Casey HATES this graph

Ship with confidence. Try Sentry: https://trm.sh/sentry This week on The Standup, we start with snack addictions and somehow spiral into one of the most unhinged breakdowns of tech, startups, and internet culture yet. TJ, DV, Casey, and Trash Dev are back—debating failed Apple products, LinkedIn nonsense, terrible data visualizations, and wild AI takes. Somewhere in the chaos, we even touch on algorithms, complexity, and why most of it doesn’t mean what people think it does. Chaotic, honest, and pretty much how developers actually talk. If you’ve ever questioned the tech industry… this one’s for you.

• 0:00Speaker 00

  Yum earth, if you're out there, sponsor us for eating these, Trash. I got your face.

• 0:06Speaker 01

  You guys wanna see what my new snack addiction is?

• 0:07Speaker 02

  By the way, this is what we have on Spotify, is TJ and Trash talking about snacks.

• 0:12Speaker 00

  True.

• 0:12Speaker 02

  True.

• 0:13Speaker 00

  Trash, that's what I'm saying, bro. That's what I have. I ordered them.

• 0:17Speaker 01

  Alright, hold on, before we start, this is my new snack addiction. So, in the back corner, I have this box of these things called popcorners.

• 0:24Speaker 02

  Yeah.

• 0:25Speaker 01

  They're so good. There's this flavor called spicy queso. I ordered a box of this like...

• 0:32Speaker 01

  I ordered it in bulk of just that flavor, and they came in this morning, and I'm just gonna destroy that whole box today.

• 0:37Speaker 03

  It's good every time. Every time.

• 0:39Speaker 01

  It's so good. Spicy queso, popcorners, whatever. Amazing.

• 0:48Speaker 02

  Uh, anyways, I'm sorry. Welcome to the stand-up. Today, we have two very special topics, and as always, we have Tej, DV, we have Casey Miratori, and we have TrashDef. All of them, legendary developers from the legendary Twitter community, here to talk about the hottest and greatest parts that are happening right now in tech today. I kind of stumbled on that last part.

• 1:13Speaker 00

  Legendary intro, bro.

• 1:15Speaker 03

  Moniker inflation is good, like, pretty super good to be like, created the entire internet by themselves.

• 1:21Speaker 02

  Yep.

• 1:22Speaker 00

  Casey, can we say though? Undisputably, this is the most listened to stand-up in the entire world.

• 1:30Speaker 03

  Good point.

• 1:30Speaker 02

  Prove us wrong. This is the greatest, most listened to stand-up of all time.

• 1:36Speaker 00

  It's the only stand-up people look forward to listening to every week.

• 1:40Speaker 03

  There you go.

• 1:40Speaker 02

  Yep.

• 1:41Speaker 03

  Okay, what is our first topic?

• 1:43Speaker 00

  Great question, Casey.

• 1:44Speaker 02

  Alright, so the very first-

• 1:45Speaker 03

  Oh wait, you know what, I just realized, I actually know what the first topic is, so I can go ahead and introduce it.

• 1:50Speaker 02

  Okay, let's do it.

• 1:51Speaker 03

  The first, the first topic for today is actually that this is the six-month anniversary. Nope, it's not that prime.

• 1:58Speaker 00

  Oh, no.

• 1:59Speaker 03

  Sorry.

• 2:00Speaker 02

  Oh, no.

• 2:01Speaker 03

  This is the six-month anniversary of Tej being 100% wrong about the Apple sling for the phone. I marked it on my calendar.

• 2:15Speaker 03

  He said in six months, you're gonna see this and I said, nope, in six months, nobody in the United States will be wearing this thing. Literally, no one. And it is, in fact, no one.

• 2:28Speaker 00

  That one's wrong though, Casey, because I got tagged, someone saw it in the wild and tagged me, like a Guitar Center, uh...

• 2:35Speaker 03

  AI, complete AI fabrication, clearly.

• 2:39Speaker 00

  No, let me find it, let me find it. It was a good one, I saw it.

• 2:42Speaker 02

  For those that don't know, this is what he's talking about, it's called the Apple, uh, holy cow, this is just so weird, it's called the Apple Sling. And this thing was marketed for like 60 bucks or 70 bucks.

• 2:53Speaker 00

  Yeah, it looks so good.

• 2:54Speaker 02

  And the idea was that you'd wear your cellphone on the side. Now TJ, or Casey Trash and I laughed about it, said there ain't no way anybody's ever gonna use that. TJ, on the other hand, said SF is gonna be filled to the brim with them.

• 3:06Speaker 03

  Yes, he did.

• 3:07Speaker 00

  I don't think that's the direct quote.

• 3:09Speaker 02

  And TJ said, every man, woman, and child will have one of these slings on.

• 3:15Speaker 00

  I didn't even see one in SF. Okay, I'm visiting it in chat right now, Prime.

• 3:19Speaker 02

  Uh, okay, Twitch chat, I assume?

• 3:21Speaker 00

  Yep, Twitch chat, right now.

• 3:23Speaker 03

  So we don't, so that was it, that was the entirety of the topic, we don't need to discuss it further, I just wanted it recognized.

• 3:29Speaker 02

  Well, look at this.

• 3:30Speaker 03

  That there, no one wants this.

• 3:32Speaker 02

  That is one in the wild.

• 3:33Speaker 03

  That's the person, that is the one that they sold. They sold one of them, they probably gave it away. They probably gave it away.

• 3:42Speaker 02

  He had to take off his body to use his phone once, because when you do it on your body, it's probably a little inconvenient.

• 3:47Speaker 03

  Yeah.

• 4:28Speaker 04

  Vain creatures, their bodies have evolved over a millennia to be able to sit in unusual postures while looking at themselves on the line. This will often last for many hours, using the excuse they're waiting for code review, repressed to why they're so inactive.

• 4:42Speaker 04

  And finally, after a long day of accomplishing very little, are keyboard warriors ready for bed? A quick read and it's lights out.

• 4:50Speaker 04

  Goodnight, little coder.

• 4:54Speaker 02

  So how do I sleep so well at night? Well, I have Sentry to help me crush those bugs. And I'm not, I'm not talking about like little teen tanny South Dakota bugs that die in the winter. I'm talking about big, mean jungle bugs. And I'm not scared of any of them, by the way, but I can squash those bugs with Sear by Sentry.

• 5:14Speaker 00

  I'll admit, I was wrong. I guess, I, you guys more correctly estimated the sanity of Apple developers. I under, underestimate, or Apple buyers. I thought for sure that because it was stupid, they were gonna buy it. We're not that stupid, okay? But they're not that stupid. Maybe, can we say, can we say, maybe in a different economy, maybe in a different, I think it was just that.

• 5:40Speaker 03

  No, we can't, because Apple's whole thing is they try to sell products to people based on the cachet. Like, they're, they are, they have a brand image, and this is totally counter to it in the US anyway. Maybe in other, you know, maybe there's other cultures where this like, was kind of more of a thing that people did, and they're just, you know, it's for the, it's for those regions or something, and that's fine. But in the US, I was like, this doesn't fly. Because Apple has a, a brand image, and this is so far away from that image. It's exactly like the Vision Pro. I don't even know why the company shipped. It, it baffles my, baffles, it blows my mind, it's completely baffling, and it's something

• 6:24Speaker 03

  that I can't imagine happening if like, if, if the ghost of Steve Jobs were still haunting the headquarters, which apparently it's not.

• 6:31Speaker 00

  He is definitely not.

• 6:31Speaker 01

  I would've bought that 30 bucks.

• 6:32Speaker 03

  No, he moved on, he's somewhere else. But, so anyway, that's it. Now, the actual first topic of the stand-up.

• 6:40Speaker 02

  So, the actual first, let me, let me do a quick intro here. The actual first topic.

• 6:44Speaker 00

  Round three.

• 6:45Speaker 02

  Is gonna have to be TJ, Trash, and Casey guesses. What does LinkedIn have to say? Now, I'm gonna say a phrase. "I had to fire Tej." Now, before I show it to you. What is the LinkedIn translation of "I had to fire Tej"?

• 7:03Speaker 03

  What does this mean, LinkedIn translation?

• 7:05Speaker 02

  Yes, it's an automatic LinkedIn translator that takes my normal phrase. And transforms it into LinkedIn.

• 7:13Speaker 00

  I've got it.

• 7:13Speaker 01

  It's like a five paragraph post?

• 7:15Speaker 03

  Yeah, it's gonna be really long, like, um, this is one of the hardest decisions that we've had to make. Uh, but the, you know, the company is strong and this only puts us in a stronger position. Uh, Tej was a very valuable employee and did a lot of great things and we're really looking forward to what he does next.

• 7:32Speaker 00

  That's great. Trash?

• 7:33Speaker 01

  I was, I was thinking more of like something happens in real life, so it's like I went to the bagel store, but the bagel person ended up burning my bagel, which somehow reminded me of Tej burning down the code at work, which made me realize and reflect upon his recent performance. And now I had to hashtag lay him off. Something, something, something.

• 7:52Speaker 00

  I, I was gonna, I was going more with, Tej has been my best friend since we founded this company. We've been through hard times, we've been through good times, we've worked together, he came to my wedding, I went to his, he's the godfather of my child. Unfortunately for him, it's time to grind even harder. #Grind, #Hard.

• 8:17Speaker 00

  And, and so, to do that in the age of AI, I have an agent that says, "You are Tej and you are good enough to do exactly what he did for me, so I laid off my technical co-founder and we're shipping 10,000 lines of code a day by my course below to also ship like this."

• 8:34Speaker 03

  Wow.

• 8:35Speaker 02

  Alright, to tell you the truth, it was in fact Casey who did the good job. Today, I'm sharing one of the hardest decisions I've ever had to make as a leader.

• 8:46Speaker 02

  I've had to part ways with my co-founder Tej. Growth isn't always a straight line and sometimes the hardest part of the journey is realizing when paths need to diverge to protect a long-term vision of the company. It's about radical candor, making tough calls, and staying true to the mission. I'm incredibly grateful for the lessons during this chapter. Onward and upward. #Leadership, #FounderLife, #GrowthMindset. TJ, that's a TJ point.

• 9:09Speaker 01

  I did the hashtag, at least.

• 9:10Speaker 03

  See, I don't read LinkedIn, so I didn't know they were hashtag-obsessed. That I totally missed that.

• 9:15Speaker 00

  They love hashtags.

• 9:17Speaker 02

  They are the biggest fans ever of hashtags.

• 9:20Speaker 03

  This is why I can't be on that website, because I vomit in my mouth a little bit. Just thinking about that.

• 9:27Speaker 02

  All right, well, this was inspired due to this post, which was unironically put on here, which is, "Michael and I are separating romantically. We're not breaking up as co-founders and we're not stepping away." Dot, dot, dot. I never got the actual full thing.

• 9:36Speaker 00

  No, no!

• 9:37Speaker 02

  This actually happened.

• 9:38Speaker 00

  No, that's not real.

• 9:41Speaker 02

  That is real.

• 9:43Speaker 00

  Quick question, did that guy happen to make an app about tracking when your wife leaves the dishes in the sink? Yes!

• 9:49Speaker 01

  Oh, my gosh.

• 9:55Speaker 03

  I love the idea of, like, instead of saying I broke up or we're getting divorced, it's like, "We're announcing a change in our chief bedroom officer."

• 10:05Speaker 01

  Oh, my, the CBO?

• 10:10Speaker 03

  We are currently going into a phase where we're going to search for our new chief bedroom officer. We have a lot of promising candidates. We want to make sure we find the right match.

• 10:23Speaker 00

  I plan on interviewing a lot in the next six months.

• 10:26Speaker 03

  Yeah, yeah.

• 10:27Speaker 01

  Dude, that's the new Tinder pickup line, "Looking for a CBO."

• 10:30Speaker 03

  Yeah, "Looking for a CBO." All right, all right.

• 10:34Speaker 01

  I like that, I like that.

• 10:35Speaker 03

  I'm serving as interim CBO right now. But it's not, it's not long-term.

• 10:44Speaker 01

  Man, that interview process is crazy.

• 10:49Speaker 02

  I am avoiding saying anything.

• 10:53Speaker 00

  Smart, next topic, quick, quick.

• 10:55Speaker 02

  Alright, so we're gonna actually get to the real first topic of the stand-up. I saw somebody on the internet post this right here. It's the complexity visualizer. Now, this was posted by STEM Explorer. I, presumably a Twitter account about exploring STEM. And they-

• 11:11Speaker 00

  Like the, like the thing in the flowers?

• 11:13Speaker 02

  Yes.

• 11:14Speaker 03

  I mean, kind of at this point.

• 11:15Speaker 02

  TJ reached deep for that pun, because it was, it was okay.

• 11:20Speaker 02

  Uh, and so they gave us a complexity visualizer. Big O notation race. In which they have O of 1, O of N, O, N log N, N squared, and then of course 2 to the N. Kind of like your classic situation. Now, this reminded me immediately of when Casey had an absolute meltdown on the internet over the ball diagram. We had several videos.

• 11:42Speaker 00

  This is actually worse.

• 11:44Speaker 02

  This is actually worse.

• 11:46Speaker 00

  I didn't know it could get worse than the balls.

• 11:49Speaker 03

  Well, it's kinda, there's two different things. So, the meltdown over the ball video, I mean, obviously, like, I always criticize ball videos. Uh, because-

• 11:57Speaker 02

  Casey, classic known ball criticizer.

• 11:59Speaker 03

  Yeah, uh, when developers feel the need to show their balls off to the entire world like that. And I'm just, I'm just sitting there going like-

• 12:06Speaker 00

  That's a job for the CBO, am I right, boys?

• 12:08Speaker 03

  Yeah, exactly. Um, Chief Ball Officer. Uh, well, internet chief ball officer Ben Dickin, who at this point is now infamous for these ball diagrams. And loves it, by the way. Like, he, as far as I can tell, he is, he is leaning into the balls.

• 12:26Speaker 03

  And, and, uh, cause, you know, he posts them to get, to get engagement. Like, that's, you know, he's trying to do, uh, he's trying, he's trying to-

• 12:34Speaker 02

  Get a rise out of him.

• 12:35Speaker 03

  Get people to interact with his, uh, social media. And so, like, you know, people, people saying these are awful is just as good as people saying they're bad, I assume? I don't really know. But anyway, the problem with those ball diagrams-

• 12:49Speaker 02

  Casey, are you fundamentally opposed to men leaning into ball diagrams?

• 12:54Speaker 03

  No, what I'm saying is that he was leaning into it. I'm not making a value judgment. If a man wants to lean into their balls, or anyone's balls on social media, that's entirely up to them, in my opinion. I don't think that we should be telling them what to do or not to do. That's my opinion. That's just my opinion, and I'm sticking to it. So, anyway, my problem with the balls was that they're a bad way to visualize data, A. But the reason, but that happens a lot. So, there's a lot of people who post their balls on the internet. And it's just like, okay, this may be the thing that they were posting was actually fine. They happen to want to show you their balls instead of just putting the data in a better format. And I'm just like, could you not do that?

• 13:38Speaker 03

  Because it's a really, it's, I understand it's eye-catching and that's why you're doing it, but it's a very bad way to visualize data. And like, there's even like, you can literally go find studies people have done, where they've looked at how people like react to different ways of presenting data. And motion is like one of the worst things to use to try and get people to conceptualize the differences between things. Like, humans are horrible at comparing like the rate of movement of things as compared to just the sizes of them, right? So- That makes sense. Yeah. So, it's just a bad idea for a chart. It's only good when you're trying to bait social media engagement, which, to be fair, like I said, I think Ben Dickinson knows. So, he's kind of off the hook, in a sense, for that part of it.

• 14:23Speaker 03

  So, really, because that's what he was trying to do, and that's fine. The reason I was so grumpy about those is because the data was wrong. Like, the thing that they were- That was also true. Ben Dickinson's data is usually also wrong, right? So, he makes a repository. He invites people to contribute to the repository. The contributions are terrible, generally speaking. They're just all over the map. And then he graphs them as if they compare like language performance, but they don't at all, right? And so, it's just not like, it's bad, it's misleading, and that's why I get upset about that. The balls are kind of, you know, insult to injury, if you will. Like, oh, now I gotta look at balls while I see this bad data. It's like, you know, it just makes things worse.

• 15:07Speaker 03

  So, I wanted to clear that up first. This is something completely different. This is like, they posted a graph that literally doesn't show the thing at all.

• 15:21Speaker 02

  It is very, very true. I've been struggling trying to figure out the meaning behind it for a long time.

• 15:25Speaker 03

  Oh, Prime was so tilted when he was like, we're gonna talk about this.

• 15:28Speaker 01

  I saw the first one and I was just like, I stopped looking, like constant time is just a wave like this. I'm like, what? It should just be like a line, right? Like, what? I don't understand at all.

• 15:36Speaker 03

  I have no idea what it's even graphing. Like, I have literally no idea. None. I don't know how they, I don't know how they came up with this.

• 15:44Speaker 00

  Complexity visualized, Casey, it says right on the graph.

• 15:47Speaker 02

  It says right there, Casey. It's pretty obvious.

• 15:51Speaker 03

  Well, I mean, you know, if I, I can't see because of course, you know, this is Riverside and so being able to see the thing. I can't read, I can't read any of that.

• 16:00Speaker 01

  Yep.

• 16:01Speaker 03

  But like, what's the top one? O, O1?

• 16:04Speaker 01

  Yeah, O1, constant time.

• 16:05Speaker 03

  And what's, okay, so, so O1, there's...

• 16:08Speaker 02

  It goes real, it goes...

• 16:10Speaker 03

  What's the next one?

• 16:12Speaker 02

  Uh, and, log n.

• 16:14Speaker 03

  Log n, and the next one is n log n, maybe?

• 16:17Speaker 02

  No, it's just n?

• 16:18Speaker 01

  This is linear.

• 16:18Speaker 03

  Oh, okay. We skipped n log n, alright.

• 16:20Speaker 02

  Then n log n, no, then n log n, because it's slower, right? Theoretically, it's slower. Then n squared, and then 2 to the n. So each one is like progressively worse by an order of complexity.

• 16:31Speaker 03

  Understood, okay.

• 16:33Speaker 02

  With a couple complexities skipped.

• 16:35Speaker 01

  So the more squiggles, the worse?

• 16:37Speaker 03

  Yeah, the more squiggles, so are the squiggles...

• 16:38Speaker 00

  It's supposed to be how long it takes to move from one side to the other, but the squiggles are useless.

• 16:44Speaker 02

  Yes, correct. The squiggles actually have a meaning to it.

• 16:47Speaker 03

  What I was trying to do is count them. I'm like, do the, do the squiggles even go in a pattern that has to do with like what the n was? It doesn't even look like that. Like, I couldn't, because...

• 16:56Speaker 02

  Well, it'd be very difficult because they erase the squiggly. So if you were supposed to deduce the squiggle density, it's being erased actively.

• 17:02Speaker 03

  Yeah, that's true. Actively. Because like the top one, so if that's O1, then the constant would have to be like 3, right? Because there's 3 squiggles, so, and we know that n is 1, so, you know, the constant would have to be 3. If the next one is log n, right? We assume the constant is still 3 probab-, I mean, I guess there's no reason to think that the constant would still be 3, I suppose. But I'm just, I'm just trying to figure out, is there any way to make this line up? I suppose you could say, well, the constant changes at each time and that's where the squiggles. I, I don't know. I, I literally don't know. I can't figure it out.

• 17:37Speaker 01

  There's no way any thought went into...

• 17:38Speaker 02

  So the thing that really just grinds my gears is that constant time... Theoretically, it does not matter upon the input. Right? It's just about whatever the coefficient is next to the constant time that really dictates how long it is.

• 17:50Speaker 00

  That's the constant part in constant time.

• 17:52Speaker 02

  That's the constant part in constant. And so, the fact that it takes about half the speed as log n really bothers me. Because constant is a fixed kind of like operating time, whereas log n depends on how much input there is. And so you can't actually make any sort of relational guess between constant and log n. But in this one, they're saying it it's about 60%. That kills me on the inside. Right.

• 18:20Speaker 00

  On average, constant time is about, or log n time is about 60% as fast as constant time on average.

• 18:25Speaker 01

  60% of the time every time.

• 18:27Speaker 00

  Yeah, like on average.

• 18:28Speaker 02

  It really hurts me so, so much. And then the, then the other part that really gets me is that at the very end, when they're kind of finishing everything up, uh, 2 to the n exponential time and quadratic time. Exponential time's like 20 or 15% through and quadratic time is like 95% way through. Yep. And so it's just like that is not true even in the slightest, right? It's like at, at, at n of 10, exponential time is 10 times larger. At n of 11, exponential time's 100 times larger than n squared. At 12, it's a thousand times larger. Like, none of this makes any sense. It's just driving me just completely insane.

• 19:07Speaker 01

  This has to be AI generated. There's no way.

• 19:09Speaker 00

  I assume it is.

• 19:10Speaker 02

  Well, they wrote code to generate the AI.

• 19:13Speaker 01

  Oh, did they?

• 19:14Speaker 02

  This is not AI generated in the sense that someone came up with this idea and told AI to generate it. I am so positive that if you told AI like, hey, generate the differences between these times.

• 19:24Speaker 00

  It would have done a standard graph.

• 19:26Speaker 02

  It would have been more accurate. It would have been this.

• 19:28Speaker 00

  Yes.

• 19:29Speaker 01

  Exactly.

• 19:31Speaker 02

  Exactly.

• 19:32Speaker 00

  Yeah.

• 19:33Speaker 02

  Which this.

• 19:34Speaker 00

  I'm looking at that. I'm looking at it. And I'm like, but which one takes longer? I can't even tell because where is the ball moving across the screen?

• 19:43Speaker 02

  Right, right, right.

• 19:44Speaker 00

  I can't even tell.

• 19:46Speaker 02

  Oh my gosh, sorry. I was just so frustrated.

• 19:48Speaker 03

  But that's kind of the point, Tej, right? Like, so, one of the important things about complexity is that it doesn't tell you how long something takes. You can have an O1 algorithm that takes a lot longer than an On algorithm for some n, right?

• 20:05Speaker 02

  This is true. Classic array versus set lookup when you're doing stuff.

• 20:09Speaker 03

  This is just how it is. And so, that's the point of understanding. Like, if you try to convey to somebody that On is faster than On log n, or something like that, you're conveying incorrect information, because it's not faster or slower, it scales worse, right? There's a difference between those things. And so, you need, like, even showing anything that implies that there's a speed difference between them is wrong just right out of the gate. Because complexity is not about telling you how fast something ran. And that's critical to understand. It's so critical that actually, in shipping code, in deployment, sometimes you will use algorithms that have worse scaling complexity

• 20:54Speaker 03

  when you know what the n is, because the scaling complexity doesn't matter, because the n is low and the constant's too high on the scaling version, right? On the lower complexity. So, you can have times people deploy an n squared algorithm instead of an n log n algorithm. You can have people deploy an n log n algorithm instead of an On algorithm, and so on and so forth. That will absolutely happen, not because the person didn't know what they're doing, but because that was actually the right choice for the scale of problem they were working on.

• 21:19Speaker 02

  A quick classic example, quicksort. Often, for the last few elements, we'll switch to insertion sort, because it's actually faster to not quicksort on a bunch of small items due to the stack invocation and recursion cost, comparatively to the n squared algorithm, or standard case, average case of insertion sort.

• 21:39Speaker 03

  I didn't even know that. Uh, well, quicksort, quicksort's n squared.

• 21:42Speaker 02

  Uh, quicksort, worst case is n squared. Quicksort, average case is n log n. And so often, inside of those, also with merge sort, you'll break into an insertion sort at the very bottom, because it is just better. And so there's a classic example where scaling actually doesn't, you know, is traded off in the middle just because of size.

• 21:59Speaker 00

  Wow. I was thinking too, Prime, of like, uh, when we talked to Tiger Beetle a while ago, of how they like, all of their, like, so many places in the code, they have like, some assert about the size of how big something can be, because they're also like, loading all of it ahead of time and all this, but... So it's like, you just say, "Hey, I think that this thing should never get over 200 elements or 1,000 elements." Like, it's cool, because then you can pick something that works really good, where n is less than 1,000. And it's like, it's really fast at that thing. And then later, if that invariant breaks, you find out, because you have a debug failure, like some warning or something that happens in your system, that can then, you're like, "Okay, cool." So now, apparently, we have big enough customers that 1,000 isn't always the case. Sometimes it's like, 10 million,

• 22:44Speaker 00

  for some reason or something. We have some pathological customer that does this, so we gotta go fix that separately, which I thought was really cool.

• 22:50Speaker 02

  There's actually a popular thing that John Carmack does. John Carmack, if you don't know, does the exact same thing in Doom and a bunch of other places. He would assert, like, "The world can't have more than 3,000 items." That's just a, that's just a reality of life. We make all of our decisions based off this one thing.

• 23:05Speaker 00

  Mm-hmm.

• 23:06Speaker 03

  There's also cases where you know that there's a limiting factor on the other end. Like, for example, if you're like, "Look, this thing is designed for 10 gigabit, you know, ethernet uplinks." So, there you go. Like, I know, I know how much data is flowing through this thing, and it can't ever be more than that, because we don't have, you know, and we're, we will fundamentally redesign this part of the code if we switch to, you know, a higher bandwidth connection or something like that, right?

• 23:30Speaker 00

  Right.

• 23:30Speaker 03

  Or, this screen, like, "We know that we're shipping on this console that only supports 1080p." So that's how many pixels are on the screen. Like, we know that a particular region of the screen can only have this many pixels, so we're making decisions based on that. Like, so you can definitely have situations too, where you even just know. You're just like, "Look, for the deployment scenario, we actually guarantee there is no physical way that it can have a different thing." And at that point, you're free to make all kinds of decisions based on n, because you know what it is. But okay, so then how do I know which one's faster? No, I'm just kidding. So which one is faster? A lot of people, like, complexity is one of those things where it's like, the, it's kind of a, I guess since we were talking about diagrams and balls and all that,

• 24:15Speaker 03

  complexity theory is definitely one of the things that I feel like is a just the tip kind of thing in computer science where it's like, do you need to know complexity theory? A lot of people ask this question, right? Like, do I need to know complexity theory to be a programmer? You absolutely don't need to know complexity theory because complexity theory is like, like, okay, how do you prove something is NP-hard or something like? Like, actually knowing complexity theory? I don't know complexity theory. I don't, I'm not going to give you a dissertation on P-space or some, like, forget it, right? It's not going to happen. The bathroom. You do need to know, yeah, you do need to know just the tip. You need to know like your basic, like, okay, how do algorithms scale with n roughly?

• 24:58Speaker 03

  In terms of how much memory they use and how much, like, how many concrete operations, whatever that operation is, will they have to do? That's something everyone should know just for the basics. Like, the ones that were on this diagram, probably not exponential because almost nobody uses exponential except in situations where... So, this may be getting ahead of my, out of my, over my skis a little bit, but I would say that in general, if you're in the realm of exponential time, so you're thinking of things that are like 2 to the n complexity or worse, at that point, you may start wanting to know more complexity. Like, at that point, you're in the realm of hardcore algorithm design

• 25:42Speaker 03

  where you probably want to know more than just the basics because you're in, you're going to start needing to know things about heuristics, heuristic-based algorithms, randomized algorithms, things that are like not commonly thought about very much in, you know, in workaday programming circles. So, once you are doing something that's beyond n squared or n cubed and you're, you know, out of polynomial time, at that point, you may want to know more. But everything below that I'd I'd say is just a tip situation.

• 26:09Speaker 02

  Casey, what happened if you have a salesman, okay? And he wants to start in Rapid City, South Dakota?

• 26:15Speaker 00

  Okay, I'm listening.

• 26:16Speaker 03

  That's exactly what I was thinking of in my head when I said that. Because like, those kinds of problems require a deeper understanding of algorithm design and the ways that we attack problems that are sort of technically infeasible. Like theoretically, you can't really solve this problem efficiently. But we've come up with lots of ways to, you know, maybe not get the best answer, but get a good enough answer that can be used for humanity to do whatever it was trying to do. Can we prove that we will get the best answer every time? No, but we can get like, close, right? And you have to kind of start to know that sort of thing.

• 26:56Speaker 02

  By the way, just, I know we're getting weird, slightly off, but FedEx and UPS and USPS, they all have an actual NP problem, which of course is bin packing. Because if you have X amount of boxes and you have Y amount of containers and you need to be able to most efficiently pack these boxes, especially along a route, now you also have like, multiplexing dykstras along with bin packing. Very, very crazy. I mean, they actually have a really hard and interesting algorithm to solve.

• 27:24Speaker 03

  I mean, even just basic things like your school. What is going on, Prime? Oh my god.

• 27:33Speaker 01

  That will literally break my teeth off my mouth.

• 27:35Speaker 03

  That's true. That's true. That's a nice crust on that, Prime. It was good, it was good. I just wanted to taste it, you know? Now the heat's gone. If you just think about like, your local school district, what the bus routes are and which order they go in, that already is like, there are companies who specialize in solving that problem because you need heuristic salt. Like, you can't, if you, even just the number of students in your district being 500 or whatever it is, is enough to start to make it be like, we can't guarantee you a perfect solution. Like, sorry, like, that's just out of the bounds. So we have to do like, things that, you know, take into account that we have, you know, that we don't have to get it perfect in order for it to be good enough

• 28:20Speaker 03

  and all these other sorts of things.

• 28:21Speaker 00

  At my high school, Casey, they wrote that down on the whiteboard. They leave it overnight, see if anybody can figure it out. Any janitors walking by?

• 28:27Speaker 03

  Right, any janitors walking by?

• 28:29Speaker 00

  Wait a minute, I've seen this before.

• 28:34Speaker 03

  Yeah, so we, you know, we always look and I don't know. Last time I looked at that sort of stuff, it was kind of generally expected that, look, a lot of these kind of problems, you can get very close to optimal solutions. You just, there are certain like, they have these regions of their, of their sort of solution space that are just extraordinarily hard. And so if the input data happens to fall in those areas, you are kind of screwed. If it falls outside of those areas, then it's okay. And you'll get, you know, you'll typically get the solution you were looking for, right? So, anyway, my point in all of that was just to say, if that's the kind of thing you're doing, you should know, you should know more than me about complexity at that point.

• 29:16Speaker 03

  And you should know more than me about like, how you go about solving those kind of problems with, you know, heuristic approaches, randomized approaches, things like that. Because there's entire theory of randomized algorithms. It's very interesting, actually, but it's like, I don't know it, and a lot of programmers don't know it, because it doesn't come up that often.

• 29:33Speaker 02

  Are you ready for, are you ready for part two? So can we say goodbye to our squiggles? Our heartbeat, we can all at least agree, heartbeat diagram, terrible for complexity visualization.

• 29:44Speaker 03

  Yeah, I mean, I think what this really is, is an open challenge to Ben Dickin. It was like, look.

• 29:50Speaker 00

  True.

• 29:51Speaker 03

  Like STEM Explorer has bested you at every possible axis. Not only is the diagram even worse than the bouncing balls, but it also is literally not graphing anything even related to the subject, right? So I think, that's why, I just reposted this, and I think I said something like, "Your move, Ben Dickin," because I feel like, I feel like at this point, he has been so thoroughly dethroned, that like, he needs to come, he needs to come back and hit hard here.

• 30:23Speaker 02

  I agree.

• 30:24Speaker 03

  Because his standard ball diagram is not gonna do it. Like, he has been shown up, like, significantly.

• 30:30Speaker 00

  And you know what? It's gonna, it's gonna be sad if he comes out with something and it's not good enough to get on the pod, Ben, okay? So Ben, if you're out there listening, we need something big from you, all right? The squad here, we're looking, Q1, that's gonna be, you have to deliver by the end of Q2, but we need it in Q1 at the very, you know, that's kind of what I'm targeting.

• 30:50Speaker 02

  That's a blocker right there, Ben Dickens.

• 30:52Speaker 03

  The world is watching.

• 30:53Speaker 00

  Yeah, we are locked.

• 30:53Speaker 03

  They want to see your response to this. Because STEM Explorer came out of the gate here, first time, and just absolutely crushed it.

• 31:02Speaker 01

  I'm having a DAX moment over here.

• 31:03Speaker 02

  You are. The one thing I will say about the ball diagram, whether you hate it or not, if the data was accurate, at least there was some level of like meaning you could kind of gather from and go, oh yeah, I can see this one's slower and that one's faster. By how much and all that, you couldn't really gather. This one, I can't gather any. There's, there's actually no information here. Why, why are the squiggles erasing on?

• 31:26Speaker 01

  I was, I was, I was gonna try to find a moment to say that. It really bothers me that they end.

• 31:31Speaker 02

  They're getting vacuumed up.

• 31:33Speaker 01

  Yeah, there's no budget. There's no budget for, uh, to keep them going.

• 31:36Speaker 02

  I just really hate that they break at the end like that. Right? Like, I mean, there's, because it does it with each one of them as they get to the end. It like wigs out.

• 31:45Speaker 01

  Oh no, we've, oh no.

• 31:48Speaker 02

  That's the I'm done mark. That's like, I finished.

• 31:53Speaker 02

  All right, all right, we can move on.

• 31:56Speaker 02

  All right, let me, let me get back here just in case. We want to cut this as a separate segment. I gotta do like a better intro. There we go. You know, on the standup, we like to come together and discuss some things that very important people say that make everybody else super upset. And this is just one of those moments where what I've just read somehow is the most infuriating accomplice salt I've ever received. And I'm sure many people have ever received in their lifetime. And of course I'm talking about Mr. Sammy Jipity Altman right here saying, "I have so much gratitude to people who wrote extremely complex software, character by character. It already feels difficult to remember how much effort it really took. Thank you for getting us to this point."

• 32:44Speaker 02

  Okay, so.

• 32:45Speaker 01

  Sam, pour it up for the boys.

• 32:48Speaker 02

  Pack it up. We are done. But I do, I really do love this idea that there is no more complex software. Cause hidden in this statement is that like software complexity is now officially over.

• 33:00Speaker 01

  Solved.

• 33:02Speaker 02

  It's actually solved.

• 33:04Speaker 01

  That's why Claude Code doesn't flicker anymore, right?

• 33:07Speaker 02

  This is why GTA 6 supposedly will release this year. Like there is rumors that it's going, that it's coming out.

• 33:14Speaker 01

  Yeah.

• 33:16Speaker 03

  I mean, I don't, it's like, almost, this is similar. These two topics feel very related to me, right? Because the previous one was like, let's have a competition to see who can post the worst graph. Right? It was like, kind of what that was, right? And this, it's like, it seems like that AI people are doing like, let's have a competition to see who, like, people will dislike the most, right? Like, how do I become someone that literally no one likes at all? And it's worth noting that it's like, including your customers? Like, they don't even want to be liked by the people who are using their software, or they don't know that, like, I mean, unless they're really this unaware

• 34:01Speaker 03

  that they don't see how insulting that is to even people who are actively using their software currently. I just, it's crazy. It boggles, like, I don't even know if I have a comment on that. I don't know what to say about this. It's like, it's shocking.

• 34:14Speaker 00

  I've got a hypothesis, Casey, which is that, you know, they sort of feel like they already won. And you know how when you beat a video game, you can play it back on like, expert mode? Or like, I've been playing Slade Inspire 2, climbing my ascension rakes, right? And each time it adds like, more difficulty, right? Like, start with a curse, you don't heal all the way, etc. I feel like this is kind of like, okay, well, we already made like, a trillion dollars. Can we do it now if nobody likes us, right?

• 34:40Speaker 03

  Right, right, right.

• 34:41Speaker 00

  So they're heaping those things on. They're like, I'm gonna win biggest company in the world on Ascension 20, OpenAI Edition. I think that that's probably what...

• 34:51Speaker 03

  So you think at the beginning, like, Sam Altman, you know, for this round of starting a company, it popped up those little like, pick which card you want to start with, or whatever, right? And one of them was, get a lot of money from people, but nobody likes you. And he picked, like, that was the modifier that he chose to play the whole thing on?

• 35:11Speaker 00

  Exactly. Yes. Right. And then, like, round one was like, it had three options. It was like, start a nonprofit to save the world with AI. But then like, later, it was like, okay, here's a few forks in the road. One was like, continue the nonprofit, everyone likes you, trying to make something for the benefit of humanity. Option two was, make a ton of money!

• 35:33Speaker 01

  When I read this tweet, I had like a visualization in my head as like, as I was reading it, as him like saying this, but he was escorting like some old person off the stage. Like, kind of just giving him this backhand comment and then just kicking him off the stage. And then it was like, here's the real show kind of thing like, because this is the stuff I tweeted this too, but this is like the exact, well, at least my emotional response was. This is a message of like, when you lay somebody off, like, thank you for all your hard work at the company.

• 35:58Speaker 00

  He LinkedIn posted it, bro.

• 36:00Speaker 01

  Dude, like legit. And I was just like, man, like, when you're typing these things, if he is the one typing these things, like, does he just hit send? Like, that's a good tweet.

• 36:11Speaker 01

  We're gonna love that one. Like, how, like, what, what is that scene look like? I would love to just see like surveillance of him posting this.

• 36:18Speaker 02

  I think, I think what you're missing is that he, I think he and like Zuckerberg and a few of these people fall into what is known as the lizard category. And so I, I genuinely believe that he was just like, you know what? There was a lot of people who I've taken all of their work from and then making the biggest company in the universe ever. I need to thank those people. I'm gonna thank them right now. Here we go, boys. And like got his team together, like, okay, hold on. Oh, man, it was so complex, step by step. Oh, my gosh.

• 36:47Speaker 00

  Character, our character.

• 36:49Speaker 02

  But dude, it's like, honestly, it's hard to even remember how hard that was.

• 36:54Speaker 03

  Does he even know how hard, like, has he ever shipped anything that was like?

• 36:59Speaker 01

  Looped.

• 37:00Speaker 00

  Looped.

• 37:01Speaker 03

  What is looped?

• 37:01Speaker 01

  I don't know what that is.

• 37:02Speaker 00

  Oh, okay. Pause. I'll do, I'll do a brown bag lunch on this in a little bit, okay?

• 37:09Speaker 02

  You did brown bag world coins. I do feel like you have to brown bag world, world, uh.

• 37:13Speaker 00

  I'll brown bag looped for you later, Casey. We'll, we'll bring you up to speed on that. Make sure everyone brings their own lunch, etc. Lunch will not be provided. Just want to make that clear. Uh, but I will be providing a presentation for that. Yeah.

• 37:26Speaker 03

  All right, all right. Yeah, I don't know. Uh, this is just really bizarre to me and it doesn't.

• 37:35Speaker 03

  So I think I actually talked about Dimitri, I talked to Dimitri about something like this at one point. He was obviously on the podcast last week, talking about AI stuff. And the way he said it, I don't want to misrepresent him here, but he, I asked him this question. I was like, given that AI companies will face like a bunch of lawsuits, regulatory hurdles, things like this, like they will run into the bureaucratic elements of society, and they will have to navigate those, right? And, and I'll point out, they also will have to navigate things that are worse than that. So, you know, I don't want to say anything that will get us marked badly on YouTube, but let's just say that like,

• 38:19Speaker 03

  a data center is a pretty easy target for civil unrest. I'll put it that way, right? True. So there's a lot of things that you want, if, if you were really serious about this, you're like, I want to make a company that does this AI stuff, and I want to make sure that it succeeds and does well. There's a lot of things you would care about with respect to public opinion, and you would never say any of the things that Sam Altman says, ever, right? If you were Sam Altman, and you had any self-awareness, you would not appear in public. You would have somebody else who was your person who appears in public, and it would never be you. That's what you would do, right? Because nobody likes him, and when he opens his mouth, he says things that make people very angry. The, in terms of the general public, right? Not talking about programmers, or whatever, talking about like,

• 39:01Speaker 03

  the general public, when they hear from Sam Altman, they really don't like it. He was recently just had another one where he talked about how, if you compare the, the cost of training an AI to how much a human consumes, the AI is actually better. Like, no one should ever say, like, if you care about this, as actually getting this thing adopted, you would never say anything like that, right? Yes. Because, because you should, you should have someone out there who understands how hu, how actual humans perceive things, right? Which is not Sam Altman. That's that lizard thing I was telling you about. It's a lizard thing. So you're right about same as Zuckerberg. They just, they have fundamentally no idea what a human is like, even though they've seen them before.

• 39:43Speaker 03

  They just aren't sure what they do, right? They're like, huh? And so,

• 39:49Speaker 03

  And so, to me, I was like, why is that? Like, how does no one else step in and go like, because I mean, you know, when you're looking at a company like OpenAI with these investors putting billions of dollars in it, someone would, in theory, come along and be like, Sam, thanks for all your hard work. You're not going to be the, they're going to, they'd Palmer Lucky him, like they did at Facebook, right? They're like, you don't get to talk to the public anymore. We're, you know, we're replacing, like, that kind of thing happens all the time for reasons that are strictly about appearances or public, you know, perception.

• 40:20Speaker 02

  What is that reference? I'm sorry, I don't know the Palmer Lucky thing at Facebook.

• 40:23Speaker 03

  Palmer Lucky, well, you know, again, trying not to get into politics too much. Palmer Lucky was on the politically wrong side of an issue early on when his company Oculus was acquired by Facebook, now Meta. They weren't obviously Meta at the time because they were just acquiring Oculus. And he was basically forced out of a leadership position there. And he ended up leaving the company. And that was purely political.

• 40:49Speaker 00

  And he was the, like, to be clear, wonder kid. Front of magazines, super, like, previous.

• 40:55Speaker 02

  Yeah, I remember him being everywhere. I do remember him being everywhere.

• 40:58Speaker 00

  The father of VR, like, everything. It wasn't just like, oh, like, some random guy got fired. Like, he's, he was the wonder kid for them.

• 41:06Speaker 02

  Yeah, he was like Jimmy Neutron is like my expectation of him. He was actually like a very talented individual that was, like, way out there when it comes to talent and building stuff.

• 41:16Speaker 03

  Yeah. Well, whatever you think about him, he got forced out.

• 41:20Speaker 00

  Yeah.

• 41:21Speaker 03

  Right? Like, no one disagrees with that, I don't think. That's not a controversial thing to say. Like, he did not voluntarily decide to leave Facebook and their VR division because he just wasn't feeling it that day. He got for, like, he was forced to leave the division that he was, you know, that he had, that was his entire thing he wanted to do. Like, this was the thing that he started in his bedroom doing, right? He wanted to see if you could use phone displays and stuff like that to produce a real VR headset now was the technology ready, right? And here, he was getting the opportunity to do that, and he got forced out for political reasons. So, something as simple as that, you can force someone out. They're not forcing Sam Altman out. Right. Nobody's telling Dario just to keep his mouth closed. I wish they would.

• 42:06Speaker 03

  And so, and my question was- Those articles are pretty funny these days. My question was, why? This is what I was asking. I asked him, I was like, why? Like, what's going on there? Because surely the investors realize that it's better for this, right? And what he said was, possibly somewhat convincing, actually. What he said was, they have two choices, is what he said to me. One is they garner public support. Or the other is they garner investor support, and they went with investor support. And what he meant by that was, if you tell a story about how AI is just going to be beneficial for everybody, and that people won't be losing their jobs,

• 42:49Speaker 03

  and that it will just be like, oh, it'll be easier to get your work done, or whatever it is, whatever the publicly, we're just empowering artists, we're not taking their jobs. That can't raise the level of investment that you can from private investors if you tell them you're putting everyone out of work. So he said they had two choices. One was to go and say that we're putting everyone out of work, and we're basically going to be a pariah, because everyone's going to pay us, and no one's going to pay anyone else, and raise their trillion dollars. Or they could actually say something positive, but they wouldn't be able to raise that kind of money, because it doesn't sound like you're replacing the entire industry with one computer. And they chose the one that gets them the investment. And I was like, okay. That's, that's a pretty reasonable, that's a pretty reasonable way to say that.

• 43:34Speaker 03

  That could very well be the case.

• 43:36Speaker 01

  That makes a lot of sense.

• 43:38Speaker 03

  So you-

• 43:39Speaker 01

  I didn't even think about it that way at all.

• 43:41Speaker 03

  So you, well, this is why it's great to talk to Dmitry, right? It's why, it's why I'm doing that podcast, is because he has a lot of really interesting things to say, and he knows these people.

• 43:48Speaker 01

  Podcast.

• 43:49Speaker 03

  Yeah. I'm just kidding. He knows those people, like, he's interacted with a lot of these people personally. He knows a lot of them in that field, and so he has, like, some perspective on why they do what they do. It's like, yeah, so they could replace Sam Altman with somebody who was much more appealing to the public, and maybe they will after they secure enough trillions of dollars. Then you're like, okay, thanks, thanks Sam, right? Now let's put a nice face on this thing.

• 44:16Speaker 00

  Then they hire Matthew McConaughey.

• 44:18Speaker 03

  Exactly.

• 44:20Speaker 00

  That's who I would hire.

• 44:22Speaker 03

  Yeah.

• 44:22Speaker 00

  I like TJ. Matthew McConaughey, that's the solution to this thing. I'm telling you, you guys seen him in the Lincoln commercials? I mean, he seems like a guy who you would trust, you know? It's like, that makes sense to me.

• 44:34Speaker 01

  I like Billy Beal, I trust him.

• 44:37Speaker 03

  He could be like, I don't stress, the Matthew McConaughey could be like, I don't stress about it. I just ask ChatGPT.

• 44:42Speaker 00

  That's what I'm saying. You'd be like, he's so right, bro, he's so right, he's so right.

• 44:47Speaker 03

  Just ask ChatGPT.

• 44:50Speaker 00

  Okay, well, when Matthew McConaughey gets hired by OpenAI, you guys are gonna, you guys are gonna eat your words right here, trash, okay? I mean, I love Matthew McConaughey too, man.

• 45:00Speaker 03

  It's Matthew McConAI.

• 45:02Speaker 00

  Yes, he's gonna legally change his name.

• 45:05Speaker 03

  Like, like, uh, like, um, oh, uh, Will I Am did, right? To the dots.

• 45:11Speaker 00

  Oh, no. Great, Casey, nice pull.

• 45:15Speaker 03

  Deep pull, right? Remember when he made a smartwatch? No one does. I do.

• 45:20Speaker 00

  Uh, Shakira, these chips don't lie.

• 45:27Speaker 00

  Come on, that would be a crazy. I can't think of anything. That would be a crazy ad campaign.

• 45:32Speaker 03

  Sitting here right now, I'm actually slightly shocked that Will I Am hasn't announced an AI company. Maybe he has and just didn't make it up, but that would be so something he would do, right?

• 45:42Speaker 01

  It's black eyed peas, right?

• 45:44Speaker 03

  Originally, yes, but then he had his solo.

• 45:47Speaker 01

  Yeah, yeah, yeah.

• 45:48Speaker 03

  Career.

• 45:49Speaker 01

  Let's try to make sure I'm thinking of the same person.

• 45:51Speaker 00

  Yeah, I have no idea who that is, to tell you the truth. Okay, I need to make a.

• 45:56Speaker 03

  Prime, there's an era that you probably missed out on because, you know, maybe you're not quite as old as I am. But there was an era where companies, especially internet companies, but not exclusively, like, companies like Intel, for example. Had sort of brand ambassador positions. And these were people who were like enthusiastic about technology, but didn't really have all that much to do with it. Will.i.am was one. Shingy, does anyone remember Shingy?

• 46:30Speaker 00

  Hmm, that sounds familiar.

• 46:31Speaker 03

  Are you trying to say Chingy? Nope, Shingy. He was, he was an experimental artist who was like a brand ambassador for, I can't, I can't even remember which company, but a major company like Hewlett Packard or something. I don't remember who it was. But like, and they would produce these videos that were just like, you know, them talking about how cool technology is, but non specifically. This was the whole thing that happened. And there's more, I'm not remembering the world. There's several other celebrities who were like in that vein and you're just like, what is going on? But that was just, it was cool for a while in like the early 2000s to have like this pseudo, like it's usually B-list, like it's not, you know, I'm not trying to think of who the A-listers are. It's not Jennifer Lawrence, who is right.

• 47:16Speaker 03

  It's not A-list celebrity, someone who's like launching a major motion picture, just below like B-tier, somebody that people have heard of, but it's not, you know.

• 47:24Speaker 00

  Prime, like Prime.

• 47:26Speaker 02

  Dude, that'd be so good, dude. I'd definitely do that, dude. Hewlett Packard. I'd be like, look at this printer.

• 47:34Speaker 03

  I got all these inks.

• 47:36Speaker 02

  I got all the inks, babe.

• 47:37Speaker 03

  I got all these inks.

• 47:39Speaker 02

  Name a color, name a color right now, I got it.

• 47:43Speaker 00

  Prime, can I tell the story about what happened on the call for nothing but nut?

• 47:47Speaker 03

  Sure. I don't even know what the story is, but you could, you could.

• 47:51Speaker 00

  Well, it does remind me of you saying Hewlett Packard and like we all call it HP and you'd like not knowing acronyms. So we're doing an event next week with Sentry, uh, in, at the Chase Center, which is where Golden State Warriors play. We have to go to a game afterwards. We're gonna do something very fun. It's gonna be a surprise, actually. We can't tell you what our actual thing is. I'll tell you, Casey, afterwards, um, in trash.

• 48:12Speaker 03

  Ooh, secret details. Sorry, chat, you're not so enough to know this. Only I get to know.

• 48:18Speaker 00

  Yeah.

• 48:18Speaker 01

  And me.

• 48:19Speaker 00

  Yeah, or they have to come in person next week.

• 48:20Speaker 02

  Trash, you might get kicked out.

• 48:22Speaker 00

  Yeah.

• 48:23Speaker 03

  After all the snacks he's provided.

• 48:26Speaker 00

  We're like 15 minutes into the call with them and there's someone from Golden State on the call. Sentry's on the call, me and Prime are on the call. They've said things like, you know, the GSW staff will do this, or like, you know, go through GSW this, blah, blah. They've said the acronym GSW about 35 times at least. And we're 15 minutes into the call and finally Prime goes, "So what is GSW?" Cause I'm like, get shit one. Like I can't say that in my head, get shit one. I had no idea.

• 49:02Speaker 03

  Yeah, is that like graphics software or?

• 49:08Speaker 01

  I had no idea you guys were doing it there. That's crazy.

• 49:11Speaker 00

  Oh my goodness. And we're on the call, like you, we've been talking about the Warriors, we've been talking. It's not like we were surprised about what NBA team is at the place.

• 49:19Speaker 03

  Golden State, what is the Golden State? New Jersey? What is the Golden State? Where is it? I don't know that acronym. What's Golden State?

• 49:26Speaker 00

  Golden State, that's, it's the San Francisco team. Golden State Warriors, Steph Curry, Golden State. That's even better.

• 49:34Speaker 03

  I don't know basketball at all.

• 49:36Speaker 00

  That's fine.

• 49:36Speaker 03

  Oh, I missed, oh crap, I missed my chance again. I was supposed to say something like, I don't follow hockey. And that's twice I missed an obvious setup. Damn it.

• 49:45Speaker 00

  I thought you were gonna say something like, I never watch moving balls on the screen. I hate that crap.

• 49:49Speaker 03

  Yeah, exactly. All these balls bouncing.

• 49:51Speaker 02

  It is a proven fact that people cannot cover head speed.

• 49:56Speaker 03

  Alright, so Golden State Warriors is the San Francisco basketball team? I believe so.

• 50:02Speaker 00

  Possibly the greatest player ever.

• 50:04Speaker 02

  Yes.

• 50:05Speaker 00

  Possibly.

• 50:05Speaker 03

  Okay.

• 50:06Speaker 02

  Maybe. Alright, so I do wanna, hey, I, in my own personal defense, I would like to say that that's actually one of my strengths. That I am willing to ask any question.

• 50:17Speaker 00

  That part was good. I was proud of you. I was proud of you that you asked even though it was like awkwardly late into the call. It was just it.

• 50:25Speaker 02

  I thought I could just pick it up. I thought I could just pick it. Honestly, I was like, oh, they used it once. I'll probably get it the second time. They use the second time. I'm like, I really don't understand it from this context. Here we go. I have to ask now.

• 50:35Speaker 00

  It just cracked me up because it was like, we're just, it happened to be, if it was just Sentry and not someone from Golden State Warriors, I wouldn't have even laughed or thought it was that funny. But it was just because we had someone from the org there.

• 50:48Speaker 01

  That's like a record scratch moment.

• 50:50Speaker 02

  It was, it was.

• 50:52Speaker 00

  They didn't care. They thought it was funny.

• 50:54Speaker 03

  Hey, man, it's better than me not knowing which that like, like thinking like, oh, New Jersey, like, cause that's like, I don't think anyone from San Francisco wants to be told they live in New Jersey.

• 51:02Speaker 00

  Yeah, yeah, they're offended by that for sure.

• 51:04Speaker 03

  Right. It's like, where are you guys? Where is this team from? It's like New Jersey or something. It's like, no, it's San Francisco. It's like, oh, well, slightly dirtier, but still okay. Yeah. Um, here's something I would like to share with everyone.

• 51:19Speaker 03

  Since you brought up Hewlett Packard.

• 51:20Speaker 00

  Great.

• 51:21Speaker 03

  I just want to underscore for the, all the young people out there, which is like everyone in chat.

• 51:28Speaker 03

  A lot of people don't remember. Because or because they were too, they were like in diapers at this time. I just want to point this out. Just to underscore how much this industry changes in a relatively short period of time. If you were to say today. I've got a brand new CPU, brand new chip coming. Get hyped, everyone. I've got a brand new chip coming. From Hewlett Packard. Literally, nobody would be like, oh, wow, like, I can't wait to see, or most people would just be like, what? Like, what are you talking about? Like, they don't do, they're just like.

• 52:08Speaker 03

  They're like Dell or something. They're just, you know, they just, they make like printers or monitors. I don't even know what they make. It doesn't matter, right? Hewlett Packard was actually the partner.

• 52:19Speaker 03

  Them and Intel were the two people who were supposed to have designed the chips that you would be using today. When x86 was going to move from 32-bit to 64-bit, they announced that the way that would happen was a thing called Project Itanium. And the Itanium was made by, architected by Hewlett-Packard and Intel as a joint venture. They were the people, because Hewlett-Packard was like a big name in computing at that time.

• 52:49Speaker 00

  Wow.

• 52:50Speaker 02

  I remember HP.

• 52:51Speaker 03

  Yeah, and the entire thing fell apart and did not end up producing chips that anyone wanted, because the only reason people are on x86 is for backwards compatibility. So Itanium was not at all what they had in mind. And so that was like the last, the last there was of Hewlett-Packard being like a big player. Like, if that had happened, they would be a company you'd be thinking of today as like major CPU player potentially, right? But as it is, now they're just completely forgotten. Like, no one cares about, sorry HP, I'm sorry.

• 53:23Speaker 02

  Actually, HP, we could turn this image around. If you would hire a B-list celebrity, I don't know who.

• 53:29Speaker 03

  Yeah.

• 53:30Speaker 01

  I was thinking. I got an HP printer right behind me.

• 53:34Speaker 00

  I was thinking, though, if HP wants to get into the chip game, they just start making new snacks and send them to trash.

• 53:41Speaker 03

  Yep.

• 53:41Speaker 00

  That'll get them on the map.

• 53:43Speaker 01

  I use my HP printer every day.

• 53:45Speaker 03

  I just had a, I just had a vision of like, of like trash going to Hot Chips, like trash going to Hot Chips 2027 and being like, there's no snacks here at all. This is the worst conference ever. This is terrible. Like, you guys don't know shit.

• 54:00Speaker 01

  Snacks at every conference is always bad, and I'm just gonna say that now.

• 54:02Speaker 03

  Yeah, that's true.

• 54:03Speaker 02

  You should go to Hot Chips and you should get a booth. And you should be giving out the best chips out there.

• 54:11Speaker 00

  Dude, this, a stand-up booth where we just hand out snacks?

• 54:16Speaker 03

  Yeah.

• 54:16Speaker 00

  Oh.

• 54:18Speaker 02

  That was hard.

• 54:19Speaker 03

  Tired silicon chips, wired corn chips. That's, that's what we're talking about here, guys.

• 54:25Speaker 00

  There we go. Well, it's popcorn chips now or something. Well, I don't know what trash is eating.

• 54:28Speaker 03

  Popcorners.

• 54:29Speaker 01

  Popcorners.

• 54:30Speaker 00

  Yeah. And y'all earths.

• 54:33Speaker 02

  Well, at the end of the day, this tweet sucked. I personally think, Casey.

• 54:39Speaker 03

  Yeah.

• 54:39Speaker 02

  The reason why he tweets this stuff is the same reason why Elon Musk tweets stuff and the same reason why Trump tweets stuff.

• 54:47Speaker 03

  Okay.

• 54:47Speaker 02

  And all those people do, which is.

• 54:49Speaker 03

  And what is it?

• 54:50Speaker 02

  Uh, that nobody can tell them no.

• 54:54Speaker 02

  And they just are like, this is who I am. I'm, and they just go out there and whatever comes to their mind is what's going out on the internet and that's just that and you just get stream of consciousness, whether you love it or hate it, that's what you're getting, baby.

• 55:07Speaker 03

  I mean, it could be, it could be. I don't know.

• 55:12Speaker 03

  I mean, I have to think that most people are trying to, when they're communicating to the public, they're trying to communicate in a way that gets the reaction that they wanted, usually, right? Like, even if that reaction is I was trying to piss someone off.

• 55:30Speaker 00

  Right.

• 55:30Speaker 03

  But I don't know, maybe they're not, like, maybe Sam Altman really just doesn't have any idea, like, or doesn't think about the fact that people will react to things.

• 55:37Speaker 00

  I feel like for a lot of people, that tweet played well.

• 55:41Speaker 03

  I mean.

• 55:42Speaker 00

  For people in AI sphere.

• 55:43Speaker 02

  By the numbers, by the numbers, 33,000 likes.

• 55:47Speaker 03

  But is that, isn't that just kind of generally what tweets from Sam Altman get?

• 55:52Speaker 00

  I don't think so.

• 55:53Speaker 02

  I don't know.

• 55:53Speaker 00

  33,000 is a lot, but we can, we can, we can do some sleuthing.

• 55:56Speaker 02

  2,000, 2,000.

• 55:59Speaker 00

  I've just, I think like for people.

• 56:01Speaker 02

  Crazy, grab 6,000, 11.

• 56:03Speaker 00

  People in AI, particularly, let's say.

• 56:06Speaker 03

  That is, that is somewhat high.

• 56:07Speaker 00

  Yes.

• 56:08Speaker 02

  Disproportionately very high compared to all of these.

• 56:11Speaker 03

  Well, 12K, that one, I guess, was popular. I don't know what it was.

• 56:13Speaker 02

  They had 10K.

• 56:16Speaker 00

  I'm saying people who come into.

• 56:18Speaker 03

  34K.

• 56:19Speaker 00

  They're doing their first coding thing ever. They're doing other stuff. They're not even, they're not software developers, right? Like.

• 56:26Speaker 00

  That they don't it's they don't have any respect like for the craft or like who came before or you know in the sense of like why and why should they right like that doesn't make any sense that would be weird for us to just be like yeah you guys should really know who John Carmack is and why he was awesome at programming and like all of this stuff and what it's like they don't it's just like it's cool that I also it's cool that I could talk to my computer and it can make something even if it's bad that's dope right so like okay we'd separate out the fact that like I don't necessarily like maybe how they got some of it or some of the people or the companies or their legislation or whatever but like it's sick that I can talk to the computer and be like make it purple make it in a purple theme no mistakes and changes the website to purple and then I'm like rounded corners and it makes rounded corners that is

• 57:09Speaker 03

  put it in a box and then another box

• 57:12Speaker 00

  in a box in a box in a box I want every div to have a border and three pixels of spacing so I know AI made that this webpage okay

• 57:20Speaker 03

  marquee tab

• 57:22Speaker 00

  marquee tab so that part you know like that part is cool I feel like for people who aren't like hey Sam you're like screwing maybe like there's some aspects where we're moving too fast you're screwing over the profession you guys are like putting out a lot of slop you're giving people the wrong expectations whatever if you don't have those like preconceived notions for it that that is like a nice tweet saying like hey guys thanks for helping us get here like we're excited for the future I got you know like for people who have a different background I'd be like that's how you would take it if

• 57:50Speaker 03

  yeah I mean it could be just a case of like addressing a specific audience at the expense of another audience and just being fine with that right

• 57:56Speaker 00

  yeah yeah I mean we're definitely in the minority yeah right there's way less people that have written code before AI than that will make some code artifact after AI exists like that's probably already true just because like people try random stuff on the websites but yeah

• 58:11Speaker 03

  right right I mean because it could just kind of spit out code so you would think right yeah

• 58:15Speaker 02

  looks like that's the stand up any blockers

• 58:19Speaker 02

  Your code that I look at every day.

• 58:21Speaker 00

  True cats trash for real.

• 58:22Speaker 02

  How much of my code do you actually run into? I'm actually like deep in your code. Wait, which one? I'm doing like TTR stuff right now.

• 58:33Speaker 00

  That's not a good sign. That's not a good sign.

• 58:36Speaker 02

  I'm gonna rewrite it all. Yeah, go for it. Please. Not even kidding. I didn't, I didn't design the current one. The, the TTR concept was mine. How it works and loads like safer evolution and for MDP versus. Leaking of secrets right now. Uh, that is all, that was, that was a different person. I won't say his name, but that was a different person.

• 58:59Speaker 00

  Casey, any blockers?

• 59:01Speaker 03

  Um, yeah, I don't want to talk about it.

• 59:07Speaker 02

  Kind of seems like you gotta talk about it. You gotta, you got something on your chest? No, it's the bad one.

• 59:11Speaker 03

  Let's get in this too. I.

• 59:15Speaker 03

  I think I've said on this podcast before, uh, literally, I think I said it on this podcast, but I don't know. Maybe I, maybe it was somewhere else. Uh, but I, I have said that the, the one time where I can relate to people who use AI is whenever I have to interact with like, uh, a web stack of some kind that like, right?

• 59:36Speaker 03

  Cause I'm just like, yeah, I, I can totally understand why you just want to type in something like, figure out like, you know, whatever this random BS is, that's like 50,000 text files of random config stuff, you know, half of which is ad hoc. And, uh, so I currently, uh, I have to move a server from like an old server that was running like this little, I wrote this little C program that just does our mailing list management for us, but it has to hook up to a web server somewhere. And like, it was on Apache and now it has to be on NGINX for some reason. And they don't support CGI bin.

• 1:00:18Speaker 03

  They only support fast CGI. So now I have to just go like re-link the program with like a fast CGI library. But of course I've never looked at the fast CGI library before. So like, I don't love having code chip that I haven't looked at and all this, like. So that is currently what I get to do later this week for no reason. This is all useless work, right? That's why I say like I always relate to people who like to use AI. Because I'm just like, this is useless. There's no, there's no point to any of this. It doesn't accomplish anything. It's literally just busy work to move something from one protocol to another with no value add other than just that's what happens to be occurring on these servers, right?

• 1:00:54Speaker 02

  I think you also said phrases such as CGI bin that most people will have no idea what you, what you're saying and the other people will shudder a little bit.

• 1:01:04Speaker 03

  Yeah, so in the old, old, old days, like wicked old days, that was just how you got, if you have a web server running and you wanted it to interface with like a C program that you'd compiled. So you wanted to extend the behavior of the web server. There was this protocol for like, okay, here is how we will invoke some executable on the drive and we'll set environment variables to let it know what, you know, what the user had set up in the form that they were submitting to the CGI bin or whatever it was that you were doing, right?

• 1:01:38Speaker 03

  And there's a lot of reasons why you wouldn't want to do that in like a serious scenario. The biggest one being that you're actually having to execute the, you have to run the executable every time, right? So, you know, in other words, there's a process that has to be started and so on. So if you were thinking about trying to do something high bandwidth, you wouldn't want to do this. And so the way that they, you know, solved that problem for things that wanted to look like this is they created a thing called, you know, fast CGI, which was just like, oh, now we just run something and we'll just communicate with it by using like shared memory or something like this, right? You know, we can, or packet passing or whatever, you know, we'll just have some way that will solve this problem so that the process stays running.

• 1:02:19Speaker 03

  And so that's how you would do it today if you had, if you have some program that's like an external program, but it's not, it's not as convenient because it's not really external anymore, right? It doesn't, you now have to have that protocol baked in.

• 1:02:34Speaker 02

  It just sounds like a web server, but worse. It's just like, hey, here's a web server, but unique. You don't even get HTTP, you just get like.

• 1:02:43Speaker 03

  Well, it's another way to look at it would be, hey, you've got the, you know, you want someone else to handle HTTPS for you because you don't want to re-implement TLS and you don't want to implement an ASN1 parser and you don't want to do that stuff, right? Which is very understandable because that's a huge pain in the butt and it's constantly changing too. They're like, oh, you know, like if you had done that back with TLS 1.1 and then they came out with 1.2 and 1.3, like now you're running back and upgrading your software to like the new cipher suites and all this other crap, right? So it's like, it's a maintenance headache too. So if you just wanted someone like NGINX to do that for you, but you would like to write the part of the code that does the

• 1:03:26Speaker 03

  rest, fastCGI is kind of a nice way to do it. You just go like, alright, you know. But you could also do a plugin, you know, NGINX has a plugin architecture, you could use that. I don't know. I don't care about any of this stuff. I'm just saying how it is.

• 1:03:38Speaker 00

  He's just blocked by it.

• 1:03:39Speaker 03

  He's just blocked. I just don't want to do it. I just don't like this kind of thing. It's annoying.

• 1:03:43Speaker 02

  Okay, okay, okay.

• 1:03:45Speaker 03

  TJ, any blockers?

• 1:03:47Speaker 00

  Yeah, about six months ago, BeeganBot promised me a funny edit of a video where there's a song where they interspersed goats screaming inside of a Taylor Swift song. And I really, my wife asked for that, but with a pan flute instead because that's Beegan's signature song for, or signature sound for his Sono songs. And I'm still waiting for it. I've been blocked on it literally for six months. My wife regularly asks me about it and every time I'm on the phone with Beegan, she asks and he always says, he's gonna get that to me soon, tomorrow, today, I don't know. He said, but so I'm blocked on that. So, thanks.

• 1:04:24Speaker 03

  If you're having trouble with your wife, Tej, I do think Trash has a pretty good app he's been working on that should probably help with that. It will resolve.

• 1:04:35Speaker 01

  It'll resolve all marital problems.

• 1:04:37Speaker 00

  Okay.

• 1:04:38Speaker 03

  Because there won't be a marital to be worried about.

• 1:04:41Speaker 00

  Right. No more problems of the marital variety.

• 1:04:45Speaker 01

  I want to, prime, last week when I was at the office, I was like going through like... Somehow I end up on one of your old PRs and it was you and Benmo. And it was like some formatting thing and you're like, "Sorry." It's like, you know, vim things. You wouldn't get it. It just made me like crack up in the office just by myself. I was like, "That sounds exactly like Prime."

• 1:05:04Speaker 00

  Hey guys, if you work at Netflix and you see Trash, make sure you say hello. He loves it.

• 1:05:09Speaker 02

  He does love it. Alright, also, hey, I have some blockers.

• 1:05:11Speaker 00

  I know, I was gonna ask Prime what's your blockers.

• 1:05:13Speaker 03

  Prime has blockers.

• 1:05:14Speaker 02

  Yeah, yeah, yeah. My blocker is that Casey needs to say NGINX correctly, or I'm gonna lose my mind for the rest of the day.

• 1:05:21Speaker 03

  No, I'm not gonna do it. NGINX. I'm not gonna say NGINX.

• 1:05:23Speaker 00

  I'm pretty sure it's pronounced "nix." That's what people are always talking about.

• 1:05:26Speaker 03

  I say "nigks." NGINX. I'm not saying NGINX. It doesn't look like that to me.

• 1:05:30Speaker 01

  No, I'm not.

• 1:05:33Speaker 03

  I'd be like, "N-ginks" is like what I would actually say if I was to try and say it. It doesn't, it doesn't look like NGINX.

• 1:05:37Speaker 01

  You're saying every letter, huh? NGINX.

• 1:05:39Speaker 03

  Yeah, NGINX.

• 1:05:40Speaker 01

  I guess it's not that terrible to say NGINX.

• 1:05:43Speaker 03

  It sounds more sophisticated, like if you say NGINX, it sounds like you're talking about some kind of like municipal telephonic, like telephone exchange or something like this, right? Like, it's like, "Oh yeah, NGINX." Like, they service all the Greater New England area's telephone service, right?

• 1:05:59Speaker 00

  Mine's N3X. That's what I like to call it.

• 1:06:02Speaker 01

  Yeah.

• 1:06:03Speaker 02

  Alright, well, that's that. That's the stand-up. Everybody get the hell out of here and get to work, okay?

• 1:06:11Speaker 00

  Oh my goodness, we gotta get to work. Bye, everyone.

• 1:06:13Speaker 02

  Bye.