Subscribe
Share
Share
Embed
Explore your curiosity - Interesting people with fascinating stories.
Life consists of three things:
-The stories we tell others
-The stories others tell us
-The stories we tell ourselves.
Jacob, welcome back to the show. So happy to have you here again.
Speaker 2:Happy to be back, Ari. Good to see you.
Speaker 1:It's great to have you back. We were just we just had a short chat. We're talking about gaming and the old games from the nineties. There was something special about them, and you were starting to say about how they were special from a storytelling perspective. Tell me more.
Speaker 2:Yeah. Right? So back back in our day, you know, those kinds of games, like in the eighties and the nineties, they had to be more content driven. Right? Because they weren't heavy on the graphics.
Speaker 2:Right? They weren't eye candy. They were mental candy. Right? You know?
Speaker 2:They brought us in, you know, engaged the story. The characters actually meant something, you know, and like, yeah. And you wanted to play it because you wanted to find out what was gonna happen next with your characters, you know? And when they died, your character dies. You're like, you know, because you had invested so much time into that thing, you know, you're like,
Speaker 1:oh. But isn't isn't there's there's some magical beauty in the fact that actually the constraints, right, there were constraints on graphic. You were playing like just with your mouse point and click, we didn't have controllers back in the day, it was very basic and would you agree that in certain way those constraints forced it to be better on content, on storytelling?
Speaker 2:Absolutely, Absolutely. You know, it kind of goes, goes back to how do you make the best software people, right? You give them the slowest computer possible because then they have to make it really efficient. Right. That's the best code ever.
Speaker 1:That's brilliant. Would you agree that we've lost something with the new games in the storytelling and these kind of elaborate characters that you've and I'll put indie games a little bit aside because I'm seeing that in indie games but in kind of your mainstream games it's up for you know the really good ones like Witcher and like Zero Down we've lost a little bit of that storytelling.
Speaker 2:I think so right you know the real story that they're trying to tell now is all about commerce right? Get you to the store, buy some stuff, you know, to get a better experience. Right? That's kinda like where they're moving towards, you know, and I can understand why because games are really expensive. They cost more than making a movie.
Speaker 2:Right? Very expensive.
Speaker 1:Right. It's like, let's, you know, let's get $3.04, $500 out of you, $1 at a time. What a business Right? You know what, personally games back in the day or the good games or even today like a zero down, you pay $40.50 dollars and that's it. Like you get the game, you're done.
Speaker 1:I would rather pay the $50 for a really good game than you know have them you know leech a dollar at a time out of me quite honestly.
Speaker 2:I agree you know, I don't have a problem paying a $100 for a game you know, knowing that it's a Bethesda game, it's gonna, you know, they spent, you know, ten, twelve years on that game. It's gonna be really good. Great experience. Right? But then, you know, they tack on the whole store and the commerce, you know, and they nag you for money, money, money.
Speaker 2:You're
Speaker 1:like Yeah. Yeah.
Speaker 2:You know?
Speaker 1:It's almost an unfair advantage. Right. You don't feel like you're at par. What's, is there a game that you're kind of waiting for that you're excited about that's coming out soon or not so soon?
Speaker 2:So let's see. I was waiting for Battlefield six cause I really liked the Battlefield series. Right? They're very authentic and very real. Right?
Speaker 2:And the new battle Battlefield six is very close quarter generated stuff. Right? So it's back to the old CQ days of like Battlefield four, you know? So everyone's got shotguns and they're just running out of shotguns. Right?
Speaker 2:Right. Right. Which is a lot of fun. Right? It's very cathartic.
Speaker 2:But on the content side, you know, what is it? Gosh. It escapes me now, but there was a new game I was waiting for. Outer Limits. That's it.
Speaker 2:Right? So Outer Limits two came out. Right?
Speaker 1:Right.
Speaker 2:So Obsidian makes really neat games that have a lot of good storyline content. Right? And I just played about. Right? And I played the whole thing, you know, and it got kinda weird at the end.
Speaker 2:But I think it's because it was a bit long for them. Right? They had never really made a game that long. And so there's a certain amount of talent that you have to have in the storytelling side to make a game that long. Right?
Speaker 2:So I'm not saying they don't have the talent. It's just that it also takes time. Yeah. And a little practice, which I don't have. I'm not a storyteller.
Speaker 1:I think that's so true. The Witcher, for example, they're in this model of, you know, the big story and then you have the side quests kind of philosophy. And I had a friend I think The Witcher is one of the best games you know made in the last ten years but my friend who played it he got lost he was like I don't know what's happening I don't know what I'm doing next he kind of got lost in the side quests so there's definitely a balance and a complexity here to make the games longer, but also keep them engaging where you're in the story. So I think that's a great point. Obviously the hot topic nowadays is AI.
Speaker 1:We're seeing it changing industries, changing how software is built. You had a game company. It's really rewriting how software is built and it doesn't seem to work quite yet. The agents are working but not well and it seems that you need to kind of master orchestrate the agents together to kind of get the good output And we talked about a few techniques that that you're using. Let's start there.
Speaker 1:Where are the agents breaking? Like, why don't they work?
Speaker 2:Yeah. So I haven't used your orchestrated agents, right? Like at Cursor and stuff like that, you know, of, you know, because I haven't really I haven't really seen good content come out. Right. I've seen like, so for instance, I'll go to GPT-five, you know, GPT-four in the past, you know, and I will ask him a question about writing code and it'll give me an answer that was mostly correct.
Speaker 2:Right? There are times it gives me answers about an API that's not public, but it knows about it because it was able to get that code from somebody else. Right? So like, that's always fun to see that. But, you know, where I see that these coding agents where these chat chatbots fail is that they they really can't really can't capture complex software systems.
Speaker 2:Right? The whole context side of it is is not big enough yet, or it's just not sophisticated enough yet for them to capture the whole essence of a very large enterprise level software. Right? The small stuff in the algorithm, you know, things like that, you know, I can get it to do DP coding all day long. It's great at that, you know, love it.
Speaker 2:But you know, the big architecture stuff, you know, it just like fails almost every time.
Speaker 1:Yeah. That, that is so true. You wanted to, to do a, to work with an API. It's going to take you hours to learn that API, dozens of tests to test out the ambiguity in the API documentation that we all know is there. But you use cursor to create just one API interface and sure it will fail six times until it gets it right, but it will do that in three minutes.
Speaker 1:And then you've got this API that just works and it knows that it works because it's looking at the logs. So there's something brilliant about it, but scaling that to full fledged architectures, there's a challenge there that I think that these companies will figure it out, but they're definitely not there yet.
Speaker 2:Hey Ari, do you remember in the 90s when aspect oriented programming was a really big thing?
Speaker 1:What was aspect oriented? Was that the was that small talk or which one was that?
Speaker 2:No, no, no. So the idea was that you got some libraries, right? So you start with the premise that all the code that you ever need has already been written. Right. Aspect.
Speaker 2:And so now what you want to do is you want to take that library and just get a piece of it, an aspect of it. And then you get another piece of a library and you chain them together and now you got a program. Aspect oriented program. Right? It was a really hot thing for like 1997, '98, right in there, you know?
Speaker 2:Then it kind of died off. But the thing is, is like what we're doing today with like interacting with APIs is very much like Aspect drawing on a program. Right? Very much like it. Right.
Speaker 2:So, you know, the thing that killed Aspect is that, you know, developers write code differently. Right? Right. Different interface and stuff. But as chat starts to write all the code in the universe, guess what?
Speaker 2:It's all gonna be written by the same developer. So you can resurrect the idea of aspect oriented program. We can just like pull pieces from other projects, other runtimes and use it.
Speaker 1:Brilliant. I love that. I love that analogy.
Speaker 2:Fast
Speaker 1:forward, and we don't know if this is, you know, six months here or two years or next mark tomorrow, but fast forward where agents are now have this capability to architect, not just to write a code snippet and you know if you're going to be the architect and the orchestrator and you orchestrate different agents to do it correctly, it can be done today and that's what I'm doing, but the agents can't architect on their own. If we fast forward to this, this is a paradigm shift. What are the software teams of the future going to look like? You're no longer going to have a backend front end developer like that won't make sense in this future. What are software engineers gonna learn?
Speaker 1:What are engineers gonna you know, engineering schools gonna teach? How do you see the the the next, I don't know, five, ten years, you know, panning out?
Speaker 2:Yeah. That's a that's a fun topic. Right? I really love that topic because I talk about it a lot. So I do some experiments just to see, like, what would an engineer look like in the future.
Speaker 2:Right? So Yeah. I would go get, an excerpt from some marketing website and just dump it in the chat and say, make me a product that does this, And it does it. Right? And it's like 60%, 50% accurate.
Speaker 2:Right? Whatever. I did that to one of Andrew Ng's things. Right? You know, he had this thing for doing PDFs.
Speaker 2:Right? So I, you know, he had a marketing page. I took that marketing page, dumped into GPT, and then I dumped into Quen, and Quen gave me a solution. And surprisingly, it worked. I mean, it was like 80%.
Speaker 2:Like, wow. Okay. That's a little scary, but it's cool. Right? So I think the future is gonna be more about model based stuff.
Speaker 2:Right? You're gonna be like, I'm gonna do the thing of like, I it's like Star Trek, you know, computer. I have a I need to solve this this problem. We need to do this algorithm to do this, and the computer and does the thing. Right?
Speaker 2:And it gives you a solution. Right? Like, like you're doing, it's gonna take more than just one agent. I can't just go to GPT-five and say, give me this. Right?
Speaker 2:GPT will give me something, which I know for a fact, they'll give me something that's wrong. Right? And I can iterate on it, but it'll just do a circle of wrong. It's done that to me so many times.
Speaker 1:I mean, the day that GPT is going to say, you're going to have one one quote unquote agent and it's going to say, hey, this task, I'm going to split it up into 10 different agents. Here's the modules that each agent will work on. Here's how they will integrate. You know, this is going to, because I'm using 10 agents, you have to have the super expensive license. Boom.
Speaker 1:And it's just going to work like that day is coming. It's going to have an experience of one agent that it's going to spawn off maybe a 100 agents because you know, you know for one task it's going to try three different ways so that's three different agents working in parallel and then it will pick the best or or merge them or whatever. So I know that day is coming but we're not there yet. But I completely agree with your hypothesis. There's something really interesting that you said.
Speaker 1:The world of business is changing, right? If you're able to go and basically duplicate a concept or a business model from a code perspective, what that means is software is no longer a barrier to entrance because you know what took old days, old school, you know, three years to do, maybe now you can do it in month or two. So software IP almost goes away from the world. So, I mean, this is a massive question, right? Like, are we going to start fighting back that, you know, AI actually is IP and the ideas are IP?
Speaker 1:Or is that a lost battle and, you know, whoever tries it is just going to lose? Or do we just need to adjust to the world where your software or your quote unquote intellectual property is just not a business barrier, you need to find other ways to hang on to your customers.
Speaker 2:Yeah. The the whole business argument is a really tough one. Right? Because there's the whole technology side, which we have solved. We know what to do.
Speaker 2:We can do all that kind of stuff. Like like you're saying, there's the intellectual property side, then there's the business side, which is how you capture people and money and revenue and support people and do all that stuff. Right? And so there's a lot of work in AI to try to automate, you know, some pieces of that. Right?
Speaker 2:That's very interesting. And now there's the part that's automating the technology side of it. Right? And like you, like you were saying, like I said, we could just take someone's idea and dump it into the technology side and then shazam, we've got a product. It's not perfect.
Speaker 2:So you gotta have some people still to make it perfect. Right? So that the business people can go and actually make a business, make it useful. So there's, I think there's a big shift coming in like how resources will prioritize. Right?
Speaker 2:So it used to be like you, you and I came from a world of like massive stacks of developers. You know, this factory of people churning out code. That's done. I can't say that enough times. Right.
Speaker 2:That is done. In ten years from now, that's not even gonna exist. They're gonna laugh at us. So you did what? You know, no, nobody does that.
Speaker 2:Right. And rightfully so. Right. We are seeing automation now on the intellectual side. Right.
Speaker 2:And we've seen automation on the business side. Right. Which is even better. Right. Cause there's a lot of mundane stuff, but then you've got the real problem to solve now is the people problem.
Speaker 2:You know, like, because if the people aren't doing stuff, they're not going to need the stuff that the business is trying to make. And so now we've got this catch 22 that we have to solve as well. Right. And that's a real problem to solve.
Speaker 1:So really we almost need to go back to the old school kind of philosophies of barrier to entry. How do you keep your business safe? What's keeping your customers to move to a different service considering that that different services software is probably going to be, you know, a lot of competition. It can improve really quickly because they're also prompt engineering software. So in a world where that doesn't exist, I don't have to fall back to those old school business principles, right?
Speaker 1:Locking of vendors, contracts,
Speaker 2:you know,
Speaker 1:social reputation, all those things. We might see a resurgence in old school marketing and branding
Speaker 2:stronger than ever before. I think so. I think, I think you're right. I think we're gonna, we're gonna have to see that. Right?
Speaker 2:For sure. Until the, until the notions of business start to change. Right. You know, the, the current way of how we do business isn't going to convey into the future. Right.
Speaker 2:It's going to have to change. How that's going to change. I really don't know. Right. Cause you know, the whole commerce model is going to have to change quite a bit, right?
Speaker 2:Because the whole value proposition is changing, right? The technology as value is going away. People as value used to be in the past and all that kind of went away. So that's gotta come back. That's where we get a lot of our value today is from people, not so much technology, you know?
Speaker 2:But then I don't know what the future is going to look like. That's the hard part is like, where's the mix between those two that keeps them happy.
Speaker 1:Right? Well, problem is also I think we couldn't make this point strongly enough is that when we said the future ten years ago, it was ten, twenty years away. The future today is tomorrow. It's the next time the chatty pity, Olama, Anthropic release a version that does something that we just we couldn't imagine. So it's you know the whole time scales is just completely out of whack.
Speaker 1:The future is and sometimes the future is right here you just haven't heard about it yet right, it's not evenly distributed. It's a quote I think it was Gladwell. So we're in this time of incredible ambiguity, right? But we're running software businesses. Who do you hire?
Speaker 2:Right. That's the hardest part, right?
Speaker 1:Is it the CS grads?
Speaker 2:Well, absolutely. You know, we have to do them right. Right? Because we told them to go to school and get your degree in computer science, right, knowing eight years ago that this was happening. We still told them to go to get your CS degree.
Speaker 2:So we gotta hire them, get them in here, teach them how to use this stuff, teach them how to use LLMs, and teach them how to code and create not so much code, but we'll call it creating software systems using AI. That's what they need to learn how to do. Right? We know how to do it. They don't.
Speaker 2:We have to train them how to do it so that they can replace us so we can get out of the way. Right? Because that's their future, not ours. Right? Ours is kind of done.
Speaker 2:What mine is anyway. I'm a lot younger than you or older than you. So
Speaker 1:A little bit. Yeah.
Speaker 2:Mine's pretty much done. Yours isn't yet. Theirs is just starting, and they're the ones that need to learn. So we gotta do that for them. And then on their side, who are they going to hire?
Speaker 2:I don't think they're going to be hiring computer science people anymore. I think that's kind of a dead end. I think that's going to be more about like AI engineers, concept people, product people, people who can conceptualize useful products and then communicate with an AI to make that product. Right? So the, the computer engineers that,
Speaker 1:and we've already heard the story, you know, of this lady who finished, you know, CS degree and then went to work at Walmart, just couldn't find a job. And you know, after three or four years of studying, that's horrific. Now obviously the universities are gonna need to rewrite their agendas and syllabuses, that's gonna have to happen. But for those engineers and until the universities catch up and they most love me, what are the computer science engineers that are fresh, how do they kind of jump into this new role and make themselves valuable?
Speaker 2:Yeah. Right. So that's kind of on us. Right? So bringing them, I have to bring them on as interns and just show them how to do stuff.
Speaker 2:Right? Give them, give them tools. Right? Give them toys to play with. Right?
Speaker 2:Cause they're relearning stuff from scratch. So they have the toys.
Speaker 1:What would you do before you've kind of found that internship, know, how do you kind of showcase yourself to be like you know, I'm the one you want to hire because you've got all these CS graduates that are thinking like oh my god there's there's all these other graduates, I want to get ahead of the curve. So before we, the kind of, you know, businesses give them that opportunity, what can they do?
Speaker 2:Yeah. What can a computer science graduate do today to make themselves stand out? Right. Absolutely. Yeah.
Speaker 2:That's, that's really hard. There's a lot of them, right? It's a, it's a really big ocean, you know, and it's hard to be seen, you know, in that ocean. So we used to tell them to go get your GitHub, you know, and create some open source stuff, you know, and show people that you do, you know, you know how to do stuff. Right.
Speaker 2:And that's, that's all fine. Right. I mean, you can go and do that. Right. You can go and say that you created these things.
Speaker 2:And they can go on your LinkedIn or whatever, and have a blog and start a podcast and do all of these things. Right. To try to do the whole social side, kind of market yourself. Right. That requires a lot of time.
Speaker 2:Right? And their time should be spent learning how to use these tools. So they should document how they're learning all these tools. Right? So that's one way of doing it.
Speaker 2:But if everyone's documenting how they're learning the tools, then no one's gonna watch them as they're documenting. Right? And so you're like, well, no one's going to watch it anyway, so no one's going to know about it. And the hiring people don't have time to really watch a bunch of, you know, blog posts and YouTube channels and stuff like that.
Speaker 1:Let me offer a scary idea. Tell me what you think. I told this to my kids and as I told this to my kids I was kind of shocking myself. There is a world or a future, I don't know if this is going to be true, where actually your reputation as a professional will be based on your social media followers because if you're creating high value educational content about whatever it is, AI or you know agentic programming or math. If you have a strong fellowship that indicates to the hiring manager that you are an authority.
Speaker 1:So are we maybe going into a mad world where your social media followers helped you get a job? And I'll give you one example which is maybe a little bit more straightforward. You know how authors are becoming authors today? The traditional book publishers are going to influencers who already have an audience and they're saying, oh, why didn't you write a
Speaker 2:book? I
Speaker 1:mean, that's that's disheartening in many ways, but on the other hand, I mean, we've gotta get with the get with the times. I mean, this is Yeah. It seems like the future is going in a weird direction.
Speaker 2:You know, it is in a way, but I have some Gen Z kids. Right?
Speaker 1:Yeah.
Speaker 2:They're just graduating from high school and they don't do anything social. In fact, they're very, mostly antisocial. Mhmm. Right? They see the whole social media side of life as very not authentic.
Speaker 2:Right? You see it as, you know, people making stuff up, just trying to sell things. And rightfully so. That that's kind of the whole point of it is to sell themselves, sell, you know, the influencer sells themselves.
Speaker 1:Right.
Speaker 2:That's what they're trying to sell. And so I do agree with you Ari, that you're going to have to have a social imprint in the future. It's just like how, how that's going to be created, you know, that has to be authentic. Right? That's the real key is that, you know, you can't be a social influencer and be an authority.
Speaker 2:Right? I mean, they're kind of orthogonal. And so, you know, you have to adopt that authority kind of methodology and mentality. And then you gotta find a way to prove that you're an authority. Right?
Speaker 2:In the sea of many authorities, including AI, right? AI's out there as the authority now, right? Mean, AI's know pretty much everything about everything. Right?
Speaker 1:I mean, that's, say such a great, I mean, you know, if we forget the answer and just, you know, stay in the question for a moment, that's the question that all of the youngsters need to ask themselves, how do you prove your authority? I said well maybe it's through followers and the fact that you have an audience but there's probably other ways to prove your authority but I think if everyone in our audience who's young or mid career just asks themselves that question, how do I prove my authority? How do I show that I am one? I think that's probably going to be the key into the future.
Speaker 2:Yeah, it is. I think so, unfortunately, right? Cause that means they have to focus on that. Right? They got to like foster that identity.
Speaker 2:They got to spend a lot of time doing that. And we want people to spend their time on knowledge, being good at something. Right. And, you know, being experts, not being socially great. And it's just, yeah, I don't know.
Speaker 1:I love that you bring up that pushback because these, all these memes on LinkedIn about like a real professional and then a LinkedIn professional who's just tweeting about it, but doesn't really know how to do this stuff. So, you know, obviously, my answers are not they're not perfect, and I I love your pushback. That's it's absolutely wonderful.
Speaker 2:Yeah. Well, there is no perfect. Right? I mean, the real answer is that it depends on where you're at at any given moment in time. Right?
Speaker 2:You're hiking up a mountain. Right? And there is no answer to get up that mountain. You just have to one step at a time look at where you're going and don't trip on a rock. That's it.
Speaker 2:I
Speaker 1:love that. Jacob, what an absolute pleasure. It's so much fun to have you back on the show One last final piece of advice to yourself, our audience, to the new CS grads, what would you say?
Speaker 2:Learn as much as much as you can about AI, right? Try to try to set stuff up at home, try to learn how to set it up at home because that's the future. People want you to be able to do that at home or in the office. Yeah. And just embrace it.
Speaker 2:Don't be afraid of it. Embrace it. Learn how to control it.
Speaker 1:Yeah. Failure is just a stepping stone to the next step.
Speaker 2:Jacob, thank you so much for joining the show. I appreciate you. Thank you, Ari.