually now are saying he really didn't actually, uh, do what he was convicted of doing. There's like a lot of stuff, like people are saying, I mean.

• 3:29Speaker 02

  Michael Jackson branded touch glove.

• 3:31Speaker 00

  He's a creep, but apparently there's a whole wave of people being like, actually, it's becoming more and more evident he wasn't. I mean, I have no idea.

• 3:37Speaker 02

  I'm a boss killer.

• 3:39Speaker 00

  I'm, I know, I have no idea.

• 3:40Speaker 02

  I'm just trying to say that it might be hashtag too soon to say, "The new Michael Jackson touch glove." Uh, so I just want to make sure that we're like clear on that, that.

• 3:52Speaker 01

  It might, it might be too soon. It might be.

• 3:53Speaker 02

  All right.

• 3:54Speaker 01

  I don't know if there's ever a good time for that.

• 3:55Speaker 02

  Okay, yeah, all right. Fair enough.

• 3:59Speaker 00

  As, as a referee, I'm calling play in bounds. Okay, that was good. I liked it.

• 4:05Speaker 01

  DJ's just like, "Oh no."

• 4:07Speaker 02

  Yeah, it's like, uh, we were getting demonetized. I just wanted to say, thanks everybody for being on the podcast and I've enjoyed knowing all you online.

• 4:16Speaker 00

  All right, all right. We got another blocker. This one's from Casey and I believe that Casey has a whiteboard presentation for us on this one. Very excited. So this is called DNSSEC and CAA Records. It's migration season. Nothing like seeing tickets fly done for the winter. Look, look at that guy go. Look at that guy go. Finally a ticketing system that can keep up with my agents.

• 4:41Speaker 00

  Hey, cut the jokes. Linear did the impossible. It's helping me stay organized and that's why I'm so glad to have them as a partner of this channel. So if you want to support this channel or just get your work organized, check out Linear today. Links in the description.

• 4:54Speaker 02

  Yes, and so this is just on the previous epis, uh, not the previous, I guess it would be two ago, cause I wasn't here on the, on the previous episode.

• 5:02Speaker 00

  Mm-hmm.

• 5:02Speaker 02

  Um, yep. I was saying that like, I thought it would be a good idea if in your DNS records, you had an additional, uh, either signature or encryption key, probably signature was my guess.

• 5:14Speaker 00

  Yes.

• 5:15Speaker 02

  So you could cross validate that the, uh, that the certificate authority, uh, signed certificate that you're using with HTTPS was actually one that you had gotten yourself, right?

• 5:28Speaker 00

  Mm-hmm.

• 5:28Speaker 02

  And, uh, and I was noticing in the comments, some people were like, "I can't believe these guys don't know about DNSSEC and CAA Records." And so I'm not a web guy. It's not what I do.

• 5:43Speaker 00

  There's also like 25 records, Casey. So it's like, and they've been added over decades, so.

• 5:46Speaker 02

  But I just wanted to point out, I did know what DNSSEC was and CAA Records were, and I don't think they have anything to do with this. So what I wanted to do was just draw what I think this system looks like. And you guys are welcome to correct me. You guys all do way more web programming than I ever did. So I'm just gonna draw it out and you can tell me where I'm missing the, how these have anything to do with what I was talking about. Is that cool?

• 6:12Speaker 00

  That's great.

• 6:13Speaker 02

  And also Casey, I have to say, this will be a perfect transition into the episode. Okay, here we go. Yeah, so that's amazing.

• 6:19Speaker 00

  Yeah, this is good. We'll, we'll, we'll get rid of all that other stuff.

• 6:23Speaker 01

  Here he comes. I love this scene.

• 6:25Speaker 02

  So here is, as far as I know, how this stuff works. So you have a DNS provider, right? And for all intensive purposes, I'll say that you are talking to it, but really there's usually like 50 million levels of cash in between you two, like, it's like a distributed record thing. So it's not, you're not literally talking probably to the exact person all the time, but for all intensive purposes, this is a person out here who is like running your name server, right? You know, this is the name, whoever's running the name server for your domain, right? And then you've got, over here, here's the client, right?

• 7:07Speaker 02

  So this is, this is who I am. I'm some, you know, random person. I'm gonna talk to your site. I am the client.

• 7:13Speaker 00

  Client.

• 7:14Speaker 02

  I am client, right? And so what I'm going to do, obviously, is I need to go out here and I need to ask this DNS provider, right? Like, hey, I've got this name, like foo.com or whatever it is, and I need to know like what the IP address is that I should be sending things to, right? This, we all know how this works. It's, it's pretty straightforward, right? Now, when they introduced HTTPS, what they did was they were saying like, well, okay, you know, for obvious reasons, we probably don't want to get like the encryption keys directly from DNS, because what if there's a man in the middle, right? They could just both tell you which IP address, you know, where you should go, and also what the encryption key is.

• 7:55Speaker 02

  So instead, what we should probably do is have like a third party, right, over here, which is like a certificate authority, which can do signing, like encrypt, you know, sign, using sort of public, private key, they can do signing to say, "I am going to validate that who you're talking to is who you think you're talking to." Okay, so how does that work? Well, got another, this is the worst diagram ever, but like I said, I'm not a web programmer. It's kind of like, uh, it's kind of like Galactus. So here is the server for foo.com.

• 8:24Speaker 00

  Not far. Not far.

• 8:26Speaker 02

  Yeah. So here's the server for foo.com, right? So what happens is the server for foo.com, they generate like a private key and a public key, right? And they're gonna give the public key to you, so you can encrypt traffic that you send back, that only they can decrypt with the private key. Now, the problem is, because you don't know if someone's sitting in here, what you're trying to do is determine whether or not this public key that you've received is actually valid. Like, is it the actual public key that server, that foo.com is using?

• 8:54Speaker 00

  Yep.

• 8:54Speaker 02

  Or is it some crap that somebody stuck in the middle here and they're just gonna decrypt and re-encrypt the, the, the traffic? So that neither of you know there's a man in the middle, right? So what the certificate authority does is they, they don't, you don't talk to them. Like, the client does not go out here and ask the certificate authority for anything. All they do is they also have like a private-public key kind of pair. It's not exactly that, but it's pretty similar. And what they do is they say, "All right, we're going to sign things with our private key, such that if you were to use the public, you know, signatures that we provide, you can verify that we, in fact, have signed it, but we're not going to ever disclose this private key,

• 9:36Speaker 02

  so no one else could sign it as if they were us," right? They're gonna give that certificate to foo.com, and foo.com is going to give it to you, right? So basically, these people are never contacted, effectively, by you. You are just-

• 9:52Speaker 00

  To be fair, Casey, don't you contain on your local system a series of them? So technically, you're not contacted, but you contain a cache of them on your local system, which is why Kearl was affected by eMudra for a long time versus other programs, because they had their own cache that they did-

• 10:06Speaker 02

  Yeah, so, so, just to be clear. So, so that's what's happening here. Now, there's a thing called, I think it's usually called a CA bundle, a lot of times they'll call it, right?

• 10:16Speaker 00

  Yep.

• 10:16Speaker 02

  And what this is, is, okay, we don't actually want to track all of the private-key-public-key pairs that are being used here, where we have to keep all of these public keys in a record somewhere so that you know every signing thing that ever might have happened. So actually, what you do is you have a chain of people who sign things. You have root-public-private-key pairs, which are like the worst things you could possibly compromise. And then, like, certificate authorities downstream of that, sort of, they have like a chain of certificates, where the root signed the certificate authority one, the certificate authority signed like a sub-one, and the sub-one signed yours. And the CA bundle gives you that chain of certificates, so that you can validate all the way back to like a root signing key that this certificate authority really was who they said they were, right?

• 10:59Speaker 02

  But still, at no time are you ever talking to them. It's just this thing that comes like with your browser or Kearl or whatever, right? Right?

• 11:07Speaker 00

  Correct, yeah, yeah, yeah. That's why browsers were updated, but Kearl was not updated, or Pearl was not updated.

• 11:12Speaker 02

  So, the eMudra hack, right? As far as I understand it, again, not a web person, was just that the private keys were very weak. Because again, a public-private key is a pretty delicate thing, because it's much harder than purely private encryption, where you just have secret keys that both sides use. When you have private and public keys, it's usually based on some kind of mathematical, like, "it's hard to go one way but easy to go the other way" kind of a function. And as a result, they are attackable. Like, you can take it and, in shorter-than-brute-force time potentially, you can mount attacks on it if the, like, the numerology will say,

• 11:56Speaker 02

  that's not really what I mean, but, you know, the discrete math is not in your favor. So if this private key just happens to be 4, for example, like it was in the eMudra hack, it might be much easier to recover that just by looking at things encrypted with the public key or what the public key itself was, or whatever, right? Usually just the, whatever the public things that are being disclosed. That's not, like I said, that's not exactly what it is, but it's what the public data that's being disclosed that has to be disclosed in order for you to actually see what's going on and verify the chain as you're supposed to. That data can basically just be, you can reverse engineer, you can factor the private key out of it if it's weak. And that's what happened. So, they had like 4-

• 12:37Speaker 03

  Casey, let me, can I say one thing here too? Which was the funniest joke that I saw about this? Which was like, "You're imagining that you're eMudra here, and then they're doing this exact thing about like trying to guess what the private key is," right? And so like, "Yeah, well, it has, you know, up to 2 to the 256 bits of randomness, blah, blah, blah," whatever, right? And they're like, "Oh yeah, well, I'm just gonna start guessing it." "Yeah, go ahead, go ahead."

• 13:01Speaker 02

  Yeah.

• 13:02Speaker 03

  "One."

• 13:02Speaker 02

  Yeah.

• 13:03Speaker 03

  "Two."

• 13:04Speaker 02

  Yeah.

• 13:05Speaker 03

  "Uh."

• 13:05Speaker 02

  Yeah.

• 13:06Speaker 03

  "No." "Three." "Oh, no, they're really close."

• 13:12Speaker 02

  And then they say 5, because they're like, "Well, I know, they're gonna pick 4."

• 13:14Speaker 03

  "Yeah, it's not gonna be 4." "Oh."

• 13:17Speaker 00

  If 4 is an even number, they would surely never pick a non-prime.

• 13:21Speaker 03

  But just the sweat pouring down the guy's face on 3 was killing me.

• 13:27Speaker 02

  So anyway, I don't know, I don't know if all of that's correct, because like I said, I haven't, I don't do web programming. I know about like how the bundle stuff works, because I had to implement it one time, and I had to write an ANS1 parser, or whatever the heck that thing was. So I'm pretty sure I know how that part works, but you know, like, the rest of this stuff, I've never run a certificate authority-

• 13:44Speaker 00

  Boxes and arrows look correct. That looks like generally, generally speaking-

• 13:47Speaker 02

  So, but this is roughly the correct thing, but there's probably some details that are incorrect here that you would want to clean up if you were actually trying to learn from this. So don't learn from this, anybody. I'm just trying to explain to you roughly my vague idea of how it's working, right? So, all I was saying on the previous podcast was, I was like, "Okay, if your CA gets compromised, the problem is that now, not the CA gets compromised, sorry. If you just, you don't even have to compromise the CA. You just, you just figure out one of their private keys." The problem is now, you can generate as many certificates as you want, because you can sign something and no one will ever know that the CA didn't sign it. So what happens is now, someone can sit like a man in the middle, do exactly this process, meaning they do, they pretend to talk to you by using the private key you're actually, by using, sorry,

• 14:30Speaker 02

  the public key encryption that you're actually expecting, and then they talk to you using a completely different private-public key pair. And when you're like, "Let me validate that," the CA bundle and the certificate they send you is completely bogus. They just baited up and they signed it with, you know, they signed the last link in that chain with the completely erroneous number, right? With a completely reverse-engineered number. And so that is what you're trying to prevent. And what I was saying on the previous podcast is, I've never understood why you don't also get a signature from DNS. So, when you go and get your DNS record, that should also, excuse me, in my opinion, that should also give you back

• 15:13Speaker 02

  one of these, you know, one of these sort of attestations about what this private key is. So you should send it, you should have it signed not only by the CA, but you should also, in your DNS record, put an independently-signed, verified version that can be verified just independently, right? And the reason for that is now, you would have to have two successful man-in-the-middle attacks. One against the DNS, which is also potentially encrypted, right? So it would have to, their certificate authority would also have to have been hacked in the same way, and so would their DNS, right? Because their DNS server would also have this, right? So it would have to be like a cascading, sorry. So it'd have to be man in the middle in here and man in the middle here.

• 15:56Speaker 02

  Maybe, maybe that's an equivalent thing. So it might be that this doesn't really provide you more security, and that would be a legitimate complaint against my suggestion. But I feel like you could have designed this in a way where it actually is a pretty good version of this. So I think it would work. But anyway, I think it would add an extra layer anyway, so you'd have to compromise two things. So anyway, my complaint with the people who said like, "Oh, you clearly don't know about DNSSEC." Well, I don't see how DNSSEC has anything to do with this. DNSSEC just secures this link. As far as I know, that's all it does. It has nothing to do with any of this. So it just means that the DNS record that you got back, you're more confident that it was, that it was not tampered with.

• 16:41Speaker 02

  That's all it does. But it doesn't do anything to prevent a false CA signing, like, someone who had, who had exfiltrated one of these keys or reverse-engineered it. It doesn't do anything to prevent that. So DNSSEC wouldn't have changed the attack at all, right?

• 16:55Speaker 03

  Right, like, by design, like, because they're sort of saying like, the certificate or authority thing should be completely independent-

• 17:02Speaker 02

  Exactly.

• 17:02Speaker 03

  -from the DNS thing, right?

• 17:03Speaker 02

  Yes. So DNSSEC doesn't, doesn't add you an extra layer, like what I'm suggesting would have added. DNSSEC doesn't do anything. It's, it's good that you had it, but it's for completely different kinds of attacks, right? And in fact, if anything, the fact that you have DNSSEC makes my idea, and it's, I'm gonna say my idea, it's not my idea, it's DJ Bernstein's idea. You can literally go, he implemented this. It's called Curve CP, you can go look at it. He invented it like decades ago, right? And so like, this idea- Would have been, would be even better if you have a functioning secure link, right? So that's all good. Now, CAA records are the other thing they said I didn't know about. I don't see how they help either.

• 17:47Speaker 02

  A CAA record is not checked by the client, as far as I know, ever. Meaning, when the client goes and asks for the DNS, it doesn't look at a CNA, CAA record. And even if it did, it wouldn't help here, because the CAA record for this domain would have said Imudra. Right? It would have said, they're the people who are supposed to be signing it. And hey, it is signed by Imudra. We thought, but it turns out it was actually signed by somebody who exfiltrated four, right? Or reversed engineered it, they didn't even have to steal it, right? Now, CAA records do do something, and what they do is they're checked here, right? When foo.com asks the certificate authority to sign a new thing, it goes and asks the DNS server

• 18:31Speaker 02

  for the CAA record and says, hey, am I allowed to sign something for these guys? I just need you to tell me. And that record will come back and they'll check it and go like, oh, no, I'm not supposed to sign it. And it'll send an error back, right? So that's what a CAA record does, as far as I know. Doesn't prevent this at all, because at no time was Imudra ever not the certificate authority for this domain, right? It was potentially a certificate authority for this domain that was being signed. Or, if you were using this to sign some other domain, so let's say you exfiltrated this key, and you used it to sign some other certificate that wasn't with Imudra. Well, again, the client receives this information, right?

• 19:15Speaker 02

  The client receives the, let's just call this the fake signer here. They exfiltrate the four, or just reverse engineer it. They send you a bogus signed certificate that says, I am warranting that this actually is foo.com, right? I'm Imudra, and I warrant that. And foo.com never signed up for Imudra. They were using a completely different certificate authority, right? Well, at no time does the client go and check to see whether, it, it, like, web browsers don't check CAA records. So they don't go and look and go, wait, wait, this was signed by Imudra, and the CAA record on the DNS doesn't say that it should be, so I should fail on that, right? Now, maybe they should, but I don't think they do. So again, I don't think this helps at all.

• 20:00Speaker 02

  What I was asking for is for the client to start doing that, right? So if you think CAA solved this problem, I guess what I'm saying is, yes, they should, but they don't currently. So I'm asking for the thing where you would do that, right? But even that is not as good as DJ Bernstein's idea. Because if we were actually having an independent encryption key here, then it would work even for sites where Imudra was the legitimate certificate authority. So once you steal the four, even for people whose CAA record has Imudra in it, this hack wouldn't work, because it, this private key, public key pair here, was used to put an independent thing in here, which this CA never knows about. Like, it doesn't know how to do that, right?

• 20:45Speaker 02

  So-

• 20:46Speaker 03

  There needs to be a direct line from the server to the CA and the DNS, so that they can independently have it, you're saying?

• 20:52Speaker 02

  Well, no, it's, it's all that, all that has to happen is, your server, when it generates this pair, it gives you a little thing that you paste into your domain record, that's like-

• 21:03Speaker 03

  Right, yeah, that's what I'm saying. Yes. I was, I meant like, there's a, we have to have a line from my server that goes DNS, and doesn't talk to CA in the middle, and that goes to CA and doesn't talk to DNS.

• 21:13Speaker 02

  Then when you come here, you get that data back, you use it, and you validate both the signature from the CA and the signature from DNS. And if they don't match, you don't load that site. Now, maybe there's a reason why this doesn't work, but in general, like, I don't see why. And I don't understand the comments who are saying that these things, that they don't do anything. And even if you extended the client to use a CAA record in the way that a CA does, it still doesn't do what you want, because then the customers of Imudra are still screwed, right? The only people who that would protect are people who weren't using Imudra, who are getting hacked by fake Imudra certificates. But, but that's not as good. That's all I got.

• 21:55Speaker 03

  Like how people were doing, they were issuing, like, the guy reverse engineered his Samsung Smart TV stuff, because he said that he was Samsung. But then, like, maybe, maybe that wouldn't have been possible if Samsung wasn't using Imudra. Oh, sorry.

• 22:13Speaker 02

  Yeah.

• 22:14Speaker 03

  Yeah.

• 22:16Speaker 02

  Does that make sense?

• 22:16Speaker 03

  That makes sense, Casey.

• 22:17Speaker 02

  So I just, I don't know, like, people in the, the comments were like, very angry that I didn't know about those things. I'm like, I do know about those things, and just, maybe I don't understand them. But as far as I know, they don't help.

• 22:28Speaker 01

  Well, they will leave a reply, for sure.

• 22:31Speaker 02

  Yeah.

• 22:31Speaker 01

  After they see this.

• 22:33Speaker 02

  Okay.

• 22:33Speaker 03

  Yeah. Which actually leads us really well into today's topic, which is awesome.

• 22:39Speaker 02

  Which is?

• 22:41Speaker 00

  Are you ready?

• 22:42Speaker 03

  Okay, see.

• 22:44Speaker 02

  Oh.

• 22:45Speaker 00

  Casey?

• 22:46Speaker 02

  Okay.

• 22:46Speaker 00

  I don't know if you know this, Casey, but there is a little website on the internet called Stack Overflow.

• 22:55Speaker 02

  Okay.

• 22:56Speaker 00

  And for a long time, me, me, myself, in 2009, 2010, I started frequenting this website. Okay, very early user.

• 23:05Speaker 02

  Okay. Like, late at, like, late at night, when no one was watching, you would go on the internet.

• 23:12Speaker 00

  I was on Windows, running my XAMPP server, programming PHP in my NetBeans editor. Okay, I was-

• 23:20Speaker 03

  Oh, Lord.

• 23:22Speaker 00

  I was living the dream life back then, okay? It had auto, it had autocomplete, it had the jump to definition, it had some good stuff in there, okay? I really enjoyed it. Now, this website right here became one of the greatest websites for budding young developers, trying to get it particularly into the web world. I'd say that that was probably where it had its strongest questions. Um, it also seemed like Superuser, which was another one of their sites, had a lot of traffic or motion, kind of helping people do sysadmin stuff on computers. So they, they also had a lot, they had math, they had writing, they had, they had all sorts of stuff. And so you could, you could go and answer all sorts of questions, or go look up questions. And there'd just be a lot of community discussion around everything on the internet.

• 24:04Speaker 00

  And this became a wildly successful and popular place over time. Casey, did you ever stop by Stack Overflow?

• 24:12Speaker 02

  Yeah, sure. I mean, a lot of times, in my experience, it's a top hit. Like, you know, if you were to search for something on Google that was sort of esoteric, like, you know, what, you know, why does the GCC compiler switch blah, like, do something, right? There would be a person asking that question on Stack Overflow. One of the problems I always had was oftentimes, the things I would search for were kind of esoteric, and there would be no answer. But-

• 24:35Speaker 00

  Yep.

• 24:35Speaker 02

  But, you know, it was the kind of site where if, if an expert had happened to look at that thing, that was very helpful, because, you know, you could look and get it.

• 24:44Speaker 00

  There was a hit rate, too. Like, that's the nice part is that you could, and then you could also ask questions, and sometimes you did get answers out. I asked a bunch of great questions. I, in fact, I have a very famous ask question, Casey, did you know about it?

• 24:57Speaker 02

  I didn't. What was the question?

• 24:59Speaker 00

  Now, unfortunately, this presentation has changed since the last time I seen it. So this is one of those, like, situations where you're giving a new presentation on a presentation you never gave. So as you can see, I was an early adopter, all right? It had 207,000 questions in a single month, Stack Overflow did. Now it has less than 1,000 new questions.

• 25:19Speaker 02

  That's a lot of questions a month.

• 25:21Speaker 00

  Yeah. People are quite curious, you know, on the internet.

• 25:24Speaker 03

  And that was before- That was before bots could ask questions, too. Think about that. That's humans typing into the browser. Beep boop, I got a question about React, and I don't know how to get the answer.

• 25:35Speaker 01

  There's a lot of them.

• 25:36Speaker 02

  200,000 of those, 200,000.

• 25:37Speaker 00

  Very popular.

• 25:38Speaker 02

  Yep.

• 25:39Speaker 00

  So it was originally launched by Jeff Atwood and Joel Spolsky. Spolsky, Spolsky? Am I saying his last name right?

• 25:45Speaker 03

  That sounds right, yeah.

• 25:47Speaker 00

  If I remember, that's, that's on Joel, or whatever, his blog post was really great. Great blog post, by the way. Or a great blogging site. So, September 15th, free Q&A site for programmers and wannabe programmers, really, you earned reputation. You could do a lot of good stuff. People could vote you up, vote you down, you can see, there's a, there's a lot of stuff. And the golden age, okay? This was in March 2014. Now, I personally wouldn't call this the golden age of Stack Overflow, but this is the most, the most amount of engagement. Yeah, for them, it was the most amount of engagement time. Today, the network spans 183 sites, that's a lot. It has 24.2 million questions. That's pretty wild.

• 26:26Speaker 03

  Casey, that's like, they have math.stackexchange and like-

• 26:31Speaker 02

  Wow, alright.

• 26:32Speaker 00

  That's what I'm saying, when they have all these super user, Christianity, Islam, I mean, they have, if they're, they even, I believe they have a specific J or token sub, sub stack? I'm not really sure.

• 26:44Speaker 03

  Stack Exchange.

• 26:46Speaker 00

  Stack Exchange, next change.

• 26:47Speaker 03

  It's like Reddit, but across 50 different sites, you know, 183 sites.

• 26:51Speaker 02

  What exactly is Tom Bombadil? And it's like, 27 answers arguing with each other, and yeah.

• 26:56Speaker 00

  I actually have Reddit, would you like the quick TLDR on that one? Tom Bombadil represented a time in which was before monotheism, he was still part of the polytheistic world that was fading away, in which he was a god of his own little kingdom, versus the rest of the world going towards a monotheistic viewpoint of God and evil.

• 27:17Speaker 02

  And is that your own answer, or is that from Stack Overflow?

• 27:20Speaker 00

  That was from a Stack Overflow thing that I read about, okay? So, like I said, I did, I did ask a couple Tom Bombadil questions out there, because I too was confused by Mr. Bombadil. And that was the best answer I heard. I also heard it repeated by a very smart guy, and he was talking about it, about like, the nature of deism and order of the rings. Anyways, okay, lots of users, as you can see, lots of questions, lots of answers. The business, essentially, learning tool for imposter syndrome, right? $1.8 billion acquisition. This was big, okay? This was, this is not during the AI period either. Like, this is during just meat and potatoes, hands-on keyboard kind of programming era. That's a big acquisition.

• 28:01Speaker 00

  2018, Team Enterprise product, Microsoft, obviously love it, seems like a Microsoft thing. Process by Stack Overflow for $1.8 billion in 2021.

• 28:11Speaker 01

  I have no idea.

• 28:11Speaker 00

  One of those private equities, I think it's in Europe, if I'm not mistaken.

• 28:14Speaker 02

  Okay.

• 28:15Speaker 00

  Okay, you get it. I mean, private equities buy these companies that are doing whatever and they just try to milk them and walk away, right?

• 28:22Speaker 02

  Well, so in that case, we don't really care. The founders got out. It's just some private equity firm stuck holding the bag. I don't care. Alright, I'm out. Done.

• 28:31Speaker 00

  I would just like to say, hold your tongue. We're gonna get there. I assume, I assume I didn't make this presentation. One would just simply assume we're gonna get there. 2022, jobs and developer story section is shut down. Things are kind of going downhill, right? 2023 is when Jipity really gets, really gets thriving, obviously. And so, why would you ever use Stack Overflow when you can just ask something that automatically gets you the answer towards what you're, you know, doing? So, writing was already on the wall. Things are going down. Stack Overflow attempts to ban Chat Jipity. Moderators are super upset. They're going on strike. They're trying to, you know, there's, there's, it gets even, it gets wild, right?

• 29:10Speaker 00

  People are not happy with this whole Chat Jipity. They start laying off people. Not looking good. Writing's on the wall. Um, there was, I think there was one more.

• 29:20Speaker 03

  No, no, this is, no, that's good. And then this is, I had some that I wanted to interject before we get to the current state of Stack Overflow. I thought this might be a fine time for us to, you know, reminisce about some of our favorite questions, answers, things that affected us. I know Trash was on Stack Overflow. I know it. So I want to hear a story from Trash.

• 29:40Speaker 00

  Yep.

• 29:40Speaker 03

  Um, I put this one in because this is one of my favorite answers on Stack Overflow, because it's not really about Vim. That's the trick about this question. This question changed the way that I looked at a lot of stuff about programming. Right? So this question is in response to someone asking like, what the heck? Why does this thing do this? Does this thing do this in Vim? It doesn't really matter. But the main point of this answer is kind of a famous, uh, famous question, is that what you're not understanding is you're not asking the question one deeper than this. People are trying to memorize Vim shortcuts by memorizing a thousand end leaf nodes, right? The end leaf nodes and not understanding further up the stack to understand the building blocks or the language that they're interacting with.

• 30:24Speaker 03

  Which this changed the way that I thought a lot about, a lot of different programming topics as well, about figuring out what are the smaller building blocks? That I can discover that when I compose them, lead to the thousand leaf nodes, right? But I don't actually have to memorize all of those. So this one, I just wanted this one in here because it's one of my all-time faves.

• 30:41Speaker 02

  Which seems like it would pay dividends as well, because like, I feel like if you're just memorizing the leaves of things, it's probably gonna take you months to exit Vim or VI, right? Whereas if you know the building blocks, you could learn to exit the editor in a couple of weeks.

• 30:58Speaker 00

  Easy.

• 30:59Speaker 03

  Max.

• 30:59Speaker 02

  Max.

• 31:00Speaker 03

  Maximum, yeah, at least.

• 31:02Speaker 02

  Yeah.

• 31:02Speaker 03

  Which is really, that's an improvement, like tenfold.

• 31:04Speaker 00

  If we're talking like an Euler level candidate, I'd say less than a week.

• 31:07Speaker 02

  Yeah, I mean, we're talking a high IQ. High IQ.

• 31:10Speaker 00

  President Trump, Euler, you know, all high IQ individuals could do this.

• 31:15Speaker 03

  The highest IQs, they're surprised.

• 31:16Speaker 00

  Highest IQs.

• 31:18Speaker 02

  I've been hearing a lot of good things about their IQs. They even exited.

• 31:21Speaker 03

  Yeah, yeah, mm-hmm, yep.

• 31:23Speaker 00

  Trash, do you have any, uh, stories you'd like to?

• 31:26Speaker 01

  Yeah, um, I tried to become Stack Overflow famous at one point. Not famous, but I wanted, not famous, but I wanted to get my reputation really, really high.

• 31:38Speaker 02

  Okay.

• 31:39Speaker 01

  Because I, it was, it was actually like a lever to get a job, right? So like, if you, if you had a good profile with a high reputation, depending on your niche, you could get a job. So I, I think, what year is this? I want to say 2000.

• 31:54Speaker 00

  This had to be 12.

• 31:55Speaker 01

  13, 12. This is like, I was doing, I was doing like database stuff mainly and.

• 32:01Speaker 00

  I literally did the same thing, Trash.

• 32:03Speaker 01

  Yeah, I don't even think I knew JavaScript at that point. But anyways, um, I was on there and I would, I would spend all day just use, using the key terms, looking for all the niche questions that I potentially knew. Because I was just doing a bunch of weird stuff and I was just stacking those points. I was being like, super friendly, you know, I was like, I was like a Reddit moderator almost, right? I was just on there, being like, this person might know, because I was, I was in these really micro communities where you kind of just know everybody within it, because there's not that many people. Um, so like, my mindset was like, okay, let me like, network with all these people that are in the same kind of interests as me, that I could potentially use as a job maybe later if I've helped them.

• 32:41Speaker 01

  Or, and then also let me just get my points up, because you get that reputation. And I always wanted just like a four-figure reputation, just to, one, because like, I thought it kind of validated my knowledge. But two, was this like, also just a lever to get a good job. And then three, just like networking in general. So I was like nerding out on Stack Overflow, like, even not work eight hours a day. Or if I figured out something that I've been trying for a long time, I would see if anyone else was like, struggling with the same thing. And I try to answer it, or I basically put my own question. I think you, I don't, can't remember if you can answer your own question, but I remember I would put really niche questions up there. And then try to answer them, either with the same account or a different account maybe, just so like, either I can refer to it or someone else will like, you know, will likely hit it if they're doing the same stuff I was doing.

• 33:25Speaker 01

  So I was doing that for like a year or two years and it was like, it was pretty fascinating.

• 33:28Speaker 02

  Did it work?

• 33:29Speaker 03

  Did you, did it succeed? Did you succeed, Trash? Did you get the reputation you were looking for?

• 33:34Speaker 01

  Uh, I think I got to like 800 or something. I never actually used it. I never actually used it to get a job, like specifically that aspect. But like, I obviously knew my stuff. So like, if I was, when I applied for a job in that domain, I was, I was fine.

• 33:47Speaker 02

  What is, sorry, for those of us who don't know Stack Overflow, well, what is 800, like, relative to, like, what's the hot, what's, what were you aiming for if 800 isn't enough?

• 33:57Speaker 01

  Uh, I think I just wanted like a thousand at the minimum. And so every time you ask a question, you get like a reputation, I think just, what, correct me if I'm wrong.

• 34:05Speaker 02

  It's gotta be just one.

• 34:06Speaker 01

  It's just, I think it's just one upvote, it's just one. So you don't get like two or three or four. So it's like a thousand different interactions of people like, uploading your answer. If you ask a good question, you get a lot of upvotes. Like if you ask a really popular question, like how to exit Vim, right? How many people are gonna look up that question? You'll get a lot of reputation from that. But I tried to get my reputation specifically, uh, from just answering questions. So.

• 34:28Speaker 02

  Okay.

• 34:30Speaker 03

  But so Casey, I mean, you mostly interacted with it as in like, it just occasionally was something that popped up in a search result, not like, like you never asked a question on there yourself or anything, right?

• 34:43Speaker 02

  Have I ever asked a question on Stack Overflow?

• 34:49Speaker 02

  I can't remember. Maybe once? I'm not sure.

• 34:52Speaker 01

  It's actually really annoying to ask one.

• 34:54Speaker 03

  It is, actually. It's, I've never, I've never asked one because it was too scary.

• 34:59Speaker 01

  Yeah. The bar's really high. They're like, are you sure it's not already on there? Double check. It better not be on there, because if it's on there, you're in trouble.

• 35:07Speaker 03

  So, Trash, just wait for the end of the episode.

• 35:10Speaker 02

  Okay. It doesn't seem like I have a Stack Overflow account, so I guess I must not have asked one.

• 35:15Speaker 01

  I can't remember my, my account. I just tried to log in before this.

• 35:20Speaker 02

  For some reason, I was thinking maybe I did, but maybe I just am remembering reading a question that I had that someone else had already, usually someone else has already asked it. So, you know, usually you don't have to, which is one of the nice, one of the cool things about the site is even weird questions I was asking. Um, you know, sometimes no one answered, like a lot of times that was a problem, but they were asked and it's just like, no one knew. And the same, for the same reason, I'm searching for it, no one, you know, it was some undocumented Microsoft thing, right? Those kinds of things. It's like, sometimes someone does know, because they were a security researcher and they were a engineer at it, or they, you know, had worked on it internally and it's okay to release, or I don't know. So it doesn't hurt to search for those sometimes. It just, sometimes you just find a person asking and no one answers.

• 36:02Speaker 03

  So I wanted to bring this out, because I thought this settled a very definitive debate online. Um, so what we see here, Emacs, 38,000 users, 28,000 questions, right? These are the different sites that are available. So there's like Emacs.stackexchange.com or whatever it is, right? And there's VI.stackexchange.com. What we see here is Emacs, 38,000 users, 28,000 questions. And what we have with Vim is 50,000, 53,000 users and 15,000 questions. Now, what does this tell us?

• 36:34Speaker 02

  Nothing.

• 36:35Speaker 03

  What does this tell us? Trash, any guesses? No, Casey, it tells us something very important.

• 36:38Speaker 02

  No, it doesn't.

• 36:39Speaker 01

  Uh, that Vim users don't need to ask questions.

• 36:40Speaker 03

  Casey, please let me have this W.

• 36:42Speaker 02

  Alright.

• 36:43Speaker 03

  No.

• 36:43Speaker 01

  The Vim users are smarter.

• 36:44Speaker 03

  What we found out is that Vim is more popular and less confusing than Emacs.

• 36:50Speaker 01

  There you go.

• 36:50Speaker 02

  That's what we found.

• 36:51Speaker 01

  Or they're smarter. Or they're just smarter.

• 36:54Speaker 02

  I gotta disagree with you here.

• 36:56Speaker 03

  Oh, no, okay.

• 36:57Speaker 02

  Here's the thing, is, you, yeah, of course there are gonna be more questions about Emacs than Vim, because Vim is a text editor. Emacs, like, the questions in there are like, my toast is not coming out as dark as I would like it to be. What are the correct commands for this? Like, oh, you have to go to meta-x toast mode, then you have to select which, which smart toaster you're using from the, you know, like, like, so the Emacs questions are like, things that, about stuff VI users would never even think about. Because it's obviously not supposed to be in a text editor, whereas Emacs, someone wrote a package for that. I don't know why, but they did.

• 37:35Speaker 01

  Never used Emacs. Never used it.

• 37:37Speaker 03

  Alright, I guess you got me on that one, Casey. That's fine. Oh, hey, Brian.

• 37:40Speaker 00

  Okay, Trash, I just want to, just, just really quickly. Trash, I got, uh, 4,700.

• 37:45Speaker 01

  Oh.

• 37:46Speaker 00

  Rap, five golds.

• 37:48Speaker 01

  Yeah. What do the golds mean?

• 37:51Speaker 00

  That just means I got a really hard one, okay? It's super hard to unlock. Look at this. Look at this question I figured out how to ask.

• 37:59Speaker 00

  Yeah, see, that's how you get some, that's how you get some points, okay? You ask my thought questions.

• 38:03Speaker 02

  Okay. Wait, wait, wait, wait. So you're telling me Trash only had 800 and you had over 4,000?

• 38:09Speaker 03

  Absolutely, Mog.

• 38:11Speaker 01

  I just got, let's, let's cut the show. Cut the show.

• 38:13Speaker 02

  Trash, I'm sorry, man, okay.

• 38:16Speaker 03

  He tried for a whole year and Prime absolutely.

• 38:18Speaker 01

  Hold on, I see Falconer on there. Okay.

• 38:20Speaker 02

  Oh, man.

• 38:22Speaker 01

  Get out of here. RXJS, get out of here.

• 38:26Speaker 00

  Alright, so it turns out, uh, Casey, that I have a little bit of a habit of asking questions that are difficult to answer.

• 38:35Speaker 02

  Cool.

• 38:36Speaker 00

  Alright, um, and so I'd just like to go to this web archives link, because it turns out that this question was so hard to answer that, that Stack Overflow decided that they wanted to contact me. So let me first read the question.

• 38:49Speaker 00

  The best way to measure your dict in bytes. A friend and I were having a boys' night out and we got on to the topic of compression and unique ways of measuring data structures. One of my problems at work is managing a very large LRU. This is actually true. This is during the Falcor days. We are frequently running out of memory and our estimation, our estimates are all seem way off. I thought it would be straightforward to measure dicks, but it's been complicated. Let me explain the approaches. Number one, I could simply measure my dict using JSON stringification, but this has its drawbacks. It does not always get the proper length of my dict, often under, uh, under measured. I worry that my lack of measurement is due to UTF eight. I also feel like string, uh, stringified dick performance would be some optimal.

• 39:27Speaker 02

  Yeah, well, actually, can I mention something here just for people who don't, uh, who might not know? So, uh, in the old days, right? Of course, oftentimes you wouldn't necessarily call a byte a byte because you use the term word in computer memory a lot of times.

• 39:41Speaker 00

  True.

• 39:42Speaker 02

  So if you, if it happened to have like 16 bit words, you would call it like a short word or S word. And so what you're talking about here where two people are comparing their dicks, uh, in terms of S words, we call that a S word fight. Uh, that's, that's what that was called in the old days. So I just wanted to add that context to people. Yeah.

• 40:00Speaker 00

  Yep. And so this is Casey, this is really where I think I got in trouble.

• 40:04Speaker 02

  Okay.

• 40:05Speaker 00

  Um, the second option would be to pickle dict it. Uh, for those unfamiliar with pickling, here's a resource pickling. I am unfamiliar with the performance of a pickled dict, but it seems to have a firm base of support. Protocol version four added in Python three, four add support for very large dicks. This seems to be, uh, seems to fix the potential UTF eight encoding issues easily by measuring the dicks outputs directly. Protocol V five discusses out of band data. So there's a potential here for being able to take a dict as it comes, no matter how much of it there is. Uh, there is a fundamental problem. You can't rely on first glance memory usage of your dict. The nature of memory allocation and expansion, uh, and contraction means it can grow substantially when you start messing with it.

• 40:46Speaker 02

  Yep, yep. There's a couple of things in there too. Like if you are using compression, which you kind of mentioned like earlier on, right?

• 40:52Speaker 00

  Yeah.

• 40:53Speaker 02

  There, there's sort of like a, um, uh, we typically call it the, the code of compression or COC. Uh, and you know, those are the things they typically work in like a rotate, like kind of like a circular buffer. So it's surely called a COC ring. So if you're using like a COC ring for this, uh, it gets even, it gets even harder to measure actually.

• 41:11Speaker 00

  Yeah, I've heard about, I've heard about that. I've never, I've never tried, but it sounds exciting.

• 41:15Speaker 02

  Yeah.

• 41:16Speaker 00

  All right. So potential measurement ideas. I could take the approach of dict measuring head on by considering the runtime implications of Python. I also feel this to be the most correct. I don't want to consider pointers and their weights to be zero. The previous two methods, uh, really only considered direct dick measurements as opposed to indirect heap management. A helping hand. Does anyone have any idea how to measure pointers or potential pointer counts within my dict? I know dicks are very complex and I don't want to miss a drop of memory. I was thinking of simply taking the sub length plus four for each key and value.

• 41:48Speaker 02

  That's beautiful.

• 41:49Speaker 00

  Anyway, so it turns out that Stack Overflow loved my question so much, they sent me an email about it.

• 41:54Speaker 02

  Awesome.

• 41:55Speaker 03

  They, they, they care about you individually.

• 41:57Speaker 02

  They were fans.

• 41:58Speaker 00

  Yeah. Hello, we're writing, uh, in reference to your Stack Overflow account. Your question, the best way to measure, uh, your dict in bytes was deleted as rude and abusive because it appears to have deliberately crafted to have multiple sex-based double entendres. How thoroughly you wove the double entendres into your questions was impressive, but it also made it clear you were doing so intentional.

• 42:21Speaker 00

  So I actually love that they, someone wrote this out.

• 42:24Speaker 02

  Acknowledged it.

• 42:24Speaker 01

  They like someone. That was an AI response.

• 42:27Speaker 03

  They did.

• 42:27Speaker 00

  That was not an AI response. Someone actually.

• 42:30Speaker 02

  It was in the before times.

• 42:32Speaker 00

  Anyways, so that's my little story of Stack Overflow.

• 42:35Speaker 01

  That was amazing.

• 42:36Speaker 02

  It's a really good.

• 42:37Speaker 01

  Back in the vintage coding days.

• 42:39Speaker 00

  Yeah.

• 42:39Speaker 02

  Yep.

• 42:40Speaker 00

  All right. Sorry for interrupting the game because I think we, I assumed we, we went on and we started, uh.

• 42:45Speaker 02

  No, we haven't started the game yet. I haven't, I haven't started that yet.

• 42:48Speaker 03

  There's a game?

• 42:48Speaker 02

  We have a, I have a little game for us at the end. Yeah.

• 42:51Speaker 03

  Okay.

• 42:51Speaker 02

  What, Prime, will you want to hit the rest of the slides?

• 42:53Speaker 00

  Oh, okay, sure. I don't even know where we're at.

• 42:57Speaker 03

  We just finished Casey, uh, correcting me that Vim is still more confusing than Emacs because it should have had much less scope.

• 43:05Speaker 00

  Okay. And you know, honestly, Casey, fair.

• 43:07Speaker 02

  Yeah. I used to use Emacs and it's like, I don't like, an Emacs user is, is like knows 5% of Emacs. It's, there's a huge ocean of who knows what the hell is in there.

• 43:20Speaker 00

  Because I do love this.

• 43:21Speaker 03

  Go ahead.

• 43:21Speaker 00

  I am a big fan of how this actually goes because in 2024, stacked over, Stack Overflow, remember, was owned by a private equity and private equity. What are they going to do? They're going to get their money out is what they're going to do. Okay.

• 43:32Speaker 02

  Yeah.

• 43:33Speaker 00

  And so they started selling the data. Your questions and answers are now training AI. Like that's what they're doing.

• 43:38Speaker 02

  Unsurprisingly.

• 43:39Speaker 00

  So in February 2024, Google signs overflow API deal. 2024, OpenAI signs a deal. Uh, users delete their answers in protest, get suspended.

• 43:50Speaker 02

  Huh.

• 43:50Speaker 00

  You know? I mean, technically Stack Overflow owns your data. Like you gave it to them. This is just like, that's how these things work, right? All content is licensed under CC BY SA. I don't know what that means.

• 44:03Speaker 02

  It doesn't matter.

• 44:04Speaker 00

  Okay, it doesn't really matter. So now, uh, it's a ghost town. New questions down 99.5% from peak, less than a thousand a month. So March 2014, at the heyday, the peak, the golden age of Stack Overflow. 207,000. September 2025, 5,000. September 2026, 966.

• 44:26Speaker 01

  That's crazy. I'm actually curious.

• 44:28Speaker 02

  That is crazy.

• 44:28Speaker 01

  What kind of questions are actually being asked on there right now?

• 44:31Speaker 03

  We can go.

• 44:31Speaker 01

  Can we look, can we look at the latest question? I'm just curious, like who's using?

• 44:34Speaker 03

  Stack Overflow and sort by new prime.

• 44:36Speaker 02

  Please tell me it's how do I measure my digits in bytes?

• 44:41Speaker 00

  Here we go.

• 44:41Speaker 03

  No, probably it will be though again.

• 44:43Speaker 00

  NF, uh, NF tables, uh, outgoing connections blocked through no country specific IP set involved. Okay. Uh, got a negative one here. Best way to attach one primary image per row to paginate SKU listing.

• 44:57Speaker 03

  Hmm. Well, what, why did it get negative one? Does anybody tell us why?

• 45:01Speaker 00

  Uh, I'm going to help that negative one. I'm just going to elbow that.

• 45:04Speaker 01

  Let's, let's, let's grimy.

• 45:08Speaker 02

  Irresponsible, irresponsible.

• 45:11Speaker 00

  Uh, you know what? Honestly, don't like how this guy is, uh, I don't like, I don't like this guy's tone of voice. I'm going to edit his own question now really quickly because I have the power of editing his question.

• 45:20Speaker 01

  Wait, what?

• 45:20Speaker 00

  Yeah. So like, I could just go down here and I could just be like, hey, you know what? Honestly, I love boobs. Like I did just, and like, that's there.

• 45:30Speaker 01

  Like good old Wikipedia days.

• 45:32Speaker 00

  I'm not going to do that. I'm not going to do that because that's honestly, that's, that's rude.

• 45:36Speaker 02

  All right.

• 45:36Speaker 01

  Man, you have power.

• 45:38Speaker 00

  I'm going to upvote the guy just to make him feel better.

• 45:40Speaker 02

  He's back to zero. He's back to zero. That's good. Be the change you want to see in the world from negative one to zero. That's a famous book, Negative One to Zero, very, very popular.

• 45:51Speaker 00

  Let's see. So, uh, Process has written down $1.5 billion, 87% of its price. Uh, Stack Overflow is now valued at roughly $230 million and now first operating profit ever, plus $1 million, $129 million revenue for fiscal year 2026.

• 46:08Speaker 02

  On licensing, I guess, or?

• 46:10Speaker 00

  One would assume licensing and also if you didn't notice, uh, when I went there and went to questions, there's also, there was also like, look at that, I just hit this guy.

• 46:17Speaker 02

  Okay, yep, yep.

• 46:18Speaker 00

  Get another guy.

• 46:19Speaker 02

  Licensing the questions and they have some, uh, enterprise stuff where they can, you can like, they will run a Stack Overflow.

• 46:27Speaker 01

  Oh, yeah, those local things. Yeah.

• 46:30Speaker 02

  Yep.

• 46:30Speaker 00

  Mm hmm. Honestly, pre-AI, really useful.

• 46:33Speaker 01

  Yeah, we did that.

• 46:34Speaker 03

  Well, maybe now it's good too, because your agents can ask questions inside of there and they can talk to each other and answer each other's questions.

• 46:40Speaker 00

  We can also now just like, hey, agents, here's all the repos. Give me the answer, right? Like, so it's a, you know, it's, it's a little different world these days, but, uh, they're making money. Okay. Stack Overflow, private Q and A for companies, obviously.

• 46:53Speaker 03

  Right.

• 46:53Speaker 00

  Uh, bots, dude, lots of money in the bots world.

• 46:57Speaker 03

  That's what I'm saying.

• 46:58Speaker 00

  Yeah.

• 46:58Speaker 03

  Bots for bots.

• 46:59Speaker 00

  The reports of Stack Overflow's death are greatly exaggerated, at least as far as money goes.

• 47:04Speaker 02

  For people, it's cooked.

• 47:06Speaker 00

  Yeah. Uh, Stack Overflow blog 2026. Finn.

• 47:12Speaker 02

  Love it. So, uh, presentation.

• 47:14Speaker 03

  Trash.

• 47:15Speaker 02

  Uh, did a great job narrating. Love the dict. It's great.

• 47:20Speaker 03

  You're welcome, Casey.

• 47:20Speaker 02

  In the, in the Discord trash, I just sent you a link since you're the most, um, try hard Stack Overflow person. If you could do this and share your screen, I made a little game.

• 47:31Speaker 00

  Yeah, I do, I do feel like Trash needs to share a screen for this one. This does feel like a trash situation.

• 47:37Speaker 01

  I don't know how I feel about that. Hold on.

• 47:38Speaker 02

  Yeah, yeah. You're kind of, you're kind of springing this on him here.

• 47:42Speaker 03

  It's okay. It's okay. He has all the experience he needs. So this is called Closed as Duplicate. If you're familiar with Neil dot Fun's password game, where there's a million different rules to make a password. Casey, you ever seen this?

• 47:55Speaker 02

  Yes.

• 47:55Speaker 00

  Oh, I wouldn't even have told him this.

• 47:58Speaker 03

  That's okay. It's obvious from the start. Uh, because I made it available for other people to try. I made a site where you attempt to ask a Stack Overflow style question.

• 48:08Speaker 02

  Okay.

• 48:09Speaker 03

  And then it, and you have to ask the right kind of question with enough things going correctly for it to work.

• 48:16Speaker 01

  Am I on the right place?

• 48:17Speaker 03

  Font size maybe a little.

• 48:18Speaker 01

  Yeah.

• 48:19Speaker 03

  Yep, that's it. Closed as duplicate. That's the site. Go ahead and log in Trash. It'll just ask you to log in with a, you can log in with a Gmail or something. It's fine. This is not a fish.

• 48:27Speaker 01

  Okay.

• 48:28Speaker 02

  Yeah, the first rule.

• 48:29Speaker 01

  I'm going to uncheck real quick so they don't see my email.

• 48:31Speaker 02

  Okay, first rule of cybersecurity Trash is if someone tells you it's not a fish, it is absolutely a fish. Okay? I don't know what they do at Netflix, so I'm just going to be your friend here and just say, never, ever listen to someone who says this is not a fish.

• 48:46Speaker 01

  Yeah. I was very, I was stressed out immediately.

• 48:50Speaker 03

  Let me tell you something. If I was trying to fish Trash.

• 48:52Speaker 01

  I can't sign it by the way.

• 48:53Speaker 03

  It would be username and password only.

• 48:56Speaker 01

  Look.

• 48:56Speaker 02

  One password.

• 48:57Speaker 03

  Oh, did I not make this? Oh, I didn't make this available for everybody. I only tried on my account. Okay, I'll play it then Trash, you can guide me.

• 49:03Speaker 01

  Okay.

• 49:04Speaker 03

  I'll have this up so that when people, other people, um, other people can play. Or Prime, why don't you do it? Because I've already done it. Well, Casey, I have to have a login for it because it does spend tokens to answer people's questions and stuff. So I have to, I, I.

• 49:19Speaker 02

  Yep.

• 49:20Speaker 01

  Oh, it's an AI game.

• 49:22Speaker 03

  Well, there's a little bit of that. There's a decision model. There's AI stuff and there's, um, some other rules. So go ahead, Prime.

• 49:28Speaker 00

  All right, and ask a question. Yeah, dude, I'm good about this. Okay. Um, react on, let's see, uh, affect usage, um.

• 49:40Speaker 00

  When is React effect usage

• 49:42Speaker 01

  Use effect?

• 49:43Speaker 00

  Considered a good idea?

• 49:45Speaker 01

  Use effect?

• 49:47Speaker 00

  Oh, use effect, that's what it's called. Sorry, it's been a long time since I've done it.

• 49:51Speaker 03

  Yup, yup.

• 49:52Speaker 00

  I keep, uh, see, I know this would be closed as opinion, by the way, or closed as, like, I already know this would be closed on Stack Overflow, but I'm gonna do it anyways. I keep seeing, uh, people talk about how bad use effect is. Why is it bad? And there, uh, see, there has to be a reason why it's good. What can I use that is, so, let's see, uh, what can I do with use effect that is considered the best?

• 50:20Speaker 03

  Okay, guys, we gotta have an apostrophe in that "it's" because we were making fun of people earlier.

• 50:24Speaker 00

  Um, I didn't. I tried to look for possessive things. Uh, JavaScript.

• 50:31Speaker 01

  React.

• 50:32Speaker 00

  There we go. Oh.

• 50:35Speaker 00

  On hold. Put on hold as poorly tagged.

• 50:38Speaker 01

  Tag regular 9000?

• 50:40Speaker 00

  Okay, okay, okay. Separated by commas, okay. TypeScript, React, there we go.

• 50:46Speaker 01

  Front end.

• 50:49Speaker 00

  Poorly tagged, okay.

• 50:50Speaker 01

  In five.

• 50:51Speaker 03

  Prime, there's comments, dude, they're telling you. They're telling you. Have you never been on Stack Overflow? Look it, in the comments, it's telling you. Go a little lower.

• 50:57Speaker 01

  Right down.

• 50:58Speaker 03

  In the replies.

• 50:58Speaker 00

  You have three tags. The number of tags needs five. Okay, we can do five.

• 51:02Speaker 01

  It's always been five. Front end. Put "front end."

• 51:06Speaker 00

  Effect, "front end." Okay, there we go. Save that.

• 51:10Speaker 03

  Wait, okay, yeah, that's fine, that's good.

• 51:11Speaker 00

  Oh my gosh.

• 51:12Speaker 03

  Yup, okay.

• 51:12Speaker 00

  Oh, 31 people already appreciate, oh my gosh, I'm going down. Oh my gosh. Please show, uh, please show your attempt. We're happy to fix your code, not write it for you.

• 51:21Speaker 01

  Oh, dude, that's classic.

• 51:23Speaker 00

  Okay, that is classic.

• 51:25Speaker 01

  Classic.

• 51:26Speaker 00

  Okay, cons, "component equals..."

• 51:29Speaker 01

  I don't even know how to write this anymore.

• 51:31Speaker 00

  Don't worry, I got you, Trash. I got you, Trash, um...

• 51:35Speaker 01

  I haven't seen that in forever.

• 51:39Speaker 00

  Uh, "effect from, uh, React." Don't worry, I got you, Trash.

• 51:42Speaker 03

  Look at this guy. Look at this guy.

• 51:44Speaker 00

  Uh, "use effect..."

• 51:47Speaker 00

  Uh, "set interval..."

• 51:48Speaker 03

  Fetch something.

• 51:49Speaker 00

  Yeah, dude, I'm trying.

• 51:50Speaker 03

  That's good, too. Yeah, that's good, that's good. Yup.

• 51:53Speaker 00

  Alright, I'm doing it, I'm doing it, okay. Uh, "console log..." Hello. Or, uh, I'll call it "update server." Yeah, see, that's good.

• 52:03Speaker 03

  Ooh, nice, nice.

• 52:04Speaker 00

  Yeah, see, that's like, that's real, you know what I mean?

• 52:05Speaker 03

  That's a classic foot gun. That's a classic foot gun right there.

• 52:08Speaker 00

  Yeah, yeah, that's good. Uh, then I'm gonna return a, uh, "clear interval..."

• 52:14Speaker 03

  Oh, there's no way that the person I see in this question knows that.

• 52:17Speaker 01

  This guy, this guy front ends for real.

• 52:20Speaker 00

  I wasn't kidding when I said I, I, I have program.

• 52:23Speaker 01

  Make sure, make sure you put the, make sure you put the props in the, in the...

• 52:28Speaker 03

  He has over 4000 on Stack Overflow, dude. He has over 4000 on Stack Overflow. He can write use effect just fine.

• 52:35Speaker 01

  I don't even, I don't even have a thousand. I got mugged so bad.

• 52:39Speaker 03

  Yeah.

• 52:39Speaker 00

  Yeah, that's why you couldn't do this, Trash, okay?

• 52:41Speaker 01

  It's true, it's true. Well, don't you need the backticks for the code formatting? Isn't it backtick, backtick, backtick?

• 52:47Speaker 03

  Let's see what happens.

• 52:48Speaker 00

  See, this feels like, um, I am not using use effect well here.

• 52:55Speaker 00

  Um, "missing the quality ways in which one can use use effect." Boom. Save edits. Okay, that is a good question. Oh my gosh, still no actual output. Please, oh my gosh, please, okay, um...

• 53:13Speaker 01

  Wait, what?

• 53:15Speaker 00

  Um...

• 53:16Speaker 03

  It's not gonna print anything.

• 53:17Speaker 00

  Okay, guys, guys, hold on. I got this, I got this, I got this, I got this. Okay, we got this. Here's, here's what we're gonna do. We're gonna go to Codex, and I'm gonna go use, uh, crap out the update. This is just the world we live in.

• 53:30Speaker 03

  Good job, Prime.

• 53:31Speaker 01

  Are you gonna make a, create a code snippet for you?

• 53:35Speaker 00

  Just hold on. Here we go. Um, and I'm gonna go like this. Codex, uh, let's see, I'm on ultrafast, and I'm just gonna say, "create a, uh, fun-looking, uh, 2018's website about, uh, phishing with some bad output HTML that looks a bit screwed up."

• 53:55Speaker 03

  Doesn't it have to use React as well?

• 53:59Speaker 01

  I don't know. We're about to find out.

• 54:01Speaker 00

  A single page so I can, uh, just take a photo of it.

• 54:08Speaker 00

  Oh, I don't know if I have photo uploading.

• 54:11Speaker 03

  I see. Now we get...

• 54:13Speaker 01

  TJ started to sweat. He's like, "Ah."

• 54:15Speaker 03

  Yeah, uh, I don't think I hooked R2 up to this bad boy for this, but that's a good idea for later. I'll add photo for when people follow this up later. Well, no, because I don't want to let people upload photos. I'm not gonna let that happen.

• 54:29Speaker 00

  There we go. Okay, good.

• 54:31Speaker 03

  Okay, good. There you go. Copy, paste that bad boy.

• 54:34Speaker 00

  Alright, here.

• 54:36Speaker 00

  Uh, I'll just copy my terminal really quickly.

• 54:39Speaker 03

  Yup, that's smart. That, that looks good.

• 54:42Speaker 00

  Yank that one. I'm just gonna close that down.

• 54:44Speaker 01

  That's a lot, that's a lot of tokens right there.

• 54:46Speaker 00

  Don't worry, uh, "test." Oh, this is our thing, "test," there we go. I'm gonna paste that in, and, uh...

• 54:55Speaker 01

  What are you doing right now?

• 54:56Speaker 00

  I just did a quick format. Don't worry, Trash.

• 54:59Speaker 03

  He wanted to get rid of the...

• 55:00Speaker 01

  You don't need a format.

• 55:00Speaker 00

  Trash, you're not a programmer, okay? You're an agent user. It's kind of obvious these days.

• 55:05Speaker 03

  He's at the prompt factory. Trash would've burned down a forest to save room, moved the leading four characters.

• 55:10Speaker 00

  TJ, there's like a maximum limit I can paste in here.

• 55:13Speaker 03

  There's a maximum limit in Stack Overflow.

• 55:16Speaker 00

  Okay, there we go.

• 55:17Speaker 01

  I've never hit it.

• 55:19Speaker 00

  Uh, just to save some space, I'll just go like that. There we go, save.

• 55:25Speaker 03

  Okay, nice.

• 55:26Speaker 00

  Uh, "49 lines of code. Please trim your code to make it..."

• 55:32Speaker 01

  That's true. That's also a very real problem.

• 55:34Speaker 03

  But it's not 49.

• 55:35Speaker 01

  That's a very real problem. Like, just literally give me the snippet. That's the problem.

• 55:40Speaker 03

  Yup, Prime did the thing, and it's not gonna work.

• 55:44Speaker 01

  You know what?

• 55:44Speaker 03

  I feel like Prime is not even...

• 55:46Speaker 01

  I think, I think those, uh, that reputation was fake.

• 55:49Speaker 03

  Ah, you think he was farming. You think he was farming.

• 55:52Speaker 00

  During the early days, okay? I was 2014 boy. I wasn't some 2018 guy still being like, what, a 34-year-old loser in his mom's basement, saying, "Come on, get the hell out of here."

• 56:01Speaker 03

  34-year-old loser in his mom's Stack Overflow?

• 56:08Speaker 03

  Just sitting around trying to measure his dick in bytes.

• 56:11Speaker 00

  Mm-hmm.

• 56:12Speaker 03

  No, can't do it. You get banned for that kind of stuff.

• 56:17Speaker 01

  I don't think that's a valid HTML comment. Isn't there a "bang" supposed to be in there?

• 56:21Speaker 03

  Oh, maybe. I would love it if it caught that.

• 56:25Speaker 01

  And it should be a closing.

• 56:25Speaker 00

  Alright, what do we get? What do we get? Uh, what can we do is, uh, let's see, "is considered the best." Best according to whom? Both of us close an opinion base. Oh.

• 56:35Speaker 01

  You called it out, too.

• 56:37Speaker 00

  Still opinion based. Oh, I do this. Okay, okay, we got this, we got this. Um...

• 56:42Speaker 03

  You're gonna get there, Prime. Eventually, a question's gonna get accepted, I'm sure.

• 56:45Speaker 00

  Why is my React use effect, uh, usage, uh, not updating my

• 56:53Speaker 00

  server stats component?

• 56:56Speaker 03

  Oh, nice.

• 56:57Speaker 00

  Got it.

• 56:58Speaker 01

  That's a good one.

• 56:58Speaker 00

  This is, okay, there's no way. I still have a problem here.

• 57:01Speaker 01

  Can you actually win this?

• 57:02Speaker 00

  You do NOT edit my title.

• 57:05Speaker 03

  Yes! I did it!

• 57:07Speaker 00

  Oh my gosh.

• 57:10Speaker 00

  Remember I showed you that feature? Okay, so I have to pause here because I did specifically tell TJ my least favorite feature of Stack Overflow is the fact that people can just edit your stuff. Like, you ask a question.

• 57:23Speaker 03

  Yeah, why is that?

• 57:25Speaker 00

  Because they're like, "Oh, this person didn't ask with, like, good enough grammar, therefore we want to make our site super good with grammar and, and stuff."

• 57:34Speaker 03

  We want all this grammar. Well, or like, there's a hidden question in the question. Like, I mean, it makes sense if what you're trying to do from Stack Overflow is like, you're curating the best question-answer combos.

• 57:45Speaker 01

  Right.

• 57:45Speaker 03

  That people don't ask the best questions. So you want to update those to be there, but it's really crappy for a website to do.

• 57:51Speaker 01

  What's the, what's the reputation you need to get that power?

• 57:54Speaker 03

  I don't remember.

• 57:56Speaker 00

  I don't know, but I can do it.

• 57:56Speaker 03

  Trash, you don't have to worry. You don't have to worry about that, Trash. Don't concern yourself with that. Don't worry about it.

• 58:03Speaker 01

  I'm gonna start grinding Stack Overflow. I want that power.

• 58:05Speaker 03

  Don't worry about it. You don't need to, you don't need to concern yourself with that.

• 58:10Speaker 00

  Or not. What, dude, it just edited my, it edited my title, and then its edit is considered incorrect. Oh, okay, that is very annoying.

• 58:18Speaker 01

  Yours is a mood. What does that even mean?

• 58:20Speaker 00

  I don't even know.

• 58:24Speaker 00

  Is in specific question.

• 58:27Speaker 01

  Name the red problem and end it with a question mark.

• 58:29Speaker 00

  Uh, does not, uh, update my site.

• 58:34Speaker 01

  It's gonna be like it's too vague. Hundred percent.

• 58:37Speaker 00

  My specific component.

• 58:40Speaker 02

  My specific.

• 58:41Speaker 03

  Measured in bytes.

• 58:45Speaker 00

  Let's see. Close this duplicate. Okay.

• 58:51Speaker 01

  The classic, the classic Stack Overflow experience.

• 58:54Speaker 00

  Yeah, this, you know, honestly, this was a pretty good Stack Overflow experience. Like this is exactly.

• 58:58Speaker 01

  That's how they gatekeep it.

• 59:01Speaker 00

  Yeah.

• 59:01Speaker 02

  I would also say that in theory, like possibly, this may be one of the few things that someone has made with AI that was properly licensed because we know they did pay for the Stack Overflow training data.

• 59:17Speaker 00

  Yeah, true.

• 59:18Speaker 02

  We don't know how much.

• 59:20Speaker 00

  Google Argon specifically to build this one.

• 59:23Speaker 02

  Yeah, exclusively Google Gemini. I asked Google Gemini and I copy and pasted the results from the AI overview out from there into my editor.

• 59:33Speaker 01

  Oh my goodness.

• 59:35Speaker 02

  Wait, are you serious?

• 59:36Speaker 03

  No, no.

• 59:37Speaker 02

  I'm like, that would be so ridiculous. That thing can't do anything.

• 59:42Speaker 03

  No, but I do.

• 59:43Speaker 02

  I'm like, how do, I wanted to know, I'm like, how did you do that? Like that thing can't even get like facts, basic facts correct.

• 59:51Speaker 03

  Prime, I desperately want to do a me versus you race of who can build something. I don't know what has to be, has to be incredibly simple. The fastest from copy pasting out only from Google AI summary.

• 1:00:05Speaker 02

  Oh man.

• 1:00:06Speaker 03

  So we're not allowed to type in the editor at all.

• 1:00:08Speaker 00

  Okay, but can we like ask follow up questions to the summary?

• 1:00:13Speaker 03

  Oh, hmm.

• 1:00:14Speaker 01

  The summary gives you code snippets, but you can just click and copy.

• 1:00:17Speaker 03

  I know, it does. That's what I'm saying. We, feels like asking questions in the follow up summary is cheating. Cause then you can just keep, you can keep prompting it. I mean, it'll be stupid, but you can keep prompting. I like, you can only type in the search box.

• 1:00:29Speaker 01

  Oh, that's gonna be hard.

• 1:00:30Speaker 00

  Let's have a competition. This seems like, this seems like a good one.

• 1:00:33Speaker 03

  Okay, that would be fun.

• 1:00:35Speaker 01

  That's really hard.

• 1:00:36Speaker 03

  Anyway, so also for the viewers out there, I'll send a few updates to this site. You'll be able to try it out. Duplicate dot term dot SH. If you also want to play Stack Overflow Simulator, um, for yourself. That'll be, that'll be available by the time.

• 1:00:48Speaker 02

  Who's paying for the tokens though?

• 1:00:50Speaker 03

  Oh, I picked, I am.

• 1:00:52Speaker 02

  Okay.

• 1:00:52Speaker 03

  But I picked, uh, the stupidest, cheapest model to do. So, so actually Casey, there's two things.

• 1:00:58Speaker 00

  I am is doing a lot of heavy lifting.

• 1:01:02Speaker 03

  Well.

• 1:01:04Speaker 03

  This one, I mean, they're not free tokens for us. So in a way.

• 1:01:07Speaker 02

  Yeah.

• 1:01:08Speaker 03

  Um, but actually there's two different kinds of models. It's actually kind of cool. There's a decision model.

• 1:01:13Speaker 02

  Right.

• 1:01:14Speaker 03

  So I don't know if you've been following this at all, Casey.

• 1:01:16Speaker 02

  Well, I know what a decision model is anyway.

• 1:01:17Speaker 03

  Was one of the first one. Yeah, yeah, right. And, uh, so that's what does a bunch, but it's, it's, this thing's all on Cloudflare. So it's clef. But, uh, so, but same thing, right? That does a huge percentage of the amount of work. Obviously some of them are just literally code. So you can just like, does it have enough tags or does it have too many tags and stuff like that? Those will all just happen for free, basically, right? And then decision model checks like a bunch of the other things, like, does this sound like a question? Uh, right. Or like, is this on topic or is it opinion based? Those are all just like decision points from a decision model. So they're very, very cheap. And then there's only a few things that we farm out to a really stupid LLM.

• 1:01:58Speaker 02

  Okay.

• 1:01:59Speaker 03

  Which is also part of the fun because the LLM is very stupid.

• 1:02:02Speaker 02

  Okay.

• 1:02:02Speaker 03

  So it comes back with very dumb things to say, which is good. Yeah.

• 1:02:08Speaker 00

  Thanks TJ.

• 1:02:09Speaker 03

  There you go. Thanks. There you go. That was trash. I'll, I'll let you know when it's available. You can try it out. I want to see what you get.

• 1:02:16Speaker 01

  That was a, that was like a pretty realistic experience though, for real. Not even kidding.

• 1:02:22Speaker 01

  Taking me back, taking me back to the way back machine.

• 1:02:25Speaker 03

  Taking you back. Mm hmm.

• 1:02:27Speaker 00

  I'm not sure how I felt about it though. I'm not, I, I, I won't say I necessarily loved it. Okay.

• 1:02:33Speaker 02

  Kind of looks like you're typing in there right now.

• 1:02:34Speaker 01

  I mean, Stack Overflow did teach you how to like articulate your questions well. So other people can actually answer them. So, you know? Learn a little something.

• 1:02:43Speaker 03

  That's it for me. That's all, that's all I had. I just thought it'd be kind of a funny, funny way to talk about, uh, Stack Overflow. Really? I just wanted to make that game.

• 1:02:51Speaker 01

  So I'll burn some tokens.

• 1:02:53Speaker 03

  Yeah, yeah, yeah.

• 1:02:53Speaker 02

  I never, I, I guess this, this is further evidence that I've never asked a question on Stack Overflow because I didn't know it was that bad.

• 1:03:01Speaker 03

  Oh, it was the worst.

• 1:03:02Speaker 00

  A TJ missed a very, very important part of the whole experience.

• 1:03:07Speaker 02

  Okay.

• 1:03:07Speaker 00

  Which is that you ask a question and it successfully gets submitted. And then someone just comments condescendingly to you.

• 1:03:19Speaker 02

  Yeah, TJ, you should probably put a thing there, like every third comment is just condescension with no feedback.

• 1:03:24Speaker 03

  Yep. Right.

• 1:03:26Speaker 00

  Yeah.

• 1:03:27Speaker 02

  Just, just whatever the LLM is you're farming out to like, write something condescending about this. That's it. That's the only problem.

• 1:03:34Speaker 01

  Pretty toxic.

• 1:03:36Speaker 03

  We didn't make it to phase two. Prime, there is actually a phase two of the site, but it's complete. It's actually hard to get to unironically. It's like actually difficult to get through the first part of asking your question.

• 1:03:46Speaker 00

  Yeah, I didn't have a good skill gap, yeah.

• 1:03:47Speaker 03

  And then, uh, you try and answer the question, your question, right? You're trying to answer it because it's like your own thing. You're trying to get an answer. Um, and there's a, it's impossible.

• 1:03:59Speaker 03

  Get downvoted, get your answer erased. Someone else comes along and answers, asks you if you have enough experience to answer this question.

• 1:04:05Speaker 01

  Oh no.

• 1:04:06Speaker 03

  Years of experience in chat GPT usage. Um, actually the best comment that the LLM left for any of them was, hmm, this looks a little too correct. We don't accept generated answers on this site.

• 1:04:22Speaker 01

  Amazing.

• 1:04:22Speaker 02

  It knows.

• 1:04:23Speaker 01

  Amazing.

• 1:04:25Speaker 00

  All right. Hey.

• 1:04:27Speaker 03

  Thanks, everybody.

• 1:04:29Speaker 00

  I forgot to do this. Can I just do this really quickly?

• 1:04:32Speaker 03

  Of course.

• 1:04:33Speaker 00

  Welcome to the standup. Today we are talking about Stack Overflow and is Stack Overflow dead? The answer is yes. We even have a super secret surprise at the end. Today with me as always is the legendary game programmer, Casey Miratori. We have with us TJ, that, that jokester, and then of course Trash. I bet you he has the best snacks today.

• 1:04:57Speaker 02

  Well, I didn't see any snacks from Trash. What happened? Did?

• 1:05:00Speaker 01

  Uh, okay. Well, let's, let's talk about this. So I was sent maple almond butter.

• 1:05:05Speaker 02

  Oh, those are good. I like those.

• 1:05:07Speaker 01

  I almost finished this bag while, while, uh, yeah, I just opened it because we just started. So. And then I got. And then I got these, uh, limoncello almonds. Like if you guys ever had limoncello, they're actually delicious. It's like, uh, like in Italy.

• 1:05:25Speaker 00

  I would like one.

• 1:05:26Speaker 01

  Yeah. So it's basically like lemon covered almonds and they're actually super delicious. I thought they were gonna be weird. And I get, I haven't opened these ones, any of these ones, wafer bites.

• 1:05:35Speaker 00

  Oh.

• 1:05:35Speaker 03

  Oh my goodness.

• 1:05:37Speaker 01

  But the limoncello almonds.

• 1:05:38Speaker 03

  Absolutely stacked today. Good job, Lindsay.

• 1:05:40Speaker 01

  Mm hmm. Stacked. Delicious.

• 1:05:44Speaker 00

  Yep. I love that Lindsay sends you treats.

• 1:05:46Speaker 01

  I know. Me too. It's great. Prime looks so bad.

• 1:05:52Speaker 01

  Prime's like, where's my shirt?

• 1:05:55Speaker 00

  That was purely a compliment, okay?

• 1:05:57Speaker 01

  Okay, okay, okay.

• 1:05:59Speaker 00

  I don't want treats.

• 1:06:00Speaker 02

  Yeah, yeah. He doesn't even like them.

• 1:06:02Speaker 00

  Yeah, gross.

• 1:06:03Speaker 01

  Loser.

• 1:06:08Speaker 02

  Onto the episode. All right, that's it. That's the end of the episode.

• 1:06:13Speaker 03

  Oh man, bye.