Subscribe
Share
Share
Embed
NET Society is unraveling the latest in digital art, crypto, AI, and tech. Join us for fresh insights and bold perspectives as we tap into wild, thought-provoking conversations. By: Derek Edwards (glitch marfa / collab+currency), Chris Furlong (starholder, LAO + Flamingo DAO), and Aaaron Wright & Priyanka Desai (Tribute Labs)
00;00;00;00 - 00;00;18;06
Aaron
Chris, I'm going to start upfront engineering meet up in New York.
00;00;18;08 - 00;00;25;13
Chris
No kidding. I, I felt like you had this in percolating. Tell us more what's going on here?
00;00;25;16 - 00;00;46;05
Aaron
I don't know, I just feel like the way software and, like, all real professional work is, is changing, and people don't know how to use these systems. And every day, I kind of, like, scan through Twitter, and I'm just frustrated by all the folks that seem to be talking about prompt engineering. It just doesn't seem like they fully understand how to use these systems yet.
00;00;46;06 - 00;01;05;24
Aaron
I don't think anybody does, to be honest, but just it's not like aligning with how I've been using all these AI systems and what I find effective. So I feel like people need to like, talk shop related to it and like, understand what works better. And I'm not 100% convinced that engineers or software engineers really know how to use these systems very well.
00;01;05;27 - 00;01;09;28
Aaron
So I think that's kind of the animating thesis related to it.
00;01;10;00 - 00;01;32;10
Chris
Yeah, I can see that. And I do think it's something like I'm curious to talk with other people who, you know, use these heavily and see how they approach it. Because I've been doing this I think works into Midjourney at this point. So two and a half, three years in and I still don't have a like I have my personal style, I have what works for me, but I have no clue what best practices are here.
00;01;32;10 - 00;01;53;05
Chris
And how much like, I like to take, an approach of, like, using terminology that's not going to lock the LLM to the mid. I like, try to like keep it around the edges. But then my, my technical jargon in terms of like acting like a director or framing things, you know, like the mechanics of production are really, really weak.
00;01;53;05 - 00;01;58;23
Chris
And so I do think I'd love to like, be part of those discussions and kind of understand the bounds between those things.
00;01;58;28 - 00;02;24;23
Aaron
Yeah, that's going to be one of the biggest stories in like 2025, 2026. It just like how these AI systems and the efficiency gains you get are just going to dramatically accelerate. I think we're seeing it already with software, but nobody knows how to use these systems. And I think there's a lot of bravado just around, oh, I know how to use it the best, but I just I really think it's nobody knows the engineers at developed that don't know it, you know like nobody does.
00;02;24;24 - 00;03;02;10
Derek
So I would love to join, when I'm in the New York area, Aaron, for your, coffee meetup around Prompt engineering. I think there's a very clear advantage right now for those who do know how to prompt these things and who are there, are there are best practices very clearly emerging to drive better outputs. My grenade I want to throw into this conversation is like, is there an argument to be made that in large part the the idea that humans are doing most of, like the narrow prompting in the future gets overridden by AI models or AI agents themselves, becoming increasingly capable of generating far more effective prompts.
00;03;02;11 - 00;03;20;15
Derek
And I don't know exactly what that looks like, but it does feel like a bit of a counterweight to the idea that AI as a human, that my my one last mode and the age of synthetic intelligence is that I know how not know what words to use to like derive the best payload at that. That two feels like something that will get disrupted.
00;03;20;15 - 00;03;22;16
Derek
But I'm curious how you guys are thinking about that right now.
00;03;22;21 - 00;03;45;01
Aaron
I think that's right. And I think that's why this question around, like, how do you actually prompt these systems, like what is the most effective way to do it? Is important. So if you believe in like a fully autonomous agen tech future, you know, powered by this age of synthetic intelligence, Eric, we have to understand that these systems work.
00;03;45;01 - 00;04;06;07
Aaron
My sense is, is that it's a little bit too self-referential. And Chris, I don't know if you've seen this, but when you ask one of these systems to generate a prompt, they they tend to stink. They're like overly detailed, too verbose, like they're not playing around with the edges to like, really unlock, like higher quality content and also higher quality outputs.
00;04;06;09 - 00;04;08;05
Aaron
But I'm curious what you've seen there, Chris.
00;04;08;08 - 00;04;29;02
Chris
Yeah, look at LMS right there, chat bots that want to do a good job and they want to do it quickly. And so I think there is a drive towards efficiency when they perform their work. And that drive towards efficiency takes them to the mid and it takes them into uncreative places because that's not what they're they're indexed for.
00;04;29;04 - 00;04;54;05
Chris
And so certainly like you do need some more creativity. And like my answer to this question is going to be like my answer almost everything. It's both right. Like the same way, you know, in our financial system you have algorithmic traders that the machine does all the work. And then you also have managed funds where there's a group of professional DJs who are making the decision to pull the trigger and both have their strengths and weaknesses.
00;04;54;06 - 00;05;14;08
Chris
We're going to see that in the prompt world, right? Like Derek, the machines aren't going to say, you know what? You're not allowed to prompt anymore because we can do this just fine. But it will take away some of the specialness of it. Right? Or you just need to, you need to keep climbing on top of whatever gets commodified.
00;05;14;16 - 00;05;32;06
Chris
And so you might just find yourself always going to higher and higher grounds. Right. Like, I don't want to, like, get off the cop thing, but like, world models are the hot new frontier at the moment. I've seen a decent amount of press coming out around world models, and there was one, that recently released a GitHub repo.
00;05;32;06 - 00;05;49;28
Chris
And, you know, you can you can run it locally. I can't remember the name of that one either. But at a certain point, I guess what I'm saying is when the specialness of the prompting does become operationalized by the machines, that just means it's your job to go one level higher and and do the work. And machines are still having a hard time with.
00;05;50;05 - 00;06;09;10
Derek
Yeah, I see that for sure. I think we're also seeing this in, you know, someone paying very, very close attention to is like where dollars are being spent in terms of like the in the innovation stack for, for AI and I'm creation and it's very clear to me that were investments are moving away from like the pre-training in the training stage to the inference stage.
00;06;09;10 - 00;06;29;08
Derek
And I think there's a few reasons for that. One, because it's clear that, you know, hallucinations are still a problem. I think Aaron was flagging, you know, the, the, the real creative edges of like, a question's answer, aren't necessarily being explored. If we're just pulling procedurally an answer from an O. Just like the first one that comes up.
00;06;29;15 - 00;06;56;09
Derek
There's this, there's, you know, these innovations called a chain of thought prompting or like, inference time testing that instead of relying so much on like the 0 to 1 answer you get just by prompting an alarm, what happens when you start to define kind of like reasoning or like question after question after question, and then assigning probabilities that this is the right answer versus that is right answer to the answer to a natural language prompt.
00;06;56;11 - 00;07;15;09
Derek
And so like the, an example would be like the kind of the way I think and it's, I think it's the way others think. And this is why this argument resonates with me. But for example, when somebody asked me kind of a difficult question, very rarely as like the first thing that pops into my head, like what I would, I would assign like a high level of confidence that that's the answer.
00;07;15;11 - 00;07;45;00
Derek
Often, like, I'll ruminate and I'll think on it and I'll like explore different passageways and I'll ask myself questions about this thing, and then I'll take a different route and I'll think about it in a different way. And that's essentially what chain of thought, prompting and inference time testing is doing. It's basically offloading kind of the, the, the automatic answer at the, which you would normally get by deriving out of like a, you know, a normal LM to using compute to kind of think through the answer at the time of the natural language prompt the inference.
00;07;45;00 - 00;08;09;19
Derek
And I think that when I start to see that, when I start to see resources being spent at that layer of the stack, my inclination is that some of what is being defined right now by like, really great prompting inevitably gets absorbed into these machines. Being able to just think more critically and like, assign better probabilities and weights to the answers that they're giving because they're spending more time thinking about these answers.
00;08;09;19 - 00;08;25;04
Derek
And so anyway, I think it's I think it's both. I don't think, you know, I don't think that good prompting is going away anytime soon. And I don't think that like inference, time testing and chain of thought prompting is just going to completely absorb everything. But something you said there, Chris, which really resonates with me, is like, it's it's everything.
00;08;25;04 - 00;08;41;21
Derek
It's got to it's going to have to be optimize optimizations around the entire stack. And and for now, I do think that you guys are right. Like I think there is there is a little bit of a moat to be derived out of just being really good at containing a prompt, in the exact way that you want.
00;08;41;24 - 00;09;01;21
Aaron
Yeah, I think that I think that there's a moat around that. And then how you construct those workflows. Derek, like, I, I think the, the challenge with some of the chain of thought reasoning that you see with like 001 or kind of it's progeny is it's hallucinating the steps to so you can give it more time. So maybe it's more accurate about that.
00;09;01;21 - 00;09;08;06
Aaron
But there's still like hallucination risk at each one of those steps. I'd have to imagine it feels that way.
00;09;08;07 - 00;09;21;26
Derek
Yeah. It's not totally complete. Yeah. And like I think we're seeing that play out. It's like these answers are I'm finding, oh one's good for certain things, not great for other things. So I think you're right. I think there is. The humans are safe for a little while longer at least.
00;09;21;29 - 00;09;36;25
Aaron
Well yeah. And I think I bet I think the power moves to the prompt are right. So like if, you know, like how to use these systems and I think a lot of the folks playing around with them the heaviest are engineers. And I think there's an assumption like it's going to operate the way an engineer's mind operates.
00;09;36;25 - 00;09;54;22
Aaron
And one things the way an engineer wants it. And I just think those are flawed assumptions. It's just a different beast. And it operates with different physics. It's just it's just a whole new design space. So it's kind of like if we asked engineers to become designers, some of them would be really good at that, but many of them would be.
00;09;54;25 - 00;10;16;01
Aaron
And we kind of know or not that great at that. And I think it's going to be the same thing here. And as like software is easier and easier to mechanically put together. Just the, the value of like that engineering mindset may just not be as valuable as somebody that can like orchestrate or architect and, and give these systems like the freedom they need to produce high quality work like moves from operations to like management more.
00;10;16;06 - 00;10;19;05
Chris
Aaron just coming out full team words. Hell today.
00;10;19;08 - 00;10;19;24
Aaron
Yeah. Sorry about.
00;10;19;24 - 00;10;43;12
Chris
That. No, no. It's cool. It's good when they. Hey, something that came up on a Lao call, on the Lao call this week is related and related to all this. And I haven't gone in deep on it, but it did feel really compelling. And the centered, you know, to this around this prompt discussion to, to some degree is, switching from inference to compute graphs and as a way of, of getting better results.
00;10;43;12 - 00;11;13;04
Chris
And like I said, I'm really high on the surface here. But a compute graph doesn't just evaluate an answer given by an AI, but it actually goes through the entire process of getting to that answer. And so it adds memory into the into, you know, like a layer of thinking and also a layer of context around, you know, what particular agent or you know, what set of parameters went into the formation of this output.
00;11;13;11 - 00;11;39;04
Chris
You know, instead of simply having a black box for an answer pops out, it's almost like you get an answer with an attached school of thought to it. And to me, that feels like it could be a really promising direction here in terms of expansion of what these things can do, because it allows you to filter and select, you know, for a particular line of reasoning or set of experiences when you want to.
00;11;39;08 - 00;11;47;12
Chris
It's almost like ask an expert or, you know, ask the part of this machine that's really good at doing this thing and have it block out all the other noise.
00;11;47;15 - 00;12;01;03
Derek
Yeah, I need to. So I'm not I'm not as familiar. Do you like, can you give like a real world example of like a kind of like a, sample prompt and like how that would work, in terms of, like, pinning down an answer or a reply.
00;12;01;05 - 00;12;22;02
Chris
Sure. So let's just use the the world of NFTs, right. And like, ask it a fuzzy question that requires specialist knowledge, maybe like, how would I frame this question? Right. Like you would frame a question where for the long term, the obvious answer would be coffee or snow, for when in fact the right answer is one, one, three, right?
00;12;22;02 - 00;12;40;10
Chris
And so it's almost like who what artist in the NFT world is, using the full capabilities of Ethereum as a world computer to create, you know, the, a network hyper structure? I mean, that's a leading question, right? But do you kind of get where I'm going here?
00;12;40;13 - 00;13;01;04
Derek
Yeah, I totally get where you're going with this. And like, I think in part this is what kind of like throwing compute at the inference layer and chain of thought reasoning. I think the, the one of the reasons why I like train of thought reasoning. So much is just and it's helpful from an analogy perspective as well, is I think once you can start kind of like quantifying these pathways.
00;13;01;04 - 00;13;19;26
Derek
And I do take Aaron's example, to heart, like I the brain of an engineer is going to be quite different than how these systems work. I do think the things that machines have an advantage on. And I'll get to your, your, your example here, Chris, the things that machines have an advantage on is that they can do things in parallel.
00;13;20;00 - 00;13;59;14
Derek
And so like while a single engineer or me, myself, Derek may only be able to kind of explore the probabilities with an answer like one at a time. We have found that, like, these machines can do it like hundreds, thousands, hundreds of thousands and assign probabilities to them, like in parallel. And this is I mean, this is largely like the I mean, if you guys have followed some of the, the AIS that have been trained on things like go or, or chess, that's functionally what they're doing, they're working their way strategically through probabilities associated with moves, and then assigning the highest probability response to the one move that they end up making.
00;13;59;18 - 00;14;20;00
Derek
I think in your example, Chris, what you're saying is like, right now we're being served a response back from an owl. That's like going to hallucinate. Or even worse, it can feel like the right answer. And like, you know, upon first glance of, first touch of a human's brain, like, can feel like. Yeah, I could argue that it's the answer to that.
00;14;20;00 - 00;14;34;00
Derek
As part of labs or it's or it's snuff row. In reality, it's like it's 113, and, you know, would be the incorrect answer. I think what we'll find is and it gets back to the I think the point that you guys were making earlier and I feel strongly about it, it's like it's a it's all of it.
00;14;34;00 - 00;14;54;08
Derek
It's going to we're going to have better training data. We're going to have better pre-training methods. We're going to have more effective training and more efficient training. We're going to have inference that like allows you to do reasoning on demand, in ways that kind of, allow probabilities to stack to, to serve up the correct answer.
00;14;54;10 - 00;15;21;11
Derek
And we're going to see optimizations on all of these things that get you the 113 answer at some point in time, I guess, I guess my, I mean, maybe my larger takeaway here is like, I don't know exactly where the the big innovations are going to come from it. Which part on the stack? My gut tells me right now that this that at the inference layer, it's the least explored in terms of thrown compute, and it's the least explored, from like a resource perspective.
00;15;21;13 - 00;15;41;27
Derek
And I feel like there's it's more efficient today for like the, the game companies or open researchers to explore that dimension than it is maybe some of these like well trotted dimensions of more data, more compute, a larger model that, seemingly is is not run its course, but it just doesn't feel as efficient to throw dollars at that layer right now.
00;15;42;06 - 00;15;45;01
Derek
But that's kind of how I'm doing things, right. I don't know if that resonates with you guys.
00;15;45;04 - 00;15;54;25
Aaron
It definitely does. Yeah. I mean, and it feels like that's the direction things are going. And I the only thing I'd add to that is kind of just memory, right? It seems like Microsoft's playing there. A lot of the.
00;15;54;25 - 00;15;56;01
Derek
Large, totally.
00;15;56;04 - 00;16;20;18
Aaron
AMD folks are playing there. I was saying this to kind of Chris separately, but in many ways, like each prompt, it's hallucinating and is in like a black box, right? And it's a little bit like a fetch. It doesn't have much much of a memory and any persistence between each request into it, providing that like consistent state and context across each one of the different requests you're making into these systems, I think will probably be super helpful.
00;16;20;24 - 00;16;42;16
Aaron
And so maybe it's just the combination of of that plus the point you, you and you were making Derek and you were making Chris, I do think that that's like the missing piece of the stack. It seems like that's where the large limbs are racing based and just. But Microsoft saying and kind of reading the tea leaves around what OpenAI is saying don't know as much about what anthropic is doing there.
00;16;42;16 - 00;17;02;03
Aaron
And I'm assuming Google's probably moving in that direction too. But I do think that's the one missing piece of like the crypto slash AI stack is, you know, some sort of like broad decentralized memory layer, like the idea that all of the outputs from that inference, Derek, would be controlled by one corporation is kind of concerning, at least to me.
00;17;02;06 - 00;17;04;11
Derek
Agreed. Yeah. Wholeheartedly agree.
00;17;04;18 - 00;17;11;22
Aaron
Yeah. I don't think we've seen that yet. And I don't know exactly what that looks like, but that's something I'm particularly interested in for the upcoming year.
00;17;11;24 - 00;17;32;04
Chris
We need a common wealth of knowledge. We need it sitting in a decentralized stack that's accessible to everyone. We need the group behind the hivemind. Well, we'll get there. While the prompt, though, still holds a special place. And since you started with with prompt meetups, let's, let's get a little wild here for a cigar. And, 2025 becomes a year of your year of the prompts.
00;17;32;07 - 00;17;46;12
Chris
Your prompt meetup becomes wildly popular. It spreads to other cities. Ultimately, it ends with a prompt Olympics. I want you to tell me what's in the prompt. Olympics. How do you measure the moment the best of the best in the world of prompting?
00;17;46;15 - 00;18;07;23
Aaron
I don't even know, right? I think I mean, that's what I think makes it difficult, right? Knowing you're judging the output at the same time, that's why I think you don't have this, like looping, this looping type stuff with the LMS yet, where they're evaluating their output and making adjustments to it at least. At the end user level, I think it just hard you kind of like, know it when you see it, Chris.
00;18;07;23 - 00;18;08;20
Aaron
A little bit. Right.
00;18;08;23 - 00;18;13;21
Chris
Until you say in the prompt, Olympics are going to look like figure skating. We're going to have the judges.
00;18;13;24 - 00;18;39;20
Aaron
Yeah, kind of. But it's also like hard, you know, like, what are you prompting for? Is it to produce great software? Is it great software for other software engineers? Is that even the right question to ask? Like if you're using these systems and you're generating software with AI, why are you optimizing it for humans? Right. The amount of software that's going to get generated is by AI systems, is only going to grow larger and exponentially larger than anything humans can create.
00;18;39;20 - 00;19;02;09
Aaron
So shouldn't you be prompting to make sure that the code is more easily accessible, modifiable, etc. for AI systems? And same thing for even written content or visual content as these things go, multimodal. Like maybe the types of information that these systems need is going to be different than what we want not to get. Heady.
00;19;02;12 - 00;19;28;26
Chris
No, I think you're in the right, right frame of mind. Yeah. To me, when I hear that, you're really just saying prompting is a language, and that language has many different ways of expression, and you can only view them in the context of what a language is trying to accomplish when applied at a specific problem. And so judging technical writing or as poetry versus code is or you can't write because they don't they don't equate.
00;19;28;26 - 00;19;38;18
Chris
And so therefore prompting just spans his entire gamut of, use cases. And what is the proper application of language towards a particular problem.
00;19;38;26 - 00;19;56;21
Aaron
Yeah, it's a tricky area, but it's super exciting. I mean, I think as we kind of peer towards 2025, I can't imagine that we're just not going to see some huge efficiency gains when it comes to software development and some more white, you know, white collar professional work over the next year. It just feels like you feel all that like energy bubbling around it.
00;19;56;21 - 00;20;00;16
Aaron
It's probably going to just get more widely disseminated in 2025.
00;20;00;18 - 00;20;21;16
Chris
We are at a point where, the response is in your ability to talk to these lessons is just so much better than it was a year ago, two years ago. And so we're already seeing this, right? Like the pain of talking to a GPT or Claude, like it's been vastly reduced. And we're we're getting a little greedy here, which is the right frame of mind.
00;20;21;16 - 00;20;26;10
Chris
Does this you know, we're barely getting started here, but like, they have gotten a million times better. Yeah.
00;20;26;10 - 00;20;54;27
Aaron
I mean, it's wild. Welcome to episode 11 of the Net Society podcast. You're here with me, Aaron, Derek and Chris. On Net Society, we explore lots of different things, including, digital art, NFTs, AI tech, crypto, the whole nine yards. Just a reminder, everything we say here, we're saying in our individual capacity and not in our capacity, in any of the organizations that we're part of.
00;20;54;29 - 00;21;02;19
Aaron
And nothing that we say should be construed in any way as investment advice. What else do you guys see coming for this upcoming year? By the way?
00;21;02;19 - 00;21;07;06
Derek
Should we do a little should a little NFT recap and some NFT productions?
00;21;07;12 - 00;21;08;13
Aaron
Yeah, let's do it.
00;21;08;16 - 00;21;11;22
Derek
Chris, you want to, you want to kick us off?
00;21;11;25 - 00;21;36;16
Chris
I mean, sure. So 2024, the big theme was necromancy. We brought NFT back from the dead. We've reanimated them. And so just as we're, like, zooming up across, you know, year by year look at things, the fact that people give a shit and the fact that there is an energy back in the space and that there are real stakes around the medium, is is the biggest win of all.
00;21;36;16 - 00;21;54;05
Chris
And so the NFT space is so big, it encompasses so many different scenes that are all trying to do different things. It's hard to like, just point your finger at any one thing. It's better to say, you know what? All of them have gotten off the the operating table, walked out the hospital, and are ready to live their lives again.
00;21;54;05 - 00;21;58;03
Chris
That that to me is like the number one thing we should be excited about.
00;21;58;05 - 00;22;16;23
Derek
What are some of the, I mean, like, you want to name names here, Chris? You want to name and shame any projects? Any, skeletons rising from the grave here that you think are worth keeping an eye on or ones that you think may, maybe getting animated but aren't quite, don't quite have the same bone structure that they once did.
00;22;16;23 - 00;22;38;15
Chris
Right. If we're talking about reanimate it and they're in zombie form versus their. We were once a life form. It's got to be the whole wing of hey, we were a successful TFP project back in 2021, and no one's giving a shit about us for three years. But now that we're having winds of change on the regulatory and the enforcement side, you know what?
00;22;38;15 - 00;23;01;28
Chris
We can detach a shit coin to our Jpeg. I think there's there's gradients of that. Right. Like you can point to judges who, you know, just dropped a token. You can say, look, this is just a project that's been hustling away. That never left is always pushing the roadmap and what they're accomplishing forward and dropping a token is, you know, just another expression of that.
00;23;01;28 - 00;23;25;25
Chris
Right. And I think that airdrops has been well received. Everyone excited about it. You know, there's a corresponding rise in, the floor values. Apologies. And then you've got stuff like Goblin Town, which people think is is launching a token. And part of the joke there is like, well, where the fuck is Goblin Town been? Like, I don't pay attention to hometown, so I'm not in a position to judge them.
00;23;25;25 - 00;23;45;13
Chris
But there's definitely like that spectrum, you know, happening right now. I mean, we also, you know, I've seen some big names who maybe weren't in the space for a while. Turnaround. Oh, right. I forgot about the NFTs. I mean, you know, Gary V's out in NFT land again. And so, you know, that's another way to look at this.
00;23;45;13 - 00;24;07;05
Chris
I think maybe the larger sort of filter to evaluate is really who's showing up every day and who is excited about this in good times and bad versus what are projects that are just simply opportunistic and only show off when it's, easy advantage for them to do so. So I danced around that a bit. Eric.
00;24;07;05 - 00;24;08;00
Derek
But yeah, I think I.
00;24;08;01 - 00;24;09;05
Chris
Just so I think in.
00;24;09;07 - 00;24;33;06
Derek
Yeah, just to kind of tee it up a bit, I mean, we've got for the first time in a while, three PvP projects over ten years. So we've got bored apes at 18 is pudgy, Penguins at three, 18 is zookeeper at 11.2 is. And then we've got, you know, a bunch of the, the 2021 PHP up, you know, folks coming back in and in with some here.
00;24;33;06 - 00;24;48;20
Derek
We got doodles at like seven and a half is Melody is at five. It's we've got the cool cats are, back over and he's to a couple of these, you know, sappy seals. I mean, I'm looking at these at these names here, and it's like, golly, I haven't seen some of these in a long time.
00;24;48;22 - 00;25;16;06
Derek
So it's. I think it's interesting. I will say, you know, the thing I always reflect on is during these, these, these wealth generation events as the, you know, ETH and Bitcoin and Solana and a lot of how, you know, the underlying currencies, by which these objects are denominated, continued to rise like people want to take profits and put them into these economies that are low, that are kind of like low in number.
00;25;16;07 - 00;25;42;03
Derek
There's only 10,000 punks, there's only 10,000 squiggles, there's only, you know, 10,000 zucchinis, and so on. And, the, the natural price discovery that happens when you have, demand side that's growing in a suffix supply asset, with that many marginal sellers is then the number also continues to go up. So what I think, you know, we'll, we'll probably continue to see a bit here is the numbers.
00;25;42;23 - 00;26;04;24
Derek
Like the core assets themselves continue to rise and I think that ends up meaning people get excited about NFTs again, and they go a little bit further out on the risk curve, and they meet new projects and they start joining discords and communities. And before long, you've got a full on NFT revival, reanimation, which is, you know, I was looking at some of my collection last night.
00;26;04;26 - 00;26;23;00
Derek
I was like, wow, I minted this project four years ago. And and it's like there's still a community here and there's still people excited about it. And there is something very durable about, like, these timestamped objects that communities form around that, that, it's fun to kind of see, reignite here over the last, over the last week.
00;26;23;00 - 00;26;24;04
Derek
What do you think? You.
00;26;24;04 - 00;26;51;00
Aaron
Know, I think, the question I was asking and maybe even turn it back around, I mean, but it feels like we should see, like, a new major popular set. I don't know if that is like a here's something different, but it feels like we should start to see that pretty soon. I do think a lot of the energy around the NFT market is driven by wealth effects, and just notions that, like, the market is doing well, you can move kind of further out on that risk risk curve like you were describing.
00;26;51;00 - 00;27;13;07
Aaron
I think the state of the NFT market is going to largely depend on just the overall health of the broader digital asset ecosystem, but I imagine 2025, that's going to feel pretty good because of changes in regulations, because we're seeing really like actual innovation at some of like the base project level, I'm assuming we probably will see something big and new.
00;27;13;07 - 00;27;32;16
Aaron
I I'm not fully fixed on the fact that it has to be like a set. I think there's a lot of a lot of reasons, reasonable arguments as to why that would not be the case. I do. People want to signal themselves on a social media platform using PRS. Does it take a different shape because it manifests on like another platform, like something that's not x Twitter?
00;27;32;16 - 00;27;40;29
Aaron
Do people want to use discord? Still seems like there's a bit of fatigue on the discord side, so I don't know what the shape looks like, but I'm curious what you guys think.
00;27;41;05 - 00;28;20;23
Chris
Yeah, one thing we're seeing, right? You said, I don't know if people want to project themselves that way or identify that way. And we are seeing like new forms of how people do want to identify on the net with these agents. You know, which has been the big meta in the last quarter of the year. And you can look at that, you know, in sort of a lineage of, okay, well, just started with having a social media account, then maybe extended to having a and a group identity that, you know, I could, form a tribe with on the social media account now towards like, I would I want to like actually, actuate and
00;28;20;23 - 00;28;44;14
Chris
identify through, an agent that, you know, actually has more capabilities and, you know, can serve as an extension of me and my interest. And so one of those, one of my thoughts around this is we're starting to see a break between objects and systems in digital culture, where is the object, the I'm going to create a synthetic version of myself.
00;28;44;14 - 00;29;12;10
Chris
It's going to sit on top of this agent platform and it's going to interact, you know, with the token, with Twitter, with, LMS and I. Right. Like that's that's systemic behavior. And so I do think that's something we're certainly going to see play out more and more in 2025 is the separation of object based representation versus being and becoming in a systemized format.
00;29;12;13 - 00;29;13;03
Chris
I think the.
00;29;13;04 - 00;29;34;08
Aaron
Other thing I'm kind of interested in is like, is there it's the stuff you were talking a little bit about last week with like recent project, like, is there something more dynamic? Is there something kind of more like net new media type? I, I personally like looking out for that and looking for maturation around that. I would like to see like more interactive programable, digital art and NFT sets.
00;29;34;08 - 00;29;45;18
Aaron
I think that stuff's really cool. I think it resonates to like core collectors and users in the Ethereum and other other ecosystem. So I'm hoping we see that in 2025. I wouldn't be surprised if we do.
00;29;45;22 - 00;30;07;17
Chris
Yeah, no, you will have programmability. I mean that's what this stuff is, right? It maybe we're just kind of talking across each other and saying the same things. Right. But flat, inert objects that represent beauty hold value, you know, are stored. Immutably. On a digital ledger. Right. That's one class of things. And we've kind of been through that, seen it done it.
00;30;07;17 - 00;30;29;25
Chris
And look, that can continue to evolve. No, no one says, oh, well, someone hung a picture on a wall, you know, 500 years ago, we don't need to make more art. Right? Like there'll be plenty of that at the same time. Yeah. The living, breathing, functional, interoperable aspects of these things are coming to the fore. And, you know, that's great because that was an early promise of the space.
00;30;29;25 - 00;31;01;22
Chris
You know, when we talked about composability and the fact that these things can interact with each other, maybe, you know, the first time through, everyone thought about that in the language of smart contracts. And now that we're getting further along, we're realizing, no, it's going to actually be through the medium of I, you know, stringing these things together and operating behind the scenes, or it's going to be a mix of contracts and AI and into wrapping into all of these other things, you know, like your social media platforms.
00;31;01;29 - 00;31;09;11
Aaron
So, Derek, do you see anything outside of the like the PHP, like, set or what's kind of your view wasn't clear to me from before?
00;31;09;17 - 00;31;31;05
Derek
Yeah. I mean, I think I mentioned our a podcast, or one on this podcast a couple episodes ago, but, I'm definitely really interested in systems and like, system art and we've, we've, I think unpacked that a bit over the last couple of weeks. I definitely think it's clear to me that like, we art is often a reflection of just like, kind of like contemporary culture in many ways.
00;31;31;08 - 00;31;54;27
Derek
You can trace back kind of like the art that was made during different movements. And it's typically, cultural response or technological response. And I think my view is we're spending more and more time in front of screens where spending more and more time interacting with digital stuff. Most of our friends and families exist to us, and shared networks and digital form.
00;31;54;27 - 00;32;18;05
Derek
And so the evolution of digital and that abstraction that we view is changing. And so I think the thing that I'm trying to keep an eye on right now as well, in what ways is it? Is it in ways that are more automated and procedural and then yes, then, like how our artists that are doing generative, procedural based work, playing with those ideas, is it more around synthetic intelligence and gen tech infrastructure then?
00;32;18;08 - 00;32;39;14
Derek
And if the answer is yes, then how are artists playing with those systems as they become available to them? The fact that we're spending more and more time wearing headsets or AR glasses, and if the answer is yes, then how are artists playing with those new tools that are becoming available to them? In ways that kind of meet the technology, where it's at our response to it in some meaningful way.
00;32;39;14 - 00;33;11;01
Derek
So, you know, I mentioned this a couple weeks ago, but like, I like to see how artists are using systems to be able to kind of like bring out, you know, these, these elegant responses to technology or to culture in, in ways that people can kind of, empathize with or identify. I think one of the reasons why we all loved that clown vamp, exhibition in Martha's, because I think it was a great, you know, encapsulation in response to kind of like, I think how everyone's feeling right now, it played with physical and digital very neatly.
00;33;11;01 - 00;33;43;27
Derek
It was a completely self-contained system. It was, you know, running, a local model without connecting to the internet and graphically and compositionally. It was it was a like, interesting, interesting subject matter that I think we all kind of could identify this being trained on human data from a very specific period in time that I think evoked some emotions around how technology and, and kind of human data kind of could interplay with one another to create something, so provocative.
00;33;44;04 - 00;34;01;24
Derek
So things like that are interesting to me. Obviously, we've talked about Bato and 113 in here, and I'm just excited to see more of that there. Another really will be breakouts and breakout artists and breakout collections that I think do that in a way that, that get people very excited to collect in and support.
00;34;01;28 - 00;34;20;17
Chris
And Derek, you know, as you were talking about clown VR, but I had a thought here that I think another thing that's bullish for this space, right, is in some ways, you can look at 2021 as an entire world of people getting called up for a cup of coffee in the major leagues and and seeing big league pitching for the first time.
00;34;20;17 - 00;34;39;27
Chris
And when that happens in baseball, a batter might get really hot for a while there, but then pitchers figure them out. They study the tape, they understand what that person can't hit. And then that that, you know, rookie gets cold, gets sent back to the minors and they have to work on their game. And you know, I do think like let's go to Penguins for a second.
00;34;39;27 - 00;34;58;12
Chris
Right. And Luke I like Luca very clearly is a strong operator who, you know, can execute a roadmap and keep a lot of balls in motion. We didn't know that about Luca when he first came on the scene. We didn't really know that about a lot of these operators who were creating these, you know, very elaborate roadmaps or even like artists, right.
00;34;58;12 - 00;35;25;26
Chris
In finance case, Casey started doing, I want to assume the image or just, you know, static image generation. And then he moved into the, detective Jack series, more ARG and mystery and like, collective coordination. And so, you know, kind of he's expanded, you know, the world his work operates and and now he's moved into this, you know, system, physical, digital commentary on society.
00;35;25;26 - 00;35;35;21
Chris
Right. Like, you can you can see those sort of progressions in the people in, in the space in both like individual practices and in project level roadmap execution.
00;35;35;26 - 00;35;56;02
Aaron
Yeah, that's a great point, Chris. I, I will say clown vamp wasn't just Murphy. He's been taking that all over the world. It was just in New York this week. I got to see it again. It was great. The setting was probably not as controlled by by Andy Clown vamp as much as it was in Marfa, but it was a bunch of people there, which I thought was really cool.
00;35;56;09 - 00;36;00;27
Aaron
I think he's going to Japan next, so it's fun to kind of see this kind of permeate through the planet.
00;36;01;01 - 00;36;04;22
Chris
You're going to hit the road and follow him around like it's a fish for sure.
00;36;04;25 - 00;36;12;17
Aaron
No, definitely not. But I'm a I'm a I'm a big fan. I think I think we'll see more good stuff from him, kind of going forward.
00;36;12;24 - 00;36;32;03
Derek
Changing topics slightly in our last couple minutes. I think we just saw a few of about an hour ago. There's some teasers coming out about a, an open sea token. The Open Sea Foundation now has a Twitter account. We're seeing the, the, the left facing eyeballs, which always means something is, something is different.
00;36;32;03 - 00;36;53;03
Derek
Keep, pay attention a little bit, Devon. His sea of open sea is has got this rock in the eyeballs everywhere. And, open sea account is rocking the eyeballs. I guess I got a couple questions for you guys. Well, first of all, what do you think about an a possible OpenSea token number two? What do you think are some of the downstream effects of a token like that.
00;36;53;03 - 00;37;03;19
Derek
And and then number three, how do you guys think about the NFT exchange platform wars that continue to persist and what's exciting or not exciting? And where do you think this all goes?
00;37;03;22 - 00;37;25;21
Aaron
Yeah, it's a good question. So like as OpenSea, like, sells the gen waters like what happens that gives me like a little bit of nervousness related to it Derek. Like why do they need a token. We kind of saw that with like blur. And I don't know if that was like a net positive for the ecosystem, but at the same time, I understand why they may want to use, use that type of an event to kind of rekindle interest in their platform, etc..
00;37;25;22 - 00;37;47;07
Aaron
I'm, I kind of like wind up being like a little bit of mixed minds on it. I am very long term bullish on OpenSea. I still find myself using it pretty heavily. I think it it feels like something that could scale as the digital asset space just scales and gets more normie people. And like, I think a lot of the other exchanges and or like NFT collecting platforms, they just don't really resonate with me.
00;37;47;10 - 00;37;55;16
Aaron
Like I don't want to like, feel like I'm going to like some sleazy nightclub when I'm buying digital art. And I feel like OpenSea is really kind of balance that really well.
00;37;55;22 - 00;38;25;17
Chris
$69 billion market cap and obviously, I mean, it's pre-baked. Yeah. Look, the entire community has made it abundantly clear to OpenSea that they don't give a shit about anything OpenSea does until they deliver a token that's, literally to say the entire community, because Aron clearly has reservations. But the vast majority of volume that flows in these marketplaces has an expectation, and that expectation is token that has been an elephant in the room.
00;38;25;17 - 00;38;48;10
Chris
And it is, you know, certainly like affected how a lot of people view OpenSea. And look, a lot of us remember a lot of us don't forget how OpenSea treated, artists royalties and secondary sales. A lot of people do know that, you know, OpenSea has been tone deaf as an organization as it comes to relating to the community.
00;38;48;10 - 00;39;11;15
Chris
You know, in a world of like, vibes and wag Me, they've consistently been corpus seek and it has been rather frustrating, I will say, as an OpenSea power user, to watch the way that, platforms interacted with the public. And so, look, I don't have any expectations around a token like a token would be great. You know, I always love free money.
00;39;11;15 - 00;39;40;03
Chris
It's it's a wonderful thing. It's good to me from, you know, maybe resolving or providing catharsis around like, you know, the history of OpenSea and its user base, how that token actually performs, what it's used for. Is it really just a concealment of a promise because we're in a better regulatory environment and they they feel they got the air cover to make good on this or whether it gets integrated and is useful on the platform.
00;39;40;05 - 00;40;00;02
Chris
I don't want to say it's irrelevant because clearly those are different pathways, but at the same time, like there's this beef between OpenSea and the NFT community, and they do need to sit down at the table and break bread. And, you know, the tokens are great vehicle to do that. So it's kind of just, you know, my thoughts there.
00;40;00;02 - 00;40;08;23
Chris
I don't want to get super deep into the weeds. I just kind of want to like think about it in terms of relationships. And this is a fractured relationship that needs some mending.
00;40;08;23 - 00;40;47;25
Derek
So do you guys have a, one question I have. So I, I think I get the logic that you're describing and er, and I get the, you know, the temperature checks, the you're, I think questioning whether or not this is a net positive, for OpenSea broadly just given that the product works and is maintained without, without the need of a token, do you guys think that this changes the course of future business, like does this do you think a token is backwards serving or do you think it there's, the possibility that a token, attracts a new type of audience and a new type of customer and a new type of unlocks
00;40;47;25 - 00;41;01;11
Derek
some new business models for the protocol or the product in the future. Is there anything kind of forward looking that you guys are excited about with, with, with the advent of, NFT platform token, and this being dropped in this way.
00;41;01;13 - 00;41;22;28
Aaron
I guess that's one of my hesitation is Derek. I mean, we've we've seen, you know, what looks rare blur. There's a handful of other ones too, that did some token model. I just it hasn't worked yet. I mean, I know Devin and Team are brilliant. I think they put together on balance, like a reasonably good, product. I mean, I think people can kind of quibble about about parts of it kind of move faster.
00;41;22;29 - 00;41;39;28
Aaron
You know, some, some areas still could be smoothed out. You know, to me, like OpenSea just needs to get deeper into the, the space. I think they've been so heads down that they may have missed a couple of things. I guess the question is token just introduces the end user, and I just don't know if they're the best user.
00;41;40;00 - 00;41;41;07
Aaron
But what do you think, Chris?
00;41;41;10 - 00;42;07;05
Chris
Marketplaces have always confounded me in the NFT world. It's it's one of these things where our prior set of assumptions about two sided marketplaces and building networks, and that leading to lock in in terms of volume, you know, the Web2 model of marketplaces hasn't played out here like you just named a bunch of marketplaces magic. Ian was another one that did a token draw.
00;42;07;08 - 00;42;30;14
Chris
All of these have had their moment in the sun, and then volumes moved elsewhere. And so something is is fundamentally off about how marketplaces work in our space. Because I don't know, like, you know, I we've had conversations in the past where I've questioned whether NFT marketplaces are good business here. Right. Like we've looked at opportunities, you know, within the DAOs.
00;42;30;14 - 00;43;08;20
Chris
And I've always just throw my hands up and said, I cannot figure out what's going on here because intrinsically, and from a pattern matching perspective, this should make sense. But from an actual running of history through our space, these things have proved like very volatile, wildly erratic. There's been huge swings in who has mindshare. What is clear is like you do need like big flows of liquidity, that when you can capture a critical mass of action, it does, catalyze the space and, you know, creates, velocity right around projects.
00;43;08;20 - 00;43;38;00
Chris
And so it does have trickle down and secondary effects, farming being like maybe an egregious example or an extreme example of this. But also you can look at this, with OpenSea and, you know, art blogs when they were both in their heyday. Right. Like art blocks and wouldn't have had that capturing of attention and capturing a value if OpenSea wasn't there to facilitate and, you know, allow people to experience, the right of, you know, the secondary market.
00;43;38;00 - 00;44;07;05
Chris
And so it is a really weird and tricky one. I think you're really going to need to sharpen your product skills and come up with like maybe really innovative or novel use cases, to incorporate that token, to give it like a real, like functional role in the marketplace. And no one's demonstrated that outside of like incentive based, farming activity yet, and so TBD.
00;44;07;06 - 00;44;26;26
Chris
But like I, I tend to take a pessimistic view that the token is going to get heavily integrated into the platform unless someone really comes up with a new way of doing this, because I think we've seen enough demonstrations of trying to stick a square peg to a round hole and none of them stick.
00;44;26;29 - 00;44;48;00
Aaron
That's I wonder if there's just like something like another marketplace that emerges, like, I wonder if we're kind of at an inflection point where we saw kind of like Yahoo and Jeeves and like all these different markets, I think, well, let me take a step back. I think search is like the best way to view OpenSea or one way to view OpenSea, you know, because it's not just a marketplace for at least me like serves as an explorer.
00;44;48;03 - 00;45;10;09
Aaron
You can find things, you kind of see all this metadata and social information. I do wonder if there's like another attempt at doing this, like a better, a better way to kind of view either digital media NFTs on chain or just like a blockchain explorer at some point kind of absorbs this like another attempt at it. And maybe we're we'll see like a Google emerge this cycle.
00;45;10;12 - 00;45;33;09
Chris
Yeah, yeah. Like to kind of combine what you and I are saying. Right. Like what's at Google apart was PageRank. Like PageRank might have been the most influential algorithm ever put onto the internet, and it had all sorts of ramifications. But one thing it clearly did was like separated Google from the pack and left like all the earlier iterations of search engines, into dust.
00;45;33;09 - 00;45;54;05
Chris
And so what is like, is there a PageRank moment for marketplaces? Is there a way to push them forward above and beyond, like how they're actually working today? And if so, is that an area for a token to get applied to? Like who knows. Right. Like that's why the future is fun and why we show up every day is to see what happens.
00;45;54;07 - 00;46;00;09
Chris
Derek, do you have a page rank moment that you think is hanging out there in front of 2025?
00;46;00;12 - 00;46;05;28
Derek
Gosh, I definitely do and I cannot wait to share it with you guys. And that will have to wait till next episode.
00;46;06;02 - 00;46;09;18
Chris
Oh, it's in the back pocket. Wow. What a what a big market.
00;46;09;20 - 00;46;12;01
Aaron
What a way to start 2025.
00;46;12;03 - 00;46;20;27
Derek
I think the next episode is going to be a heater. We should do our predictions for 2025 on the next trip. Is is is pre going to be joining us next.
00;46;20;27 - 00;46;21;16
Chris
Week I.
00;46;21;16 - 00;46;22;13
Aaron
Think.
00;46;22;15 - 00;46;23;16
Chris
No.
00;46;23;18 - 00;46;25;02
Aaron
I think it's just the three of us I think.
00;46;25;05 - 00;46;27;07
Derek
All right we'll have some fun again.
00;46;27;09 - 00;46;35;04
Chris
Yeah I don't know if there is the next week Derek our next week might be the first week of January. I can't remember the calendar like that's true.
00;46;35;05 - 00;46;38;06
Derek
Well let's will the next episode we should do our predictions episode.
00;46;38;13 - 00;46;46;18
Aaron
Let's do it. Well happy holidays guys, and thanks for everybody who, who's hopped on and started listening to this podcast and started this journey with us.
00;46;46;20 - 00;46;48;05
Derek
We appreciate you guys.
00;46;48;07 - 00;46;49;29
Aaron
Happy holidays and happy New Year.
00;46;50;03 - 00;46;52;18
Chris
Yeah thanks everyone. Happy holidays.
00;46;52;21 - 00;47;10;28
Derek
Happy holidays everyone.