Artemis Returns, AI Compute Wars, and Codex Control

Andrew Mayne: Hello and welcome to the Weird Things Podcast. I'm Andrew Mayne, joined by Brian Brushwood. Hello, hello. Justin Robert Young. Hello, friends. Andrew Mayne: So today we're going to talk about, guess what? Hey, Artemis, Artemis, Artemis, they came back. We didn't talk about it last week because we didn't do a show. Awesome. Justin Robert Young: Yeah, I left. We had an open AI podcast shoot and I went to SFO right afterward. I walked into the Amex lounge. They called me the most relatable man in America, for the record. And there at the lounge, gathered around the television like it was the 1960s, everybody was there watching Artemis splash down into the ocean and then everybody clapped. It was probably the most wholesome moment I've seen in public in the last 10 years and I was very, very excited. I was in a public place when that happened. Brian Brushwood: That's good. I think the price tag came in at just shy of a hundred billion dollars. Worth it. Worth it. Andrew Mayne: I, I, listen, if my friend blows his credit card to throw a party, I'm going to still show up. Andrew Mayne: I'm not going to stand on the curb and lecture him. I'm going to go show up. I'm like, man, this was an amazing party. It was a great party. I, I, listen, we, we have, you know, we, our criticisms of SLS are, uh, for longtime listeners are well known. And we probably could have done this, I don't know, 10 years ago, a lot cheaper, but we win it. And kudos to the astronauts who did it. It's an amazing day. It was an amazing experience, amazing time for space. And we're also in this great position now where we have SpaceX, Blue Origin, Rocket Lab. You know, it's a very, very different world than it was 10 years ago. Artemis, uh, you know, there was some questions about the heat shield. Was that going to hold up? Looks like it held up just fine, which, you know, thank God, you know, our astronauts are fine. Um, looking forward to the next artist missions. And listen, we could have boots on the ground again on the moon soon. And I like, I like of us, I like it when we think of ourselves as a space-faring civilization. Justin Robert Young: Yeah, I don't think the world is prepared for a moon landing operation in the social media era. Like, it's going to dominate everything. The high-definition video from the moon, instantaneously, pictures, videos, memes, jokes. From the moon, like, it's going to suffocate every news story. It will be the biggest thing, and it will be the most compelling moment in human history. Even more than 1.0. Also, my favorite use of the phrase, boots on the ground, is when the moon shows up. Yeah. Andrew Mayne: Yeah, it's the good kind of boots on the ground. Uh, so that's exciting news. I, I am, again, I am, it was, it was, it was heartwarming. It was great to watch them. The astronauts seemed great. I watched the splashdown. I watched them as they got pulled aboard the, uh, the carrier. All of that was just amazing to watch. Incredible and incredible. And, you know, uh, we're back in space in a big way. And that's exciting. Justin Robert Young: Yeah. Anyway, now that we've gotten that out of the way, we're going to talk about AI for the rest of the episode. Um, so, uh, uh, a bunch of stuff happening. Justin Robert Young: Anthropic, big week and a half, Mythos, relief, announced. Brian, I, I called Mythos the steamed hams of AI models. Uh, uh, it is, it is happening in the kitchen, but Superintendent Chalmers is not allowed to see it. Yeah, no, uh, it's the Aurora Borealis. Yeah. A frontier model with those capabilities. Capable of destroying and hacking everything. Skynet embodied in a piece of software right now in your model inside your HQ. Can I see it? No. Justin Robert Young: Uh, but still, I don't, I don't think that, look, I'll put it this way. If Meta or Google said that they had a model with those capabilities, I would, I would give it a, give it a scance, a scancy pants look. I don't think Anthropic has that reputation of, of, uh, overestimating. I think it is everything that they say it is. I think that we're not seeing it because they can't serve it because they're having real compute problems right now. And that's a really large model. And if they're going to release it at any kind of scale, even at their highest tier, uh, it would probably cause problems for them. Well, let's, let's, let's do a fun game. I call pretend the audience is hearing this for the first time. What is the story? Justin Robert Young: Uh, well here, I'll, I'll, I'll, I'll do the dummy version and then Andrew can fill in with, uh, the actual facts. Uh, so the biggest problem with AI in terms of serving it is compute. It takes a lot of just the, the raw processing power to both train the models and then serve the models. And so when you look at all the, the tremendous build in data centers over the last year and a half, that is trying to solve that problem, that the demand for AI is a arrow to heaven and, uh, the ability to serve it is becoming more and more challenging. Well, about a year and a half ago, Sam Altman was scrambling like his hair was on fire, trying to get as much compute as possible. Line up deals. There's a lot of criticism for circular deals with all these chip makers and stuff like that because it was a complicated structure. But the idea that open AI was always saying was we just need to get chips. We need to secure commitment to chips as fast as possible and as, as big as we possibly can because the compute is going to be a problem. It's a problem for us now. It'll be a bigger problem in the future. That's where the, all the data centers and stuff are. About six months ago, Dario Amodi of Anthropic was on a podcast commenting about this kind of voracious compute demand and saying, well, you know, does it really make sense to be securing all this compute? Because, you know, you're making a lot of commitments down the road. If you are too aggressive six months from now and you've committed to all this data, well, you could wreck your entire company. And now, Anthropic is seeing a gigantic spike in its consumer usage of Claude because of all the publicity they got for the Department of War kerfuffle. They also have continued to be a big enterprise player. They were leading in that field, have been leading in that field for a little while. But a funny thing happened, they hit their compute ceiling. And so now, I'm not going to name the friend of mine that routinely sends me these things, but there's a way that you can see the status of how up Claude is. And what you want is fields of green, like a spring day, just green, green, green, green, green. That means that it's up. What you ultimately want in any internet service is 9.99% uptime. So aside from some random speck of dust, you are always available. They have not had that. Instead, it looks like a fall day oftentimes. It is a lot of reds and greens and yellows. Not good. There's also been a lot of complaints with Claude super users that all of a sudden, processes that used to be very easy or effective with Claude, that now don't work the same way, that they might be restricting their models. Now, a lot of that's anecdotal. Anthropic has denied that the things that are being alleged have been done. But there's enough smoke there to assume that there is at least some kind of fire to them trying to correct or at least soften their compute problems. And so now, you know, you have this big mythos model, which, again, I believe is everything that they say it is, that it can create tremendous security vulnerabilities to the point where they need to keep it in the hands of professionals. I also wonder whether or not they could serve this if they wanted to, because they're having a hard time keeping the things that they currently list on their menu available to customers. Brian Brushwood: So I got a question. Let's say alternate Sideways Universe, they did release Mythos or whatever. And let's say somebody pulled a gigantic hack that never could have been done without the use of Mythos. I know that not only are the models frontier, but also the legal landscape were in frontiers, because wouldn't that be them providing the tools that caused damage? I mean, all of this is being determined. We're still figuring out copyright. Justin Robert Young: The biggest stuff that Mythos pointed out when they launched it was vulnerabilities in either proprietary software for companies or open source software that is used to serve things across the Internet. These are what are known in the hacking world as zero days, meaning they are unpublished vulnerabilities that the company does not know about. And so you can exploit that zero days currently. I don't think that we would see a major hack. What I do think is that the black market that sells zero days, which is, by the way, a multi-billion dollar industry. would or is. I'd actually be curious. I should talk to some people that know some people in this field. Brian Brushwood: Basically, the shortest version is the moment a hack is used, then people know it can be used, and then they go to work fixing it. And basically, it's like a starting pistol goes off. So zero days are, especially if you're a state actor or planning a massive takedown all at once or a heist or something, then you want to purchase those zero days. Justin Robert Young: So you can exploit a zero day for as long as you can before somebody finds out, right? Like, and there have been some that have been devastating that are at the underpinning of a lot of different tech that we use every day, like Heartbleed. But oftentimes, like iPhone zero days or iOS zero days, they will sell for a lot of money, a lot of money, because there are state actors who want to punish their enemies and now have a way to do it. That's what I think mythos would probably be most used for. I don't know about, it's funny, I was talking to a friend of mine in the hacking world about how on-device AI models would work, because there's a lot of open models now that are really capable and effective, and I was curious their take on it. And, you know, right now, that's still a pretty nascent world. People are still trying to wrap their head around exactly what that would look like in terms of custom code spun onto a device if you were to physically plug something in. But I think mythos would more be used on the zero-day side from what they were talking about. Andrew, how'd I do as an AI expert? Andrew Mayne: Great. That's why you're the host of the Attention Mechanism and a producer of the OpenAI podcast. So, Brian, your point about, like, I don't think it's really a liability issue because we've been finding exploits with these tools before, and some other people went to replicate what Mythos said it found and said they could use existing software to do it. I do think that Mythos is probably much of a capability jump in doing that. Andrew Mayne: My take is that, like, yeah, I think they trained a very powerful model. I do think it was very good at finding these exploits. I also think that maybe they didn't finish fine-tuning it to make sure that it kind of deters from that, because OpenAI released their GPT Cyber, which is their own initiative using 5.4 to make it a more broadly available tool for finding exploits. I think that in training a model, you do what you call your red teaming and you have to figure out, like, how you avoid it being misused or whatever. It feels like they didn't finish that from Mythos, and it was just easier to say, hey, because Andrew Mayne: right now, 5.4, OpenAI's GPT 5.4, my opinion, better than Opus, probably better even the new Opus model. I'm willing to give Anthropic the benefit of a doubt that the new Anthropic model they released, the new Opus model, is probably better than people are reacting to it right now, because I've seen that happen with chat GPT models, where people try things and sometimes it's a harness, sometimes other stuff. I do think Anthropic, they've admitted to doing some things to try to optimize stuff, and that can change the way, you know, things work from, like, how long it lets it think and other stuff. And there's always just sort of a messy part there, but they've been consistently good at replacing, you know, releasing solid improvement models. That's their reputation, as Justin pointed out. So I'm not ready to say, ah, they completely screwed up this time. I think Mythos is being a bit hype-y for them, and I think that that's a bit of a tone change, and I see a lot of people picking us up and like, oh, Mythos, and it's like, it's cool, like, great. When are you shipping? Can you ship a version that's usable? I think that part of this is reactionary because, you know, again, hype right now is OpenAI Spud, the model that's supposed to be due next week, you know, and that right now 5.4 is winning a lot of hearts and minds when it comes to coding. It is a cheaper model to run, and it's also apparently, you know, my test is better, and it's doing great things right now inside of Codex, which will go, I'm going to wax on about. But I think that Anthropic knew that they needed to do something because they've got great headlines. Their headlines are great, and I'm watching people get excitement. Their stock is doing great, and I think they're building great products. They released a new design product today that, like, just further stomped on Figma stock, which is, you know, because I think that they have the capability to do a lot of great things in different areas. But I do think when it comes to frontier model stuff, I think they may be feeling a little bit of anxiety about what's going to be coming out from OpenAI. And better to come out with Mythos and, hey, this and Opus now this week, say, hey, look what we have. Look, we're still good. We're great. Then to release it afterwards. And that's the problem that XAI has. Justin Robert Young: Opus is also the premise for why Anthropic might be trying to patch things up with the government. Report today from Axios that Anthropic hired a new, Axios didn't report this, but this has been reported in other places. Anthropic hired a lobbying firm in D.C. that has ties to the Trump administration. And so they put their CEO, Dario Amodi, out in D.C. to try to make nice with people that he wouldn't normally make nice with, including, like, the Heritage Foundation and a few other right-wing think tanks. And he, according to Axios, meeting with Susie Wiles today with the premise being, oh, Mythos is so powerful. Of course the government would want Mythos. So what do you say we make nice? And I'm just, this is my own analysis, but just the way I have read these stories feels to Brian Brushwood: me like a golden bridge. Yeah, it seems like a face-saving get-back where nobody's retreating. Justin Robert Young: That Anthropic can be, the, Anthropic can say, well, it was because of Mythos that we came together, and they can just stop talking. At any moment, they can stop talking, take the deal that OpenAI took with the Pentagon and just not talk about it. They don't have to talk about any of it. In fact, usually nobody talks about any of this. They can just say, we came to an agreement that we are all Andrew Mayne: comfortable with and move on. Yeah, it'll be interesting to see what happens there. Sure. I, we, we need to have all of our AI companies in good graces with the government and, and particularly, you know, cause it's, it's, you know, all hands on deck. And it is, they have done an incredible PR, Anthropic PR, the, the, what they've been able to do, you know, from like the amount that they've been spinning on influencers is insane. Like, like my wife started noticing these people just raving about Anthropic and very, very tiny print paid for by, and nobody notices that. Like, that's the crazy things. Like they've done, they've got a great product. First, let's say that people are very happy with the product. So it's not like it's a BS product that's being pushed by, you know, but it is, they have been spinning incredible. I saw like, you know, some big influencer dinner and all the swag and I'm very sure everybody's doing that. So they've been spinning incredible amounts. They've been doing a lot to sort of one, and I think they've won over kind of the blue sky mindshare of things. Uh, I would say the reality is probably different than what people think. You know, I'll have people describe a situation like, well, that's not what consistent with this. And people are like, oh, the White House is talking to Dario. Dario wins. Like, I don't think that's the way that it's working. I don't, I don't think this is, you know, that the White House is like, oh, well, we're so afraid of Mythos. We have to go talk to Dario. I think it's like, they're like, hey, can we, can we reset? Because they're in the middle of a court battle and it's not good for either party. And I think that they want to reset that. But, um, I, I think that they could make the same deal. I think we'll have to make the same deal that OpenAI did, which, which Sam, by the way, said, hey, they could go make. And I think a lot of us will look kind of silly and stupid because if they do do that, it'll be curious to see what do people say? Because people were upset with Sam before they even knew what the deal was. People might just assume like, oh, Anthropic won and it was the same thing. And maybe there's so much love for Anthropic that people won't look at the details. They'll just squint and do that. So. Yeah, we'll see. There still Justin Robert Young: are a lot of hurt feelings on all sides of that. I'd be curious to know. I think it's in everybody's best interest if every, if make the deal and everybody shut up. We just don't, it does not need to be litigated like this, that kind of stuff. I genuinely believe when you're that rancorous in the press and we understand how much, uh, the, the nation state enemies of the United States, and I'm just going to stay here, Iran and China, right? How much they are incredibly keen on the minute by minute rancor that happens within the United States and how much they use that to, uh, to, to, to fan the flames. Like you don't need to be doing that in public like that. There's Andrew Mayne: a reason why these contracts are behind the scenes. Yeah. But it wasn't just being a public originally. And then Anthropic internally, people found out the source of the deal and had broke ranks and complained. And I don't see that going away. I don't see a personality shift happening there. I think the first thing that's going to happen and they make a deal is that internally they're going to like, well, what was the deal? Dario, tell us what the deal is. Dario can either be like none of your business, go back to work slave or, you know, uh, well, we did this because it Justin Robert Young: was the best thing and we'll see. I mean, this is growing pains for that company though, isn't it? It is growing pains for them because like the part of getting to that next level, part of being a pillar of the economy is discretion. And you can't be a slack warrior in the way that Dario was. And I hope that he's learned his lesson from that part of this year, which has been a kind of best of times worse of times for, for Anthropic, uh, in like they've had this tremendous success, but they've also had. Some of this corrosion that has been brought on by bad behavior and I hope that they understand, I hope that they're learning because I do want them to be successful I think moving, the speed we're moving right now with AI, I want to keep going for a minute This is exceptional, the things that we are on the cusp of achieving Brian Brushwood: It's interesting because having less interest slash access to the specifics on the technical side I'm just sort of watching the battle for hearts and minds So to me, all of the PR moves have been largely political attack ads And in that regard, let's set aside the quality of the product or infrastructure strategic good decision, bad decision on the actual compute side of things In terms of competing for public mind share even this phase, even the mythos thing feels to me like predictably wild swings that you would expect from a third place competitor Like if we're in a Coke versus Pepsi between OpenAI and Google then this is, you could either be RC Cola a forgettable also ran or you could be the spicy cola, your Dr. Pepper on there So in terms of competing for the leftovers because again, the haves and the have-nots there's going to be a big, big, big, big drop-off after the third place slot Then we get it to and the rest And so, I don't know Justin Robert Young: Actually, Brian, you said something really interesting I'm curious about that So in your mind, Brian Brushwood's rankings This is apropos, we're not asking you to show your work This is just you looking from your point of view into the world of AI Rank the labs one through five Understanding that there is a drop-off but one through five, where would you put that? Brian Brushwood: Sure, sure Well, OpenAI, number one for Chad GPT Dominant player, you know top tier Second place Brian Brushwood: My parents might think co-pilot or whatever But to me, that's just a rebrand of OpenAI So I don't really think about that Instead, number two is definitely Google Gemini mainly because it just has so many moves that others don't It can watch your YouTube video and parse it and tell you this part lags This part you repeat yourself You know, you could try a different riff on that What if you used a metaphor, Brian? Oh, you have my You have me But also I also like that it's a different environment So like, for example because I've been stung by the occasional hallucination not only do I have Chad GPT show its work for example, I make a lot of codes and encryptions I will always take the output from OpenAI and I will paste it over in Gemini and say, decrypt this with this keyword And then, and that's where I'll find the cracks in the system So I tend to have personally a foot in each And I gotta be honest This week was the very first time I went to anthropic.com And I'm like, so, what do you do? And it's like, I'm the thing I'm like, do you know who I am? And it says, nope And I'm like, should you? It's like, I don't know What do you want to use me for? I'm like, I don't know But I did rather enjoy the level of for lack of a better term self-awareness that Claude had when I asked him like So why are you here? And it's like, well OpenAI's products are like this and I'm more like that And if you want to do that kind of stuff talk to me Justin Robert Young: Alright, so there we go That's your top three Brian Brushwood: Yeah, and then Copilot is a rebrand And beyond that Shoot, then there's Chinese spyware And then there's I don't know, piracy bot Brian Brushwood: And there's meth All the Claude bot stuff Like the crazy people Justin Robert Young: And then local models It is an abyss And the reason why I asked you and thank you for doing that is that I think we are You know, the way that The world that Andrew's built in is from the Frontier Labs perspective And you're talking about like pushing the actual boundaries of these models in very specific kinds of ways Benchmarks, evals, that kind of stuff I very much believe that there is a whole nother world that is much larger in terms of bodies that is consumer product and that's really what I wanted from you is like from the Brian Brushwood perspective which is very much rooted in the like I'm curious I want to give time and effort to this What are the best consumer products that I can use that leverage this technology The one, two is Catch EBT The Coke The reigning champion And then Google which if anything to me from the very beginning I'm almost it is a testament to OpenAI that they're in that slot because nobody had more of a head start Nobody still has more of your data than than Google When I'm I'm on Codex today trying to figure out cool ways that I can triage my inbox and I'm like why wasn't this a Gmail product two years ago They still haven't done anything to meaningfully make Gmail a better experience by way of AI So that's very interesting because I think from the frontier model perspective Andrew this is really it's a two company race between Anthropic and OpenAI right? Andrew Mayne: Well I mean AI is being so pervasive in so many other places when you look at the website traffic the biggest competitor to ChatGPT is Gemini like Gemini has been growing like a lot it eroding away at the lead that OpenAI has had OpenAI is still in the lead when you look at Enterprise you know coordinate entries in Horowitz data OpenAI is still in the lead in Enterprise but Anthropic has been growing by leaps and bounds and some people will tell you oh Anthropic owns Enterprise like well not by the last numbers but doesn't mean they're not going to because they are on a great trajectory and they've had you know part of what happened was because of OpenAI's deal with Microsoft was that Microsoft was able to do certain things in Enterprise is that OpenAI couldn't so OpenAI was a bit hamstrung and OpenAI focused on one their API growing user ChatGPT growth and then building out data centers and Anthropic was able to you know hey we don't have that deal we can go in and we can do enterprise sales you know night and day and they did that and work with partners like you know AWS and you know Microsoft as well and everybody and that really created a great opportunity for them to do a lot of really you know great inroads into enterprise and get a you know do gangbusters there of course what just changed was OpenAI now has a new deal with Amazon through Bedrock to be able to make way more impact in enterprise and you know I've talked about this before the models that get released today were planned years ago and started training maybe six months ago or so right and those were planned and they decided to train those things based upon what they think the market needs six months ago Anthropic is doing great with Opus because Opus was the number one code model arguably the number one code model right better than OpenAI's offering better than Google's offering there were places where OpenAI was good because it was cheaper whatever but if you just wanted to make sure that it got done done right Opus was the best model and I liked I still use GPT for a lot of stuff because I would like I'll do a lot of iterative coding but Opus overall was the best one just expensive though but anyhow it was still the best one but what Anthropic found was that companies that have a lot of software engineers were willing to pay tens of thousands of dollars a month per seat per engineer to give them access to Opus because cost wasn't the biggest factor it was still cheaper than hiring another software developer and it made them more productive so Anthropic was making more per token so for every token that OpenAI sold maybe they were making 10 cents Anthropic was making 50 cents I mean not really but you get the idea there right they're making maybe three to four times as much so Anthropic is like oh just create the best model the best model charge more for it and you'll be fine and they did that and they've been doing that and that's worked out pretty well but then when they said what do we do for the next model like well let's make this bigger model do we try to make it more efficient do we try to do that no we didn't do that before with Opus we don't need that that so they create Mythos their next big huge generation model but by the time Mythos finish training they're in a world where OpenAI's 5.4 model is better than Opus at coding and more efficient and cheaper they're in a world where OpenAI had said hey we still believe in efficiency we still believe in having like fast cheap models and the bigger model and this sort of stuff we also think speed is important where OpenAI is now about to unveil their new spud model which you know we'll see what that does the benchmarks for Mythos were pretty good but some people were comparing Mythos to GPT 5.4 on like max high like the highest SETI max reasoning in some places they were pretty competitive and it wasn't like this big huge it was a leap let me make it very clear I think Mythos is very capable a lot of places but in some of the areas it wasn't like as clear as you would think meanwhile OpenAI is now going to come out with their their spud model next week, which is rumored to be every bit or more capable than midfell's, but will be served to the public. And you see kind of where things land from where expectations are. I think, you know, the challenge that Claude and Anthropica's had is like, those outages are real. Those rate limits are real. Like you get, people will switch to an open eye subscription or surprise that they get way more usage out of that. A lot of people like you and me, Brian, we have Claude subscriptions too, but we only use it a little bit and we're not even really aware of the rate limits. A lot of people there who started there don't know how rate limited they are compared to somewhere else. So it'll be, it'll be interesting to see how things play over the next six months. Justin Robert Young: Yeah. Where can people go to definitely find the Weird Things podcast at all times? Patreon.com slash Weird Things. It's where you go. You support the show. It's a good time. You feel good about yourself. Feel good about life. All right. One last thing, Brian. The reason why we were joking about you buying a Mac. Like, oh yeah. Is because Codex is the app that OpenAI released that they use their coding models and you can make things in it. And it's great for vibe coding. It's great if you are a coder. It's just a good app. But yesterday they debuted computer use. This is a way that it can click on things inside of your computer. Whatever you can do on your computer in terms of clicking and typing, it can click and type. Justin Robert Young: Problem is it's for Mac only. And so if you want to set up automations, if you want to do those kinds of things, you're going to need to get a Mac. And I think Brian is, what, is there any time in your life, Brian, where you've come close to buying either a Mac laptop or a Mac desktop? Yeah, several times. And oftentimes it's been an inflection point where I've needed to buy a new laptop and I've said, all right, I'm going to do it. And then I do go to one of those websites that tells you the, oh, great Oracle of Mac. Am I going to feel like a dipshit for buying a Mac or a cool surfer on the technology wave? And then the Oracle comes up and it says, you are buying two-year-old technology that is very slow if you buy today. And I say, see you in a couple years, Mac Oracle. And then. Oh, gotcha. Oh, just the things that say for where it is in cycle that you should probably wait six months because they're going to refresh you. Yeah, the sucker weather report. Gotcha, gotcha, gotcha, gotcha. Now, meanwhile, I remember it was the wise sage, Andrew Mayne, who said, what's the point of having a job if you can't buy the latest Apple iPhone every single time it comes out? That I have taken to heart. Which, by the way, I'm off that. I settled my loan. I was on the buy every new iPhone loan package. And then my phone got stolen in Las Vegas. They wouldn't give me a new one because they had updated AppleCare two months after I had bought my most recent phone. Brian Brushwood: Where they now AppleCare, if you lose your phone, you just basically everybody gets one free. You probably don't want to be saying that, but that's what they explained it to me. And they were like, oh, yeah, you're just going to get one free. And then they're like, oh, they updated AppleCare two months after you bought it. And I'm like, can you look at my history and just look at the road that I've paved with iPhones that I've purchased? And like, you know, give me, you know, Willy Wonka me, homie, and fudge a little. Like, let's just move on with the iPhone. And they were like, no, sorry, can't do. And I'm like, okay, well, then I'm going to cancel my loan. Like, what's the point of being loyal to you guys if you're not going to scratch my back after? I mean, that punch card had to be full of at least a dozen iPhones. Andrew Mayne: Yeah, I, uh, did you ask him where they were on the day the, uh, the first iPhone was made available? Yeah, exactly. Were they there? Have they been there from a deck deal? Brian Brushwood: Based on the age of the woman I was talking to at the Las Vegas mall, uh, she might not have been born by the time that we were waiting in line for the first iPhone. But, um, yeah, yeah, I don't know. So, I, uh, I'll tell you, I've actually had more of my, my head turned by, uh, the, the, the trifold that our friend Darren Kitchen has. That is a, that is a full-ass iPad that just folds up in your pocket. I, I will admit, I'm pretty stoked about that, that upcoming iPhone fold. It looks pretty, pretty dope. I don't think I'll ever be able to break out of the iOS jail, but. Yeah, me either. I always get a little, but then I, this is probably the tension to get back to this, where it's like, everyone's in awe, you're tempted. You see some cool stuff. You hear some kids playing over the fence and you're like, oh, I want to play kickball right now. And then you get close to it and you're like, ah, but I can't really leave. That's the reason why I'd probably never buy an Android phone, um, aside from the fact that the hypocrisy would just be too much. Like, I've made fun of Android people for too long, but, uh, I don't know if I'd be able to show my face. Uh, but then also, I think for you with the Mac, it's like, ah, are you really gonna? But for this, though, you just need a cheap one. to run stuff on. Yeah, yeah. And, and especially for like, like, uh, I deal with a lot of media and, uh, the idea of agentically, you know, sorting and putting things into sensible file structures or whatever. Take a giant external media drive and like, chew on that, sort this according to so-and-so or likewise, like here. Andrew Mayne: Well, you can do that with codecs on your Windows now. You can do that. Brian Brushwood: Yeah. Well, then, uh, well, uh, tell me some of the use cases that become opened up. Andrew Mayne: So here's, here's the thing. Okay, so let me explain. So computer use has been available from other providers, like Anthropic has had their version of it, um, but it's slow. It's always been slow. That's been the problem. Computer use is very slow. And what Apple, uh, Apple had a group of engineers, very, very, very smart, cracked engineers that left and went off, created other companies. And one of those companies they created was, they created a, a, uh, kind of like a, kind of a computer using platform. And they understood a lot of things about how things work at the, the OS level. And OpenAI just bought them. So great. We like what you're doing. And so the way computer use works and why people are calling it magical on codecs is that instead of you just like using regular accessibility and you see your mouse cursor moving around and it does this, they basically are running a second layer behind the scenes. Where they basically can go into things that are behind your, what you're have focused on and they can run a cursor can go in there and operate that and go around and open apps. Like I can have it open up my Apple notes app and put things in there. A lot of things people that were using, you know, some of the other agentic stuff for, but basically they figured out a way to sort of do this at a different layer. And it looks magical when it happens because it's not like you're fighting for control of your computer. Or you're doing your thing and then codecs is doing its, its thing, controlling a cursor and doing it in there. Like I can go into codecs right now and say, Hey, draft me some tweets about this podcast. And it'll just open up X, go in there, draft the tweets while I'm talking to you and doing this. So that's, what's cool about that part. But if you say, Hey, I've got a bunch of folders with media, just take codecs, point it at the folder and say, I need you to go through these folders and go process this. It can already do that. Justin Robert Young: Well, so to my question, what, what, what are some things that become possible now because of on computer control? Andrew Mayne: Well, like I said, you have a computer that can control your computer. What, what would you like to do, Brian? Well, okay. That's you kicking the question back to me. Justin Robert Young: Let me give you an example, Brian. So for whatever reason, I very much, my brain connects to physical things and not digital things. Or at least they connect differently. I am a much more productive person when I write out my schedule on my whiteboard than I am when I look at it on a screen. A thing that I have long dreamt of was to have essentially my own presidential daily briefing that was tailored to my life every morning. Where I would get just a one pager of the most important things that I would need to know leading into the day. Justin Robert Young: I saw a computer use. And the reason why I thought of that was because this process, if I were to do it, would take a lot of different, pretty disparate actions. It would be summarizing my calendar. It'd be summarizing my Gmail. It'd be summarizing websites. It'd be summarizing newsletters that I have inside my email. And they would all have to go in their own correct little buckets. Justin Robert Young: So the hidden thing, I've tried to do this a few times in ChatGPT with very, very not great results. I tried to build a version of this in N8N. And it, you know, the best thing that I could do, even if I got it to work, which I never really was able to get it to a way that I liked, that it would do, is it would just send me an email each morning. But even then, it's like, okay, well, great. I have another email that is amongst all my other email. Awesome. That's a, it's a cool thing for me to also ignore. Now, and this was vibe-coded last night, I just, I described what I wanted to do in Codex. I asked Codex to put together a plan for me to do it, and then a prompt that would execute on that plan. I then applied that prompt, and it goes through the entire thing. And not only does it use the connectors to get into my email. I actually did it use the connectors to get into my email. It also is able to look at my messages because that's the computer use, right? There's no easy way to plug into your iOS messages through Codex. It's not going up to a server. It physically can look at my messages because it is also existing on my desktop. Then it spits out a PDF and then goes to my print center on my computer and prints it out onto my printer. And so at 6.30 this morning, I woke up to a piece of paper that had most of what I wanted. Some of the connectors were broken, so I had to figure that out. But still, that was something that was vibe-coded in about 15 minutes last night. And at least two of those steps, including the one that kind of differentiates it from an email, which is printing it, was something that now I can do. So before anybody comes at me, it's like, well, you could also do an air server that you could push to yada yada. It's now it's all clean. I don't have to go to another server to do it. It does it right on my PC. And I could not be more thrilled. But that's an example of something that was just not possible before that now is possible. Andrew Mayne: Yeah. I can mention before the two examples were like, if I wanted just to open up Apple Notes and go add notes to my Apple Notes, go find my recent notes, go look for that, create a checklist for me. You can just go into my Apple Notes and put it, and I open up my phone, it's there, drag images in there, open up the browser. I can create a Google Doc, you know, obviously programmatically, but I can say, hey, open up my browser, go create a Google Doc for this. And it's able to run different applications, you know, with stuff that's sitting there, anything that you can kind of mouse click around. I could ask it to check, you know, my text messages and go inside there and say, you know, I don't know, let's see if I can do this. Andrew Mayne: We'll see if we can do that. And you can do a lot anytime you have anything, you have to interact with an app. You know, it's able to do that. And, you know, let's see if we can get it to do this. So I think that a lot of it just figuring out your workflow and breaking it down to say what steps. So, you know, I could just say, hey, every day, figure out what's the most important email and put it in my Apple Notes or get some news briefings, put it inside of there. You know, obviously just in app, it understands Gmail and calendar. So right now I use it to triage my emails inside of there. But when it comes to running things, that's the nice thing is you can just open up your other apps and do stuff. Justin Robert Young: Yeah. Now, what I haven't done yet is start to automate, like, let's say, world's greatest content production. So Brian and I are in the kind of last, last furlough of an episode of World's Greatest Con where we had, like, a good re-record session yesterday. But because we're trying to get it perfect, there's a lot of different takes. Like, between the two different sessions, it's probably about maybe an hour, 45 minutes total of, like, if you just drop both files into audition. That's a lot, right? And it would probably help immediately, right off the top of my head, something that I could do would be, hey, go to my Google Drive, get these two files, ingest them, diarize them, or transcribe them, and then cut them up into separate audio files that are just by take. So it's like, if he's repeating the same things, that's one take, I need you to separate it out into a bunch of takes. So at the very least, I now know, okay, this is all this, this is all this, this is all this. So I have a better idea of exactly how many different holes we are trying to fill because this re-record session wasn't just for one part, it was for spot-checking a bunch of different things. I could do that. If we're starting, so whatever, this episode is going to be part one of a part two. If I want to then say, okay, here are the beats we need to hit for part two, I can go to Codex and say, all right, go into Atlas and use Suno to create these kinds of themes for these moments that are going to hit, just download all of those and lay them, automatically put them into the file, create a new audition file that you automatically lay these things into. And it can do that. Anything that you can mouse and click, now I can just automate a lot of that stuff where it's just going to save time. You know, that's, that's really the biggest, the biggest thing to think of with any of this stuff is like, all right, what in my, what is, what is the task I'm doing? What would be the moments where my human brain matters the most? Like, where is it really, really, really important? Where is my taste, my judgment, where does it really matter? And then there's other stuff where it's like, okay, well, that doesn't matter as much. It doesn't matter as much that I'm hitting a button or doing these mechanical things. It matters a lot that I pick the right song. It matters a lot that I edit things the right way. But if everything else can be laid out, that's what personal assistants do, right? And that's what you're available. Brian Brushwood: Well, a, a interesting comp is sometimes I'll record something and I'll get three different camera angles on it. But then what I have to do is after getting all the footage, getting the microphones, the three different angles, you have to import all of it into Adobe Premiere. And then you have to synchronize all of them. And then you have to set up the multicam shot. And then you have to, and the one that really slays is oftentimes the very last thing you have to do, and there's no two ways around it. We haven't found anything to automate this is you've got bleed over. If there's two characters talking, the mic is picking up both. Now you subjectively know this is the Brian mic that should be picked, that should be used. This is the Alex Rangel mic and back and forth. And so you have to go and you have to manually duck it back and forth in there. It sounds as though it's an elementary enough task that that can finally be automated because that part is... Andrew Mayne: Yeah, I, yeah, I went through a thing and we haven't had a chance to, to, uh, beta test it. But you know, the thing I set up for the weird things podcast now, cause like we were, so right now what I did is I have a, I have a Gmail or excuse me, a Google drive where Brian can upload the file to there and then, uh, worked in testing. But again, we'll, we'll try it on this episode. Um, what I have an automation that keeps checking that folder and if there's a new episode there, it goes in, creates a transcript, creates a title, creates the album art. And actually it processes, I found a thing called Alphonic, which is an API, which improves the audio and it will go through there and clean up the audio and then do, then put it into WordPress as a draft. Justin Robert Young: Andrew, by the way, is just doing the weirdest ventriloquist act of all time as, uh, while he was talking, our group chat just pinged with hello from Codex. So I can only imagine that Andrew, Andrew, while drinking a glass of water is having his Codex sing, uh, uh, sing a song here. Justin Robert Young: Uh, yeah, so it did that, um, uh, but that, that's an example of you are able to text via Andrew Mayne: that. So let me tell you what it was doing. This was the funny part. I had to intervene to save it some trouble. I said, send a text to Justin, Robert, Young and Brian and say hello from Codex. And so what I love about this was it goes and says, I look at this and it says, okay, yeah, I didn't say Brian Brushwood. So it literally said, Hey boss, can I have permission to open your contacts to see which Brian you mean? And it went through my contacts to go look to see which one I said, Oh, Brian Brushwood, save it the trouble. I've had it. I've done stuff where I'm like, Hey, Oh, catch up on this email. I need to intro these two people. And I don't know the, I don't say the email address and I'll search my email and find the email address and then create the email as in my draft folder. Like it's insane. The fact that I watched it, it went like, all right, I need to figure out this Brian. And then it says, let me see if there's a conversation between the two already. Oh, I found it. And then send it. Bing. Got him. And I can have it check too. I can have it check that thing too and say, Hey, every, every 30 minutes, check my iMessages to see if, you know, uh, you know, Brian was mean to me back at him, you know, as the Brian Brushwood: frequent Brian thunderstorm of meanness comes blowing through exactly. Justin Robert Young: Yeah. Regular malcontent, Brian Brushwood, uh, just, uh, stomping around your messages, furling invective. Andrew Mayne: When are you going to take your pie in the face for being wrong about SpaceX? Come on. SpaceX pie. Brian Brushwood: They still, they still haven't attempted this. We still don't know about the, the Starship. Do we? Andrew Mayne: Uh, yeah, I don't. Justin Robert Young: I think that's the current standing bet is, uh, the first time. Might, might, might, might, might be saving that one for after the IPO. Maybe, uh, maybe, maybe let that IPO roll on. Andrew Mayne: I, I, I, yeah, I don't know. Um, I'll accept my pie. I get it. I'm, you know, I would, I would assume that my losses on backing Elon, uh, my, my takeaway from this folks is, you know, the way. AI advise people when I talk about how to figure this stuff out. And again, I'm not going to, oh, the weird things workflow works great. We need to just put an episode we got through to test it and that's be part of the test it. But you break things down. And for that was, we had an interface portal that Brian could go use, but I'm also like, is that really a good use of his time? Can I just, can I offset some of the stuff to maybe the final testing to me so he doesn't have to go do that? So I'm like, what's the most minimal thing we can do? Well, I could have him take the MP3 and drag it into a folder and then let the AI take over from there. In theory, text message me when it's ready, et cetera. And it's great too, is that we have like, if you go to, you can see one of the things I did on the back end was I created weirdthingspodcast.com, which is, we have the weird things website, but instead of trying to like go through and redo a bunch of things with WordPress, I said, well, let me create a shadow site that has all the episodes. And stuff there with the transcripts, et cetera. Andrew Mayne: So I've just updated my transcript model. So the transcripts are going to get better from here. This is just doing a kind of a pretty rough diarization, but this was just how I wanted to just make it easy to put that stuff. And then also like when it creates a new episode, it creates a landing page. Brian Brushwood: Yeah. You guys got any pics? Andrew Mayne: I found a cool YouTube channel. The guy's done a few videos that I dig. Let me get the name of it here. It's science stuff. He'll do like, some of the lists are a bit repetitive, but it's Astromat. And I look at like, gosh, there's already a Chinese knockoff of this. So the channel's called Astromat YT and it's very small. It's got like 8,000 subscribers, but like 10 reasons why the universe might end soon. Seven terrifying deep space signals. He uses these formats, but there are some really cool ones in there. You know, theories on why time doesn't exist. You know, the whole block universe theory. I don't know if you've heard of that one, but like just some really cool stuff. You know, sometimes you'll see the same ones repeated in other videos. But I do like a lot of this sort of like, you know, 10 things. Like I, I, I don't have a lot of it. I don't have a very big attention span for 30 minutes for, you know, 10 minute videos crammed into 30 minutes. You know, and some people are, some of these people just take a long time to get to the point. And I'm like, I'm kind of tuning out. Brian Brushwood: Yeah, that is a, it is a strange set of incentives that YouTube's have set up where it's like, keep people on one thing for a long time. And so you end up with sort of a dopamine inducing same, same rhythm, but also a long time to say something that didn't need, didn't need to take that long to say. Andrew Mayne: Yeah, I choose my watch habits by run times. When it gets past 15 minutes, I'm kind of like, maybe not boss. Justin Robert Young: Hmm. I want, Brian, is there a name, and if not, can we create one for when the algorithmic demands shape? Like you can just see the stretch marks on a piece of content that is like very obviously algorithmically incentivized. Brian Brushwood: Yeah, well, I know that when content is, stretches and shapes to what will engage the audience, they call that audience capture. So I'd imagine maybe some version of algorithmic capture or rhythm capture or format capture. Justin Robert Young: It describes the fact that for a while, like the number one video, all the number one videos on Facebook, which just had a pure time spent watching, were all these just garbage, like, I just found my wife cheating on me. And it's just this guy in the back of a car, like edging, and it's saying like, ooh, I'm going to get my wife, my wife, I can't believe she's, she's not answering her phone. And it's like, it's literally just there to edge you for whatever, right? Because that's what moves these things forward. It was like very long arguments were like an entire genre of Facebook video. And YouTube being the king of that field, I think, is a little bit more artful with their algorithm and certainly changes it a lot more often. So you get different versions of it. But I think that's a fascinating thing to look at and be like, oh, why are all these explainer videos really slow? And it's like, because the algorithm incentivizes you to make your first point at least seven minutes. And if not, it will like not recommend you. And people have figured that out. And so now everybody's first point is seven minutes. Brian Brushwood: Yeah. And meanwhile, it's interesting because there is a temptation to believe that YouTube is shaping the content. But really, all YouTube wants to do is shape itself to the natural rhythms and flows of dopamine, oxytocin, serotonin, endorphins, you know. And in fact, I speculate that some research into like what is the half-life decay of dopamine or whatever might yield very familiar rhythms for what we like for the way stories progress. Andrew Mayne: Point of order, sir. I think yes, and but I think that YouTube is trying to maximize towards ad revenue. And that may be getting that dopamine thing in there. But sometimes that I don't know. Sometimes we overestimate how good they are at figuring these optimizations are. That was the thing we learned about Facebook because we thought they're way better. They were just doing cohort bucketing. I mean, I think you're right that that in one way, that's kind of what they want. But I do think that like the other side of the algorithm is literally, you know, I mean, Brian, I've got a pitch for you. I want you to talk about luxury real estate and jets. And you're going to watch your ad revenue for a while. Peek up, you know, there's a lot of that of just different audiences. Brian Brushwood: Oh, I didn't even think about that. Well, I know that that pretty much there are valuable demographics for targeted for ads in like there is a specific demographic of people who watch my stuff, our stuff, and they maybe are not as profitable to advertise to as other demographics. And so as a result, even though your content is just as engaging, just as blah, blah, blah, blah, blah, blah, blah, it's just those are or are not the people advertisers are trying to reach. Andrew Mayne: Yeah, it's hard. I mean, like I would say that even more so like Mr. Beast, the average Mr. Beast, you know, viewer isn't worth as much as, you know, some 45-year-old executive in the market for buying a new car or a watch or whatever. But they do add up. And we've seen that. We know people like, you know, that's one of the things is like, you know, we know people who do tech shows who like, you know, do like do very, very high tech focused stuff that, you know, have big revenue. Open AI bought TBPN, you know, Tech Bro Podcast Network. If you look at their numbers, I, you know, I talked to a friend like, well, their numbers aren't that high. I'm like, it's who's watching it. It makes the difference. You know, if you want to sell a data center, that's where you go. And that's, that's, that's, you know, understanding that like not all numbers are the same, you know. Justin Robert Young: Somebody, I saw, I saw on Twitter yesterday that TBPN was Cocomelon for VCs. Like it's just, it is just the thing, you know, you put up, but that's the audience, right? That audience watches that. Here's my pick. I finally got around to beginning. I have not gotten to the end of it because it's a four hour long video. But the defunct land on intelligent characters about the broken promise of Disney and intelligent characters. But here is the fascinating part. I'm watching it at the very beginning. It starts talking about how early Disney was in terms of buying computers, specifically pointing out that Walt died before the first personal computer. Really, he, he does not get to experience, uh, the, the computer revolution that I'm sure he would have been fascinated by. It kind of reminded me of like, man, Walt dies before the personal computer and Steve Jobs dies before AI takes this kind of gigantic leap forward. Uh, what, how different would our world be if, if either of those two didn't happen? But what was fascinating to me is that they begin talking about the first, uh, computer run characters. And one of them is smart one, a, uh, character that they have that is run by Sperry Univac, one of the first big computer companies. And my grandfather, uh, worked at Sperry. It was always described to me as Sperry gyroscope, but that was, you know, the, the version of the company that existed before. But I texted my mom and then my uncle about exactly what my grandfather did. And this is what I got back. My grandpa, Tony was an electrical technician doing the work, uh, of an electrical engineer, but without the college degree, he wasn't eligible for a more, more prestigious title. Cause he didn't go to school. He worked on what was called the high power room. It contained one of the first computers then made, uh, made up of God knows how many mechanical switches and they made a racket. Uh, that was the beginning of his love, uh, for computer and interest in computers. And that was the first person that got me into computers was my grandfather who had an Apple II GS, who loved, uh, uh, computers his entire life. And it was really a crazy moment to think I mean, my grandfather worked there in the early 60s So by the time that it's like the late 70s, early 80s That I think Smart One rolls out in Disney But certainly he was at the dawn of kind of the Univac supercomputer revolution And it was cool, it was cool to see that connected Andrew Mayne: Where was he at that time, where was he located? Long Island, at their Long Island facility Yeah, you know, my grandfather was out in San Jose around that period Working with, I don't know if he actually worked with Spartan and some other ones It's a crazy time to think about that period and what happened early on I have not watched this, I'm embarrassed I'm a huge Defunctland fan, and fun fact, I backed a movie by some of the people Who work on Defunctland that's going to be in theaters this summer It ends by Alex Olem, a couple of his producers worked on the Defunctland stuff That's awesome Justin Robert Young: Yeah, it's, you know, there's something about Defunctland of like The style of Defunctland kind of breaks a lot of the rules That I have for content, for infotainment-y kind of content Because you on one hand don't want to just have everything be a Wikipedia page Just this plodding linear thing And then this happened, and then this happened It just sounds like it just reads like an Old Testament book Like, and then Jada bagat gala, and gala bagat bala, right? But yet he does it, and it works? It's like, I don't know whether it's Kevin Perger's voice Or just the subject matter is light and interesting enough Or it just is something that I'm really into That I'm like, yeah, literally, just read me the phone book of like And then Walt bought the this computer And then they bought the this computer Which was interesting because it used the this tape And it's like, I don't know what it is It crushes Andrew Mayne: Four hours, they're like, zero F's to give We're going to make a four-hour documentary on this topic And I watched the whole FastPass thing, whatever Riveting, great, real research We're going to do research, all sorts of stuff You're like, this is great This is not low effort, this is passion And this is why I just don't watch TV I don't watch YouTube because there's going to be some YouTubers out there Justin Robert Young: I saw that four-hour runtime And I'm like, you know, there's been a lot of conversation About what's next with movie theaters And you see all these moments of music stars Like Sabrina Carpenter and Taylor Swift And people that are like, we'll have these big, gigantic nights They'll roll out their albums And it'll be, go see it in a movie theater With a bunch of fans, big moment Movie theaters love it I'm like, if Defunctland said Hey, our next YouTube video We're going to launch it on YouTube But also, it's going to be in theaters Brian Brushwood: I would so go Like, would you go see it? And I'm like, I think I might Like, it'd have to be the right one You'd have to have one that would be like No, this is going to be awesome It's going to be a special cut above But I don't think it's crazy to say Like, yeah, would you go see a two-and-a-half-hour Defunctland, would you go see a four-hour Defunctland video In a movie theater? I don't know, maybe Yeah, I would Hey, I got a pick for you Andrew, are you familiar with the full experience That we do over on the Cord Killers podcast? Brian Brushwood: Oh, this is great No Okay, so the full experience For a while, we had Cord Killers Which is, of course, about the news of the cord-cutting revolution People who want to watch their stuff on Netflix Hulu, Amazon Prime, etc But then we did a watch companion piece That we called We used to call Spoiler in Time But we would do entire seasons of things But then people kept just suggesting You know, your Babylon 5s You're like, oh my gosh This is a multi-year commitment We can't And then we came up with the full experience Where, you know what? We're going to see if we pretty much get what you love about this show By watching the first The ultimate The highest rated episode The lowest rated episode And the last episode Does the full experience tell the full story? And it's been great Because some of them we've seen before Others, it's like I'm never going to watch Bonanza But I'll give it the full experience And see what I feel about it And then in this case We ran across one That I've only seen the first episode of But boy, oh boy Do I love it Is it great? The production value is incredible I re-watched the pilot twice It's in the sweet spot Where it was shot on film So it's very high resolution Even though it was intended For just standard definition broadcast television And that's Magnum P.I. That show has it all The original It is great The pacing It exactly hits all the archetypes It is peak 1980 I'm loving the experience Go watch the pilot of Magnum P.I. Andrew Mayne: There is a channel I want you to watch Let me get the name right Because They go into The history Of a lot of these shows And they just had one A while ago On I want to say I watched I watched one on Magnum P.I. It's all about the history of Magnum P.I. How it came together The production, etc. So Double Reel Originals Double Reel TV Let me see if it's that one Double Reel TV That's worth checking out So Double Reel TV This guy does these deep dives Into TV shows And He goes back into Like That So let's see we got Matt Houston Heart to Heart Hunter Riptide Simon and Simon Wise Guy Ambiguous: Baywatch Presco County Andrew Mayne: So anyhow Yeah Magnum P.I. He did that one year ago The history of Magnum P.I. The shirts The Ferrari And everything in between Oh It's really worth checking out Andrew Mayne: Really cool So that's Double Reel TV And I'll tell you something I watched by the way I was thinking about And it was I remember seeing It was a TV movie When I was a kid And I'm like Whatever happened to that Brian Do you remember Hangar 18? Wait a minute Supernatural Is that right? It was These astronauts Are on a mission in space On a shuttle And they launch a satellite And it hits something And something crashes in the desert And it's a spaceship And it gets taken into Hangar 18 And the astronauts Are blamed for a mishap Because somebody died And so They kind of have to like It's kind of like Capricorn 1 meets Close Encounters Sort of thing I guess But like They basically Gotta go find You know They're trying to Clear their name And find out Where the spaceship landed Meanwhile they're exploring We get the scientists Exploring the spaceship You know It keeps cutting back To Robert Vaughn And some other guys In the room Plotting stuff As a kid I remember it being Kind of quirky And cool And interesting It's not a great movie But I watched the whole thing Yeah Brian Brushwood: It looks like it's Yeah It's just on YouTube That's amazing Very cool While we're talking about Hey did you see Hey Brian Did you see The Street Fighter trailer I have not I saw that the official Full trailer came out Is it good I think it Rules I think it's Awesome Yeah It Alright Let me pitch something To you Brian Alright A Street Fighter movie With a Hung in cheek Wink wink Kind of Super Balls to the walls Energy And all of the action Looks like everything Everywhere all at once Oh I'm in Is that what it is That's what it is Hells yeah 100% Hells yeah That's great I may have to bring back The spikes for it And it's a period piece In 1993 Oh is it really Get out Alright Alright Alright I wanna watch it But I don't wanna We got demonetized Last time we watched A trailer Gotcha Alright That end Guys They showed a Doomsday trailer At CinemaCon This is for Theater owners In Vegas The description Of what it is Is out And it sounds Awesome Yeah The reaction In the room Was apparently So intense They played the trailer Twice That's amazing I wonder if we can get Give me a Play by play Description Of the Doomsday trailer That just dropped There we go We'll see if a robot Can give us a I would tell it To read Reddit Reddit has the The good play by plays But Brian Brushwood: It involves Here I'll just I'll say this It's better that you Don't read it I'll just give you One thing Alright I'll give you one thing Just to wet your whistle Shang-Chi Versus Gambit Okay Just gonna throw that That's a little flash There's Characters Returning The first time You see and hear Victor Von Doom Brian Brushwood: It's I don't know It apparently ruled I'm in Just whenever That's a thing That thing was Ambiguous: Strange I made Assim…? Ambiguous: It's just kind of awesome. When Marvel's at its best, especially with the big team-up movies, it's just like, how are you going to see these characters interact? What characters, what things have you never seen before that would be really awesome to see? And it's like, man, it feels like they're back in good hands. These are guys that, they know how to make a Subway sandwich. Brian Brushwood: Yeah. Right on. All right. How's it been, Maine? It's been weird.