The following is a rough transcript which has not been revised by The Jim Rutt Show or by Peter Wang. Please check with us before using any quotations from this transcript. Thank you.
Jim Rutt: Howdy! This is Jim Rutt, and this is The Jim Rutt Show.
Jim Rutt: Today’s guest is Peter Wang, co-founder and CTO at Anaconda, a leader in Python technologies for scientific computing and data analytics. And yeah, that’s Python the computer language, not python the snake.
Peter Wang: Well, hi Jim, it’s great to be here.
Jim Rutt: Peter is not only a very accomplished technologist in the Python space, but also is a deep thinker about our present world and how our technologically driven network culture and humans interact today and how they can better interact in the future, at least if we’re going to survive. He’s also a leader in thinking about next generation distributed web technologies. We’ve got a lot to talk about, Peter. So let’s start with an update on what’s going on in the Python world. One observation I have, I’m a Python user. I tried to learn Python at the Hello World level in 2004. Soon thereafter, you replaced my use of Perl with Python for text mangling and what have you. I then started using it more seriously in 2013, and now it’s probably my number two language after C#. One of the things that’s always been an issue in the Python world is version two versus version three. And now it looks like we’re decisively past that gap, and it looks like version three from now on out?
Peter Wang: That’s right. The train is not only left the station, but as of December 31st of this year, the version two platform will no longer be maintained. All the developers are moving to only supporting and maintaining version three.
Jim Rutt: That’s been my rule for at least three years is in lessors, an absolutely indispensable library that wasn’t version three compatible. I always ran with version three. And now I’m seeing a lot of the big main libraries have actually started to drop V2 support. So finally the decision is clear.
Peter Wang: Yeah. It only took 10 years. But…
Jim Rutt: In retrospect, a screwy decision to made that fork over such minor issues. But hey, it is what it is, and it’s now finally rejoined and it’s version three hereafter. Now, one question I love to ask a guy like you who probably sees absolutely everything in the Python space, what are some interesting new libraries and capabilities that have entered the Python space in the last couple of years?
Peter Wang: One thing I will say up front is that the Python space is so huge now. It is, I think by many counts, the number three most popular programming language. No one really can see all of it, but the last few years have in the general Python space reconsumed by the two versus three transition and what happens to two. And then also the async IO and asynchronous programming. A lot of those things have just had some growing pains they’ve been going through. In the data science and machine learning space however, there’s a ton of new stuff. And that’s where I do have a lot more depth and understanding.
Peter Wang: There, I will say that in the last few years we’ve gone, obviously TensorFlow started happening I guess about a couple of years ago for deep learning. Then we had a few other framers that came out and what we’re seeing now as the glow, I think the initial burst and glow from that phase, we’re seeing more libraries come out that deconstruct what a deep learning library is and create more general vector computing and optimization stuff. So I think that’s some really cool stuff related out Rapids from the Nvidia folks using some of our technologies like Numba and Dask and what else? Oh, there’s some automated machine learning tools that have come out for just getting people of a faster start over all these different possible machine learning models. I think those are some of the key ones. There’s just so much going on. It’s actually an interesting problem that we suffer there where there’s almost too much happening for anyone to keep track of.
Jim Rutt: I think I always worry, and we know about how evolution doesn’t always converge to the right solution, are some really good tools being ignored due to evolutionary drift essentially? Where for whatever reason, three tools, let’s say higher level rappers around TensorFlow, which has been a hot area of late, and maybe the wrong one winds, I don’t know. How do we keep enough diversity in an ecosystem to make sure that guys who should win do it?
Peter Wang: Well, gosh, there’s so much to unpack in that question. I think there’s a defense of the ecosystem that’s separate from putting our finger on the scale and saying, this library should win. In fact, if you have a generative ecosystem, then no single person can say what is the right thing. And we almost, we have to roundtrip it and run a loop through the user space, right through the innovation community and the people actually applying the stuff in practice. So we might, developers, we always have internal intrinsic attributes of a library that we optimize for. We’ll say, Oh, this was more elegant. This one has newer techniques, so and so forth. But until you actually put it out and run around trip through user space, you don’t know what’s really “right.” But to actually answer your question, I think that one of the things I’ve really very strongly feel we have an opportunity to do well in Python space, especially on machine learning, is to articulate a more modern defense of what used to be called open source.
Peter Wang: But it’s something different. And it’s a generative contribution gift economy around source code sharing that’s not inhibited or anchored by legacy approaches to IP and software monetization. So this got really deeply fast. What I mean is, we can talk about like, people know I’m an open source advocate, people know a lot of Python stuff happens open source. But I think the term open source is outmoded, and we should almost stop using it because a lot of corporate citizens in the opensource community will come out with open source. But their governance models are very limited. The governance models basically retain copyright. Only the corporate contributors and the employees of that corporation can push commits and things like that. That’s open source technically by every measure by OSI and FSF and everything. But it somehow manages to not retain the ethos of the open source code sharing of this generative innovation community. That is a struggle that I do worry about.
Peter Wang: When we’re talking about how do we assure that the space evolves in a good way? For me, trying to come up with that articulation of what are the values we’re trying to defend there, that’s one of those things. The second thing is then ensuring that lots of things have a chance to compete. So when you’re talking about rappers over TensorFlow, that’s all great, but 25 different rappers over TensorFlow at the end of the end of the day, it’s still the TensorFlow core. There’s other approaches to deep learning frameworks like PyTorch, like Chainer, MXNet, many others, and they have a different construction. They’ve more modular construction.
Peter Wang: An observation that my senior director of community innovation here has put so eloquently was that the reason the [Py 00:06:51] data community, or the side Py community won while we’re able to create all these wonderful innovation things is because the people who are doing the work, the actual heavy lifting of making code, making new modular libraries, they designed them with a certain economy of scope. So they could play with each other, we use pieces that are already there, nothing tried to boil the ocean. It was a really interesting innovation community. And I argue, actually one of my talks that I’ve given, I argue that the Py data [inaudible 00:07:17] community is one of the very few things that I’ve seen that is actually out innovated capitalism. Basically the return on your dollar there has been better than any kind of capitalist firm.
Jim Rutt: I certainly have found it is unbelievable ecosystem. I’m not a hardcore commercial production programmer, but I do like to try new things. I’ve tried out the various TensorFlow rappers. I’ve tried PyTorch. I’ve tried this, I tried that. I do a lot of stuff. And what I’m always amazed at, if I’m looking for some little piece of glue wear or something to move data from here to there, half a line worth of Google search in the Python space will usually turn up something for me. The space is amazingly well populated.
Peter Wang: It’s a blessing and a curse. We’ve got about a hundred thousand packages, so good luck.
Jim Rutt: Now, I’m talking to the man himself. So I should ask this question. If say someone who is not a true deep Python guru, are there ways to validate packages, libraries, systems? How does one think about choosing between nine different machine learning frameworks, for instance? What advice would you give to somebody who is entering this world?
Peter Wang: That’s a great question. As far as I know, right now there’s not really a standard way to validate it. There is not a central credentialing authority. One measure that’s popular is people look at popularity rank and how many downloads off PyPI, how many downloads off of Condo or Anaconda, how many questions are there on Stack Overflow, how active is the GitHub for that particular project. Those are all these proximal measures of the healthiness of a project. But unfortunately, the way this goes, some software or some projects that are very useful, very solid, they don’t need much fixing. They actually do a very small thing very well, and they sit around, they have no activity for a few years, but they’re still perfectly fine. So there’s a bias source of newness and activity there, which is unfortunate.
Peter Wang: So we definitely lack for that. And I think any, further on this conversation, we’ll talking about decentralization, we are suffering exactly that problem here. Anytime you have an abundance of options, now you’ve got a curatorial need. And the curation becomes very important. We’re thinking about ideas to improve that here at Anaconda actually. And we do play a very interesting role in the ecosystem. So we do have visibility into a lot of things that could inform such a curational system, but we want to make sure we do it in a way that’s not [hamfisted 00:09:46]. We want to do in a way that does surface, again, [inaudible 00:09:49] I talked about earlier, which is that sometimes projects don’t release very often are actually very stable and should be the ones that you do use.
Jim Rutt: Yeah. It’s certainly not as simple linear problem. It’s a multidimensional long linear problem. You mentioned a lot of things I use, how many downloads, does the community seem alive? Are there mailing lists or discussion groups? I do use, how often is it updated? But I don’t use that as a binary. As you point out, some very nice mature packages really don’t need any updating. So why should you care about that? The other one I like to look at is look at the issues list on GitHub. And if there’re a lot of issues out there that nobody’s addressing, that’s usually not a good sign.
Peter Wang: It’s interesting. It can be, but it could also be a different part of the quadrant, which is this is an incredibly useful library that just suddenly found massive popularity, and you have one dev working on it in his or her part-time and 10,000 people are on GitHub asking about it. Now, normally is not that much of a ratio, but we do see this really interesting phenomenon now where because Python’s gone mainstream, people, their jobs depend on it. You talked about yourself being a casual user, but there are people who literally, Python is a very high rank skill on indeed.com and these other LinkedIn and job search places.
Peter Wang: People, they get a job and the job requires them to use these libraries. And these are open source libraries maintained by volunteers. So they’ll show up on the tracker saying, hey, your software sucks because it’s not doing this and then the other. And the volunteer maintainers are like, well, we’re not getting paid for it. And there’s this incredible mismatch and disparity between the expectations of, what I would say, the standard labor economics around white collar work that uses this kind of software and then the generative abundance gift economy economics that is part of the sharing tribal culture that produced the software. And at the intersection of these two things, there’s not many tangent points between these two circles in fact. That’s one of my things I like to nerd out on, which is actually also an intersection of meta-level, is an intersection between the circles of my day job and all this Python stuff and open source community. And then also my thinking about future of human civilization and collective dynamics and things like that.
Jim Rutt: Yup. It sounds like you’re the right guy to be there to be thinking about this. The Python community is amazing, but if you could make it even more amazing if you could come up with some way to help people sort through that without putting the thumb on the scale prematurely or at all right. How do you get an emergence signal that’s a good signal? We will come back to that problem again and again. I mean, it’s funny one of the things that we look at, this distributed search problem. People say, well, why can’t we have a distributed Facebook? Well, it turned out it’s a pain in the ass to be able to make it searchable if everybody has their own copy. So these problems come up again and again and again. Now, where I’d like to go to next to something I know you’ve thought about and talked about a little bit. And that is the match and mismatch between human cognition and our worldwide technologically driven networks.
Peter Wang: Mismatch between human cognition and our worldwide network.
Jim Rutt: Technologically driven network. You’ve talked about things like speciation.
Peter Wang: Ah, okay.
Jim Rutt: Well, where would you like to start? Start saying what you think about this problem? That human cognition was not designed for fractal, huge, instantaneous worldwide networks.
Peter Wang: There’s two pieces of this. So one of them is tied to the limitations of human cognition. I was listening to the Krakauer interview that you did, and he talks about it really well there as well. He talks about all of our analytical models, we are really solving for how to have a shorter model description than the phenomenological observations. A thing I like to say about this is that humans really are narrow band sensors that have a preference for structure. We can’t help but see the world in terms of structure. And that structure is how we compress sensory input, structures how we compress our description of phenomenon. And we really have very narrow bandwidth to process these things and to retain them. And we remember them, we remember these compressed versions of them, almost like a little reduced hashes of them, not the full description.
Peter Wang: So human cognition is really, I mean, our brains are just these patterning engines and that works fine in three plus one D space that we grew up in, had to hunt animals in and greed in. But what we’ve done in all of our caloric excess in modernity, we’ve created now technological infrastructure that exposes all of this extra cycles that we have in our brains to a completely different massively high bandwidth set of perceptual inputs that are all virtual. It’s all virtuality, but through our eyes and our ears, I mean even this podcast is an instance of such a thing. We are exposed to a space that, or to a set of sensory things that had habit dimensions. They carve out dimensions. They are well beyond the three plus one D that our brains really were evolved to succeed in. And these worldwide networks, the communications networks, what they’ve really done is now they’ve created a place where signals and messages and ideas bounce around and essentially have a life of their own.
Peter Wang: What I like to think about is this is, if I could use a model that maybe we come back to over and over in this podcast, but let’s think of the human being is almost like a connected series of rods or masses suspended in water. And different layers of water have different densities, maybe different temperatures, and we’re loosely connected where each of us were just, every single one of us has multiple planes that we live in, but at the highest plane, the intellectual plane, we tend to think of that as this very rarefied air. But actually, our intellectual brains and our cultural brains, that tier of us, it is essentially substrate for massive amounts of wave patterns and signals that bounce off each other. Some cancel each other out, some reinforce each other. So we’re really like little particles of water or atoms, molecules of H2O in this giant ocean.
Peter Wang: That’s really my model for what’s happening now. Why it’s so hard for people to talk about it. Because people still think of themselves as essentially individual, whole atomic things that maybe read a book or watch a TV show or listen to a soundbite. But actually, we have to almost deconstruct what it is to be human and realize that there’s so much more signal pressure on this cultural intellectual part of us. And that actually causes a disconnect in neuroticism as it disconnects with our internal narrative going between these different connected pieces that comprise us. So anyway, that’s a lot to unpack there, but I just want to put that model out there is how I see the world and why I think right now we’re in such a weird place. We have metaphysical problems, we have a crisis of philosophy, it’s what I like to say. Because our model of self, our model of agency, our model of what even a human person is or human reality is or how important consensus reality is, all of those are breaking down.
Jim Rutt: Do they have it? Do you think this is by necessity because of the new modality or is it part of learning, lack of learning? And if we look back at the invention of the printing press, apparently the various frauds and bullshit indulgences were a big thing that the printing press was used for. The early days of newspapers were full of ads for quack nostrums. I remember email quickly got inundated by spam.
Peter Wang: That’s right. And even Marcus cookie recipe. Right?
Jim Rutt: Exactly, exactly. But fairly quickly, humans adapted and learned to filter out the most egregious bullshit most of the time. Is that what’s going to happen with humans in this high dimensional high speed network world are we entered a new regime where it’s just beyond our capability to deal?
Peter Wang: I’m going to say that it’s the letter, and the reason I say that is because well, and we see this in data science too, this argument that quantity at some level, quantity takes on a quality all of its own. And the difference between the misinformation and fuzzy information, fuzzy propagation of high quality signal and all these kinds of things that we’ve had in the past is we used to be in a regime, civilizationally or socially, of bandwidth scarcity. There’s just was not enough bandwidth to construct a high quality simulacrum of virtuality. We are now past that point. There really is a binary point. There’s a hard limit to how many pixels and what frame rate you need coming into your eyeballs for you to see something that’s very, very real. There is some audio encoding bit rate below, which it sounds like crap and above, which it’s basically like you’re sitting next to me. Anything beyond that is gravy.
Peter Wang: There’s a hard transition point where we have enough bandwidth to create a distributed, it’s not a thousand plateaus. It’s seven and a half billion plateaus. We can create this virtuality and now we’re in a space of attentional scarcity. So I think it is a very different thing. Yeah. And in fact, I got a really long discussion with the guy who was asking as I was talking about my ideas for decentralized web and things like that. And he was saying, all this stuff on social media, how is this different than all the pieces that people were saying about TV back in the day? People are saying TV would rot your brain. Neil Postman had this, Amusing Ourselves to Death book.
Peter Wang: People say the same stuff about TV. Why is it different now than social media? And I think the difference is that, there’s a couple of things, but number one, we’re just in a different bandwidth space. Number two, it’s not broadcast, it’s not broadcast. And that’s the problem. TV and radio and newsprint, at least their power came from broadcast. They can influence a billion or a million minds, but it was the same message going to the same million. What we have now is aggregated, narrow cast. So every single person that can get a deeply tailored virtual experience of the world that is then able to manipulate them however we want. And maybe you’re right, maybe it’s a learning issue where this generation of young kids that are growing up with all this stuff, they will learn defenses. The term [inaudible 00:19:32] to use is malware. They’ll have been inoculated to the kind of malware, but we have a long way to go before then.
Peter Wang: And number two, who knows what the world will look like by then that all of their interactions with each other, the human to human interface is already going to be intermediated by so much virtuality and digital modification. We may really be putting ourselves into a new phase of human evolution, and we’ve not contemplated really what that looks like structurally and what that means. And that’s very dangerous.
Jim Rutt: We’ll come back to that a little bit later. Let me push back just a little. A year ago, when they, when I first started getting some publicity, I became very concerned about the possibility of deep fake videos being used for very pernicious purposes. And yet as far as I know, there are zero examples or there are a little corner cases, but no major exploits have been made around deep fakes. That at least to me, indicates perhaps that people are developing these anti-malware immune systems and say, not really isn’t Hillary’s face on that porn video. That’s just impossible. Isn’t that interesting that all this concern about deep fakes a year ago has so far at least not been met by actual problems from deep fakes?
Peter Wang: It is interesting. I was also concerned it. Actually when deep fakes first came out, I went and downloaded the, you could download the image via BitTorrent, and it was about like two, three gigs was the software to do it. I was looking through, and they actually embedded Anaconda to install a TensorFlow and other stuff. So I’m like, yay, I’m very proud of this. But here’s the problem with that point of view. And I agree with you. We have not seen profusion of deep faked political whatever, take down videos and stuff like that. We are seeing some of it, we’re seeing social media sites and whatnot, having to have policies around, is it okay to take the likeness of a person and TensorFlow it, or PyTorch it on top of a porn actress and is that fair use? Is that art, is it slander? What is it? So we got that kind of stuff going on. That is happening. That’s not deep fake video, but it is happening.
Peter Wang: But zooming up a little bit, to answer your question more broadly, I think the danger here is that what deep fake represents is it tells us, hey, we’re almost good enough to be able to use a pilot GP use to trick our iconic visual perception system, which has been very good. That’s a very well honed system. Our appropriate septic system, our visual systems, these haven’t honed for a very long time. We look at a loved one, they raise an eyebrow, and we read a ton of meeting. So these are very well honed systems. The fact that we can take a bunch of vector hardware and fool them is like not great. It’s kind of like, huh, but think more deeply than this. We are actually entering a zone where we don’t actually know what to believe. We are going to exceed our own capability. We can create more virtuality that’s more real than reality. To make this more concrete, maybe that was the last track, but make it more concrete. You say you haven’t seen any deep fake videos yet. How do you know?
Jim Rutt: Well, actually I have, but I haven’t been fooled by any.
Peter Wang: But how do you know?
Jim Rutt: No, I don’t know certainly, but you would think still do forensic ability to detect deep fakes is pretty strong. That if we had been fooled, we would probably know at least one example thereof. And as far as I know, there’s not been a single widespread exploit of deep fakes that has penetrated any significant distance into our world. And as far as I know, the forensics can still detect a deep fake at a high level if you’re willing to put the effort into it.
Peter Wang: Right. So to push back on that a little bit, number one, again, how do you know? Because the way that we’re going to test for deep fakes, besides Amazon tracking a whole lot of this stuff, is we’re going to use probably guns and other kinds of deep learning neural nets to try and detect deep fakes. So you can see the problem there, because we’re going to take those, turn right back around to make better deep fakes. So our actual ways of detecting defects, the forensics that we can do, I think that sandbagging, that’s not going to be a perfect approach. But even more than this, I don’t know if you read the articles that were coming out after the Trump victory. People talk about fake news and stuff like that. And there was a really interesting, fascinating profile done on this guy who was just this random dude in like, I don’t know, Louisiana, somewhere.
Peter Wang: He just decided for lulls. He was going to create a fake website and make all these bogus so far right wing, outrageous articles. It’s like troll his conservative friends, and they started getting a lot of traction. He started selling ads, he started making money. And then before he knew it, he was becoming one of these main founts of misinformation or right wing like nonsense. It was just a random dude with a WordPress. So the bar that you set for yourself in what you effect, what you and I might see and say, Oh yeah, that definitely is actually Trump on a bed getting peed on by a bunch of Russian hookers. That bar may be very high for us. It ain’t that high for the vast majority, the electrode, I’ll tell you that. Some random dude with WordPress can make a significant percentage point impact on an election. You bet deep fakes can be a problem for us.
Jim Rutt: That’ll be interesting to see what happens in 2020, to what degree has the immune system against malware improved? For instance, I think the most successful fake news story, which is still like to your point, guys like you and me wouldn’t have been fooled by this one. But apparently the number one fake news of 2016 was, Pope Francis endorses Trump, got 20 million views. Will people be so gullible in 2020? I would assert without being able to prove that the answer is no, they would not be so gullible. That if someone were to refloat that or the analogous or equivalent story, it would not get 20 million views. 2016 was very early in the aggressive and moderately skillful application of fake news. And since then, just like spam. How many people fall for Nigerian letters these days? Not very many, but it was quite a good business back in the day.
Jim Rutt: So we shall see. I’m not saying that we don’t have an issue here, but I’m saying that the doomsayers, I believe, underestimate the ability of humans to learn adaptive immune systems against information malware. Now of course, it’s an arms race, as you point out. And there is no guarantee that the technology won’t eventually overwhelm the ability of the human immune system and just like 2016 showed, it will lag so that even if in the long term our immune systems catch up, a lot of harm can be done in the meantime.
Peter Wang: I think this is actually a really important thing to drill in on just a little bit because the way I said it earlier, I’ll just reiterate that. We are getting better at making virtuality that exceeds our perceptions of reality. Furthermore, more and more of our experiences of reality are intermediary by electronic means. Once you have this interface point that’s some kind of like electronic device, you can go and plug in virtuality, you can plug it in reality, it’s like when they go in the hospital and they put a little thing in your arm, they can just swap out syringes and swap out different kinds of drugs and stuff as they need. We’ve got that port in our eyeballs or ears now, we’ve got it, everyone’s got one.
Peter Wang: I think that when you talk about underestimating the human ability to adapt, I would ask you, adapt how? Besides turning back into the equivalent of a Neanderthal where we just have tribal, hey, what did you hear? What did you hear? Beyond that, what is your adaptation? Everyone’s going to learn how to debug TensorFlow code? Everyone’s going to go and go through the training datasets for these things? No. So this is where I think the arms race of machine learning and machine augmented human decision making and cybernetics, all these things are creating really speciation. Because I think you’re right, people will respond. They’re not going to sit around and be lied to all the time. But they will respond.
Peter Wang: And what I think is that the response for a lot of people who are below a certain tier of expertise, they are going to turn into the vassals. They’re going to be digital vassals essentially. I mean, they’re already doing that now. They are to becoming that now. So it’s not only elections that deep fakes matter, it’s not only deep fake videos that matter, but it’s the [anti-vax 00:27:58] stuff. It’s people with all sorts of financial, like weird financial fraud, little Ponzi schemes that show up all the time on Facebook. You and I, maybe we move into more rarefied circles, but for a lot of Americans out there who are, a lot of them are struggling and desperate financially, there is a ton of random stuff they’re exposed to and that stuff is going to become increasingly weaponized by machine learning, by maybe deep fakes, by whatever.
Peter Wang: Those people, after they get burned a couple of times, what are they going to do? They are not going to be a burden a third time, they’re going to maybe shut off their devices. They’re going to maybe just trust any kind of information to get through a digital means. And they didn’t revert back into, like I said, a more Neanderthal state. They’re going to revert back into, not homo sapiens digital, they’re going to basically become part of a substrate for other people to manipulate and control. That’s why I talk about speciation, and that’s why I deeply believe some of Jocko rules, critiques of the technological society, I think we’re really, really setting ourselves up for a weird caste system. And that’s my concern there.
Jim Rutt: Yup. It’s speciation. I think that’s interesting. [inaudible 00:29:02] maybe somewhat unusual that I still have lots of interactions with my childhood friends. I grew up in a working class neighborhood. Half the adults had dropped out of high school, so this was not a particularly sophisticated area, but the other people were generally hardworking, had good values, and have been pretty successful.
Jim Rutt: But it’s amazing that there’s lots of people my age who hardly know what Netflix is, let alone have ever used it. And Facebook, I have never be on that goddamn thing. Isn’t that just obsessed pool of shit? So they literally, in some sense, are almost speciated out out of this new thing that we’ve all been doing all these years. I’ve been doing it since I was in my twenties, and I’ve been doing it continuously. So I’m maybe a little anomalous for a mid boomer, but there were a lot of people who just aren’t engaged at all. Or as you point out, some who have been burned or disgusted and become information nihilists with respect to the networks. And frankly, that may be better in some ways in being duped again and again and again.
Peter Wang: But it’s a loss of agency. It represents a loss of agency. And I know that all nations, was it Eric’s great quote that, “A nation is a group of people who’ve decided to forget something together.” We maintain an illusion of democracy in government by the people and all this other stuff. It’s a useful illusion to maintain, but at the end of the day, there is value to having some reality behind that illusion. And when these people opt out of all this stuff, they’re opting out of the governance system. That’s why I call them vassals.
Jim Rutt: Well, or they build a governance system in a different modality. Perhaps they decide, fuck what’s online. We’re only going to believe what we see face to face. I think that’s an interesting alternative. You’ve probably heard me talk about in the past my distinction between weak links, which are done online and strong links which are done face to face, and it strikes me in the United States at least the strong link networks had been underinvested in in the last 30 years as we’ve all been ever so infatuated with the weak links. We look at other places, the five star movement in Italy is a very interesting example. It was basically built around meetup, meetup.com. Face to face interactions at every scale all over Italy. We could see that from the rejection. So yes, speciation but doesn’t necessarily mean our species will win.
Peter Wang: Okay. I don’t disagree, but I don’t agree. You’re right. That doesn’t necessarily mean that our species will win whatever our means. I don’t disagree with you about the strong versus weak links. In fact, I would say that the past 30 years, as you said, the past 30 years of underinvestment in strong links, I think that was your phrasing. I think actually that’s your touching. You’re brushing up against the root problem here. Vicky talks about the crisis of meaning and this last 30 years, it’s been more than that. I actually say it’s something closer to about 60 years. I peg it to basically the postwar in America and the beginning of really mass media driving mass consumption. What’s happened is that every single step we’ve removed, we’ve been weakening communalism and community strong links between people.
Peter Wang: Then we’ve been replacing those sources of meaning and those interactions with celebrities, brands, things that are delivered to us in a broadcast fashion. And even with the avenue there from radio and local FM stations getting snapped up, same thing with cable TV that went from community local to being global broadcast Turner. So you have this broadcast mass media phenomenon that has essentially destroyed a lot of that person to person in person sort of thing. And I think the strong links, the strong breakdown that you’re talking about, it’s, well maybe not intentional, but it’s been an active thing.
Peter Wang: It wasn’t like, Oh shoot, we forgot we were supposed to see other people. No, it was, hey, your kids should watch more TV. Hey, did you see that TV show? You should read the newspaper, the newspaper’s talking about the stars on this new TV show, this pilot that showed all of us are tuned into broadcast mass media. And that by itself has been a deeply weakening thing. So when you talk about the people doing face to face connections as a way of building movements, I think that’s fantastic. That’s true and correct. However, however, I think that kind of stuff, if it’s actively being competed against by the digitally networked and digital mimetic systems, it doesn’t hold a candle.
Jim Rutt: Yup. And this is what I really speak towards, is a hybrid of strong link nodes linked by weak links. So they can coordinate and sense make using weak links and act to using strong links and getting that architecture. I mean, those are just words. Getting that architecture right where you take the benefits of both strong links and weak links and create a movement that uses both appropriately strikes me as the winning strategy. Nobody’s done it yet, at least that I’m aware of, but I put it out there as a challenge for movement builders, party builders, memetic warfare people to take that combined idea and see what you can make out of it.
Peter Wang: Yeah, I agree with you. But I would just push back on that concept of it being a winning strategy because that explicitly sets up, and maybe you’re not really saying this, but when you talk about this being a winning strategy, it almost sounds like there’s a goal. There’s a thing we’re trying to achieve and this would be a way to win it. But I want to be very clear that when we say that, this is a really very difficult thing to articulate. I think you’re right that in a state that you and I would be happy to see, would be a system that could be described by that description. But the goal wasn’t to win. The goal right now is to create a better architecture for humans to build shared values and shared meaning.
Peter Wang: Joe Edelman talks to that. I think on the Emerge Podcast, he talked about having goal directed versus value directed approaches. Even though I agree with you from a phenomenological description of what a winning state looks like, I want to make sure that we don’t convey the wrong idea that hey, we want to take over the world. Here’s Jim and Peter’s grand plan for megalomania. That’s not what this is about. It’s more about having communities where there’s high meaning and high conviviality, which we don’t have right now. We have a lot of individuals, we’re, I think, a peak individualism. The mass media stuff has played into that very well, and we’re just done with it. Liberty is not enough.
Jim Rutt: I think that was very well said, Peter. And maybe I will riff off that a little bit. Yeah, that’s idea of this weak link, strong link community when I say winning, let’s say wins enough to build a safe ecosystem to develop an alternative to the status quo. Where people living within this new force field can basically reject the game A programming and have a space in which they can build game B. We don’t know what game B is yet, but we know that it avoids many of the pernicious aspects of game A. And to my mind that’s winning.
Peter Wang: Yeah. So by that definition then, avoiding game A failures, we can pound to find that as game B winning. With that, I would agree.
Jim Rutt: All right. Let’s talk back a little bit here to fake news, networks, et cetera. One thing that’s very different, big difference between 1975 and today, is in 1975, there were a very small number really, a very homogeneous gatekeepers. There were the three TV networks. There were frankly a small number of newspapers, the New York Times, Washington Post, LA Times, Chicago Tribune that drove the news agenda. There was Newsweek and Time Magazine. So there was a very narrow choice of views of the world. Now, this had some big negatives. I mean, frankly, the idea that gay liberation would have happened under that kind of system is relatively unlikely and lots of other good things that have come from the breakdown of the control of the discourse by a small number of people. But at the same time, having those small number of professional quality discourse mediators also kept out a lot of general horseshit. Stuff like anti-vaxxers and flatter and God knows what other crap circulates around our unmediated networks today.
Jim Rutt: So it seems to me that what we really are in need of is a new synthesis where we can get some of the quality of mediation, discourse, filtering, et cetera, that we used to get from this handful of big bland mediators. But instead, is dynamic and small scale and fractal and emergent and all those good things. Sometimes that goes under the name of sense-making. We need to learn how to exist in a world where there isn’t a small number of big brother mediators yet still have a whole lot of the useless crap filtered out that we otherwise would waste our attention on.
Peter Wang: Yes. I do say that when I talk about what I think we really need to do, we need a new infrastructure for sense making and meaning making and really collective correct version. But sense making is the first part of the collective action.
Jim Rutt: Could you define sense-making? This is a word that gets thrown around a lot and I love to hear what your take on what it means is.
Peter Wang: Well, the way I think about it, I’ve in the last couple of years really, I’ve come to a point where I have a lower and lower sense of esteem for our baseline human hardware. Even as I maintain great faith in the nobility and the potential of homo sapiens and people, I just have a certain sense of, I don’t know what the right term is. It’s not nihilism, but certainly a sense of like, man, most of what we deal with this construct. Most of what we tend to even hold up as being, I don’t know. The reason I say all this is because on the one hand, we can talk about sense-making in a way that let’s say the default perspective on sense-making is, hey, all this stuff was going on. We live in a VUCA world, volatile, uncertain, complex and ambiguous, and we need to figure out which side is up.
Peter Wang: You look at the situation with the downing of the airplane by the Russian missile. Well, was it the Ukrainian rebels, was it the Russians? Who was it? And then you look at when Trump authorized the airstrikes on that Syrian airfield. Even hardcore like actual fact things, we can make sense of now because our world is now so big, so complicated. We see so many different things. And so in this VUCA world, we have to have something, we have to get new tools. You talk about the collapse of the gatekeepers of the fourth estate. How do we figure out something that suffices the same needs that those people sufficed? That’s one view of sense-making. Your average Joe Schmoe at home needs to understand they live in a world that has a continuous past and has some sense of a possibly coherent future.
Peter Wang: My view on sense-making is a little bit, I view it in a way that maybe less privileges the individual human viewpoint. Which is to say that humans basically lie to themselves all the time, and we go through the world relying on a series of lies and myths and constructions and fables. All of these things we call models, model, stories, memories, myths, whatever you want to call them. And sense-making is the process of compressing this gigantic amount of stuff coming in into something that patterns into or is resonant with, or is not completely out of tune with the various different models that are playing in our head, that’s what most people solve for. They solve for, this doesn’t offend me. This doesn’t surprise me. It doesn’t make me feel like I’ve been living a lie, et cetera, et cetera. That’s what most people solve for when we talk about sense-making.
Peter Wang: Now, when we talk about sense-making in an aspirational sense, it’s that we want to get high quality models that we can then feed into a control loop that we can then manipulate the environment, taking courses of action that lead to outcomes that we desire. That agency, the cybernetic loop is a critical part of useful sense-making. But the vast majority of people, and I gave a conference talk about this a couple years ago. It’s actually where I met Jordan for the first time, and I gave a talk about the story of stories and how I call it the making of sense.
Peter Wang: Most people demand that reality conforms to their models more than they demand that their models have some semblance to reality. That’s unfortunately, when we talk about sense-making, we have to understand there’s many different species in the tribe of human that need to engage in a sense making activity. So when we talk about fighting fake news or coming up with higher quality new sources or having decentralized or distributed ways of patterning and says it’s all these things, we also have to understand that a large part of humanity still runs on malware and firmware that requires stories that make sense more than information that leads to higher quality courses of action.
Jim Rutt: Actually, if we want to get historical, a little philosophical here, we’re essentially talking about the difference between enlightenment thinking and pre enlightenment thinking. The enlightenment was the first full break away from just believing random horse shit that somebody made up and instead attempting to make sense that was actually congruent with reality. And yes, I’ll put my metaphysical cards on the table and admit to being a naive [crosstalk 00:42:11]. I actually believe there is a single reality, and we can learn more and more about it. Though I doubt we can ever learn everything about it and that the enlightenment toolkit was all about sense making to bring our models as congruent as possible to that reality. It is a little disconcerting that on both the left and the right we’re seeing people explicitly pull away from the enlightenment as a core value and retreating to, I’ll just be happy to believe horseshit that it’s been handed down through the generations because it’s the horseshit that I feel comfortable with and that strikes me as exceedingly dangerous and very counterproductive for the future of the human race.
Jim Rutt: Now in terms of how this might actually work, I had an example today. I mentioned my hometown buddies. We go hunting, we go fishing, we go out and do crazy shit a few times a year. They know that I’m well connected into the networks and have reasonably good sense making skills and also other nodes near me. I got an email today from a friend asking me, “Is this real or is this bullshit?” It had to do with something to do with a gun control in the federal government or something. I did five minutes worth of research, and I said, yeah, this is real. And it does seem somewhat unprecedented. So yes, this is something you should pay attention to. So as I started thinking about that as just a very baby example, one might think of sense-making being in small to medium sized groups that are interconnected across, call them secretary nodes, like the secretaries of correspondence in the continental Congress and the pre-revolutionary America where every big community had a group of pre revolutionaries, and they had a corresponding secretary who wrote to the other nodes.
Jim Rutt: So my buddy asked me, I asked somebody else, I did the research. And he now has my power at least at second hand to make sense of some random thing that came in over the internet. And that may be the model. So long as there are a sufficient critical mass of enlightenment thinkers in this connected network that a person is part of, that provides their sense-making apparatus.
Peter Wang: Yeah, I agree with that wholeheartedly. And I think that this is the thing you talk about there. I also, in my own conference alighting taught that I gave a couple of years back, I talked about the difference between a human network versus a computer network. A human network, the intelligence actually is in the network in that it’s a trusted network. There’s trust in the connection. So your friend, if they got an email from me or from Joe Schmoe saying, yeah, this thing is real, they would discount it completely. But if they get it from you, they believe you. The fact that it’s you, that connection itself, trust and connection is 90% of the signal. It doesn’t matter what it says on that wire actually. It so strongly modulates the signal going down that connection.
Peter Wang: But the internet is very interesting. It’s as you know as well or probably better than most, the internet is just as ossified frozen, like you could call it frozen accident, I think is great term. It’s the frozen accident of a command and control computer network where the network is supposed to be dumb, where trust was not entity we cared about at all on the network back in the day. We’ve overlaid our expectations, and our software provides affordances as if the network we use is a human network, but it’s not. It’s a computer network and the modalities of communication, the broadcast, the amplification, the kinds of gamification we do on it, all those things exploit the characteristics of a computer network and completely underserve the needs of a human network.
Peter Wang: Also, they don’t surface this kind of thing where you’re a trusted expert on a couple of things, and your friends trust you on that. That’s very, very hard to surface in a scalable way. The human networks are very different than the computer networks. So I think in the long run we’ll have sense-making networks that look more like these organic human networks. That’s what I hope.
Jim Rutt: Just as a sidebar, I have talked about this a lot, but our architectures, our technical network architectures are all built on foundations of sand. TCPIP, HTTP, HTML and Unix. All three were designed for a non-adversarial world. These were all designed for essentially research communities of people who trusted each other and did not assume bad faith exploits. And here we are trying to build high fidelity, reasonably coherent signals on top of these bases of sand. And it is A, ongoing problem, which we will probably not get out of anytime soon, but we have to do it.
Jim Rutt: And again, this is where I come back to the weak links, strong links. When you say scalable, it doesn’t have to scale. Think about it as a fractal network where it can scale in smaller nodes. Let’s say it’s a community of a Dunbar number of people, 150 who have some reasonable level of interconnection amongst them, but some of the members of that network also reach out to other Dunbar clusters. So that when you aggregate up, you eventually get to the whole world. But most of the communications are within relatively small groups. That might be the right way to go.
Peter Wang: Yeah. And actually, I just want to clarify, when I say the word scale, I think it actually can scale. And it’s one of the reasons I say that the biggest social network and the most resilient social network in the world is still email. But actually the scale that I’m talking about there, I should have air quoted it. What I mean is, scalable in the sense of modern like VC backed, scaled monetization model kinds of things. I think that these kinds of, it may very well be the case that really good, humane communication technologies almost by design cannot be monetized in the fashion they are been monetized and now. The business models around them are not scalable even though the networks, and the meaning making is.
Jim Rutt: I love that distinction, and I think actually that may point us in some sense of where to be thinking. Don’t think in terms of yeah, I got to build another unicorn. That’s what I’m doing this week. I built two, let me build a third one. Fuck them, the unicorns. Instead, let’s solve the actual human problem of how can people do sense making of reasonable high fidelity even if they personally don’t have huge sense-making skills. How do we allow people to find their advisors or how do they create a virtual set of advisors from a relatively small number, maybe Dunbar number-ish, a set of network connections? How do they discover, how do they knit that up?
Peter Wang: Yeah, and I think the closest we got to something like this was in the, there’s this like a period of a few years in the, I think the late nineties maybe, when the internet we got rich enough to have embedded audio and video kinds of things. You had some flash animation stuff and we then had web rings. We had web rings out the Wazoo.
Jim Rutt: I remember those. Those were actually pretty cool.
Peter Wang: Yeah. Those are evolution, Darwinism vibrating I was on and there’s a Babylon five web ring and all these things. Those are great. That’s organic socialization happening. The thing about this is I think that we’ve forgotten, we’ve forgotten just how powerfully leveraged an activity software development really is. And that in the last 20 years, the creation of software, especially enterprise software, the fact that people can get employed stitching together, writing more bandaids of Java on top of other like Java crap that some other consultant build. We have so much financial activity happening around building essentially pretty crummy software that we forgot that software actually is quite leveraged as an activity. I’ll tell two little stories.
Peter Wang: Now, one of them is when we talk about in the open source world, the sustainability crisis. This massively open SSL, for instance, turns out the Heartbleed bug without results because there’s only four part-time people working on open SSL, and it’s a core piece of infrastructure for the world. In Python, we’ve got software like NumPy and Pandas and Matplotlib, for instance, the most popular plotting packs to Python. That’s maintained by maybe a couple of part-time people in there part-time. So we talk about the sustainability crisis and open source, we can wring our hands about it and it’s a problem that we’re not funding these people very well.
Peter Wang: But if you think about the flip side of that, how many billions of dollars of intellectual value has been provided by just a few people in their part-time? That’s amazing. Do you imagine if you went back to the middle ages and get one farmer cutting wheat for a couple of hours a week could feed all of Paris, that would be insane. So the leverage we get off of software, open source shows how incredibly powerful and how impactful the power of software leverage can be when it’s unencumbered by scarcity business models.
Jim Rutt: I love that. But if you have more to say, I was going to transition to a discussion of non rivalrous economics.
Peter Wang: Sure. Well, I think this is all part of the same conversation. And I will just say one more thing. The thing about all of this is that most of the software activity that happens in the world today is pretty much bullshit. And it’s bullshit because we’re building on sand. Where we have a lot of stonemasons and architects trying to build cathedrals and Notre Dames on literally on quicksand. Actually if we were to build good underlying pieces that solve problems, they don’t need to get modified very much. We can actually build incredibly powerful and fun things. And you can look at Notch, Who built Minecraft, which is one of the biggest, biggest gaming phenomenon that’s happened last 10 years. It was one guy working on this stuff and he put this thing together in like six or seven days.
Peter Wang: Real underlying software innovation hasn’t really been happening in a real way, and I think in this way I’m also resonant with Eric and Peter Thiel’s ideas about stagnation and all of the as if stuff. We have as if innovative software industry. But actual software innovation, it’s really not happening. We’re still building in the long shadow of the ’70s, and it’s starting to really crumble.
Jim Rutt: So there are some very clever things, like I still recall that was about five years ago, the first time I used Uber, I got Holy shit, somebody taken a bunch of the pieces and connected them with tinker toys. Now I would not want to see the back end on some of this stuff, but this is really a remarkable concept and execution and we’ll comment on the wonderfulness of the business model. But in terms of type of logical tour de force, it was pretty damn amazing. It worked. It was intuitive. It was easy and go, wow, someone’s actually innovated in a major way here. Though, as we say, the foundations are crappy. God knows how many levels and stacks of crap there are to make that thing happen, which a lot of the engineers themselves don’t even understand.
Peter Wang: Yeah. And to be clear, I’m not saying there’s been no innovation. I’m just saying relative to how much money we sink into it, gets really terrible. When we look at the job ads in San Francisco for entry-level front end programmers, you just look at what they spend most of their days doing, is garbage. They’re reinventing Windows 2.0 APIs. If the browser was actually a UI framework instead of some attempt to present information, people could build apps with a hundredth of the effort.
Peter Wang: Hardware engineers in Intel and Nvidia and other places, they spend so much time making these perfect chips and they all know that people on average don’t even get 1% of the maximum computational power out of their CPUs. And I know I’m just winching now, but I would say that we have way overshot on hyper specialization and we have not made much progress in thinking about building up from a position of economy. Like if we only had five FTEs to throw out this, what is the simplest way we could build this and in a robust and resilient way? That’s the kind of thinking that we don’t see very much, I think, in the software world.
Jim Rutt: Back to the earlier part of our conversation where I threw out the idea of the best solution doesn’t always win a technology. I was actually quite interested in, in the early nineties, at the time that HTTP HTML was starting to take off, and I was gagging on it. It’s bad now. It was horrendous then. Sun had a very cool alternative called News, which essentially was a windowing environment in which the native markup was postscript. You could do amazing things by sending postscript to these windows. That even to this day, you can’t do an HTTP HTML in a reasonable fashion. You’ve got to all these horrendous AJAX, E. Klugs, it’s still unbelievable shit shower, the HTTP HTML world and it’s somehow News had one that brief horse race in 1991 or ’92. We would be in a very different world where people wouldn’t be trying to do the equivalent of moving telephone poles with chopsticks, which is what I think a front end development is these days. It’s just, why is it so God damn hard to do things that should be easy?
Peter Wang: Yeah. For all of the wedging that I do about Trump and everything, nothing drives me to drink faster than reading people complaining about, looking at people trying to build stuff on the front end and seeing how much reinvention there happens of stuff that I was doing, God, in 1993. It’s like, why is this hard and how many hundreds of megabytes are being thrown at making an enumerated list that collapses? Why is this hard? Why is this a hard thing? It’s just, like I said, nothing drives me to drink faster than that because that is really, well, I think if we go through it, if we don’t drink and we stiffen up and we say, well, what is this really? What is this really happening here? This brings it back to the little thing. This brings me back to actually a little bit when you talk about enlightenment versus pre enlightenment and the left and the right of banding stuff.
Peter Wang: I think we are at the coming to the culmination or at the culmination of a cycle of glut. We’re just at a point where the city, when the Mongols sacked Beijing, the city was overflowing in riches. Civilizations and cultures and tribes, when they get rich and they think they’re in control things, they get fat and sloppy. And you’ve just put all this energy into excess that has nothing to do with the hard economy of like when you’re frontier person. I think Jordan calls it the Arc Heidi Smith model of the domestic, the feral and the wild type. And I think we have just a whole lot of old domestication, a whole lot of mandarins and these front end devs that spend three weeks putting together risk control, that’s just a symptom of us having essentially turned off gravity and allowed these things to grow, these [inaudible 00:57:07] to grow too high.
Jim Rutt: Of course, the money signal says it’s worth doing because even an incremental improvement in efficiency in some crappy, big ponderous corporation is worth more than it cost to do. And if it’s even just a little bit more valuable than the cost to do, it ends up getting done even though it’s some higher level. It’s a ridiculous thing to do and that comes from our signaling systems. That’s actually a good pivot to the idea of non rivalrous economics. For our audience, I’m going to introduce a non rivalrous economics a little bit, and get your reaction to how you think it might be part of the solution going forward or maybe not. This is a concept that again, our friend Jordan Greenhall, now Jordan Hall, was the first person I’m aware of who brought forth, or I believe there were a couple of precursors, but he’s the one that in my space at least made it real. And that’s the idea that we have missed the thought that there’s a big difference between rivalrous economics and non rivalrous economics.
Jim Rutt: A perfect example of a rivalrous economic item is a ham sandwich. Either I ate it or you eat it. We can’t both eat it. A classic example of a non rivalrous economic artifact is a MP3 file of a popular song. It’s essentially in this day and age, free for us both to have it. Our economic model has not really taken advantage of the fact that we could have a lot more distribution of non rivalrous economic goods than we do. For instance, many medicines fall into this category, very inexpensive to actually make the pill. But when you have to navigate through the intellectual property levels and the corporate ownership and all this stuff, these two cent pills suddenly costs $200 and that means a whole lot less people are able to take advantage of that pill than would otherwise.
Jim Rutt: On the other hand, and you point this out on the, what was the term you used for under supported libraries and stuff? What a non rivalrous or radical non rivalrous approach misses is there has to be a loop back to incent creation and improvement. And that part has not really been thought through yet where there is some economic signal that comes off of the distribution of non rivalrous goods that goes back to incent those people who created the non rivalrous good though it doesn’t make them billionaires. And also sustains a reasonable level of maintenance and improvement in those goods.
Peter Wang: Yeah, that’s a good way of talking about it. I think that definitely captures it. I’m just making some notes here while you’re saying that. I think maybe this is one of the other things, one of other controversial things which I think we should just do away with copyright altogether. I think copyright is this radioactive albatross that we’re dragging into the information and digital age and I think it would be better if we got rid of all together and people from demand side actually created the poll for curation and for dissemination of popularization. But that’s just neither here nor there. But that’s just one example. Because you said the MP3 file. I would say that for me, my view on economics is from the perspective of dilettante and amateur. I don’t have a degree in this stuff or anything like that.
Peter Wang: But as far as I can tell, economics as an activity for trade and specialization or how the route aspects of economics and those two fundamental activities are so different. When we talk about the space of objects and ideas that give rise to non rivalrous interaction dynamics that it’s almost worth asking, is the term non rivalrous economics, is that term itself an oxymoron? That is to say, is economics really as we think about it fundamentally only useful when we have these rivals, scarce or scarcity mentality things? And if we move into this metaphysical space where non rivalrous things can happen and predominantly it’s not rivalrous things happening, is that worth talking about economics at all? Is it more about coherence and residents? Is it more about phase resonance and phase constructive and destructive interference? Maybe those are better terms.
Jim Rutt: Frankly, I guess I don’t really care what we call it. I would still say that it’s the technology, or it’s the toolkit for creation and distribution. And yes, we call it economics in game A, but there has to be an analogous toolkit that is coherent for an alternative mechanism to emerge. Whether you call it economics, or you call it something else, I don’t really care. There are some people that have done some interesting work on analogs to economics and the game B or an alternative world difference and see how that Daniel Schmachtenberger and his four essays on his blog about our future economics is a good place for people to start. Right?
Peter Wang: Yeah. The reason I bring that up is just because the way I see it viscerally, where I have to deal with this stuff on a very practical level is in the intersection of the open source software communities and the business world that’s starting to adopt to that as the fundamental building blocks. The business world still thinks about software artifacts as being valuable. It still thinks about its employees on the basis of a labor economics model. And when we look at what’s made the opensource software movement successful, the values are completely different. The units that we talk about, people talk about the values that are espoused there are completely different set of values.
Peter Wang: I think if we really want to pull this back to the metaphysical basis, there’s not that much that people fundamentally have to tokenize, to parcel out, to quantize and give to other people that can constitute the basis of an economic system, or at least one that’s worth talking about. Because if it’s entirely bespoke in any given pairwise relationship, then there’s no model possible that we can build. But if it’s something we can abstract, some interaction dynamics wouldn’t obstruct outside of that, then maybe we have the basis for an economics. But I don’t know, I just don’t want to put too much technology and math into something that is a very deeply human inter human kind of activity.
Jim Rutt: Certainly, I think talking about open source, one signal, it’s not a fungible signal, let’s say a completely differentiated signal, is the reputational signal from working on an important or even an unimportant open source project. A part of your resume, and it certainly impresses me when I’m looking at somebody. I was a contributor to project X for the last seven years. Not only does that show that they know how to deal with other people, but they’ve actually pointed me to a significant body of their work that I can go take a look at. So there’s an example of signaling but does not use a fungible token. Maybe that’s the way forward, is realizing that the signaling may not be in terms of simple tokens.
Peter Wang: Yeah. Because at the end of the day, the fungible token aspect of economics is just there to scale transactions beyond the zone of trust. Because if you’re [inaudible 01:04:14] cows and chickens for fishes and fishing nets, your zone of trust is like your clan, or a tribe level. But if you want to trade silks with people across the mountains, you’ve got to have something else. So the tokenization as a way of scaling transactions beyond trust barriers is fine, but that’s not as useful when it comes to these kinds of activities. It’s almost like every project has its own kind of token.
Peter Wang: Every person has their own kind of token. If you get a thumbs up, and it actually our software systems, they don’t create affordances for this actually. It’s really unfortunate. You look at a project like, Oh, this project only has three stars. Another one has 10,000 stars. But the ones, the 10,000 stars, some professors taught a bunch of students, they all thumbs up his project. But this one, three stars, it’s like we dove in Rawsome and Linus Torvalds and somebody else, some big computer science person. Well guess what? That three star project, it’s worth a whole lot more. So these reputational systems are, I think, certainly a way forward. I do wonder if there’s better ways that we can just scale interactions beyond the civil naive web of trust model. But I don’t know.
Jim Rutt: It’s certainly an area where there’s a crying need for innovation. Think about an emergent bottom up filter that could be there on Facebook. One of the things that drives me crazy is that Facebook provides no real tools for nuance, full response to content that flows by. I’d love to put every piece, not every, ones that either pissed me off, or I really liked, I like to place them in some two dimensional space of quality and importance. If Facebook just allowed that and then use the little bit of tools for smartly aggregate them and doing some communitive structure across those, we could tune up a much better flow what we saw than this current crazy world where the only signal is like, like.
Peter Wang: Oh my God. This could be a whole series of podcasts because the thing is that people don’t understand just how much a little bit of a software tweak can have an absolutely massive impact on the usability of communications and infrastructure. Building communications and infrastructure is such a nuance thing. It’s such a nuance thing. And then we just have interns go and try this and try that. It’s absolutely, I don’t know what kind of malpractice it is, but it’s some kind of malpractice. I mean, it’s ridiculous. You’re talking about the things that we would like to see on Facebook. There’s so many things. There’s so many things we can do on Twitter and Facebook that would, they’d tweak Twitter to talking about hiding retweets and likes as a thing. You think about that. Or if you added a downvote button, or a dislike button. Or if you had simply socialize K files. So every single person on Twitter has the ability to add block lists and turn on essentially some weight of all of the people they follow their block lists.
Peter Wang: That would instantly, instantly change dynamics of the vitality on Twitter. But of course, it would screw up their evaluation, but it would create a much better conversational space. And all these different tools, all these tools we might use, what they would do is they would give the end user the ability to participate in a conversation around the norms of that communication space. One of the things that I talk about is that every interaction between two people is a space. Every conversation is a space. And the space has norms.
Peter Wang: The problem with the social media platforms that we build, we call them social media, they’re really attention marketing and virally platforms. But they’re there to… They’re completely unnormed. They’re unnormed spaces, which is why it’s so screwed up. And it’s just deeply, again, I don’t know what kind of malpractices, I don’t know what the term to use in the future. We might have someone write a PhD thesis about the nature of harm and communications and the way that these things are perfect examples of bad design. But I don’t have the words for it right now. But it’s something I feel very deeply and viscerally.
Peter Wang: I had this idea actually last year about putting in a concept of equal time. We used to have this concept of equal time on the airwaves and now you have places like Twitter and people on Twitter with larger followings than the broadcast networks used to have. And there’s no equal time concept, but they have the same kind of reach. How’s that right? How is that going to be a good thing? All these UIs systems that we build, they have so many millions of variables that we could tune and tweak to make for much better for meaningful human interaction. But none of them are incentivized to do this. None of them are monetized in a way that would be aligned with that. It’s a really terrible situation.
Jim Rutt: The unspoken attractor is the fact that the way they make their money is selling ads. And when the world went down that route, this shallow but massive exploit is exactly what one would expect. All they’re trying to do is optimize eyeballs x attention x excitability, and then sell that to people per switch basically. The alternative model, I thought about this a lot, but the world really just seems to reject it, is suppose we had a discourse platform that you had to pay for a little bit, five dollars a month, $10 a month, they’re essentially a trivial amount. It was built on the principle that we are not here to maximize your eyeballs or maximize your twitchiness. We’re here to maximize the value you get from the system. What a concept? And we would build in many of these clever networks, smart tools so that we’d all have a better experience. And because the company is paid by the monthly subscription, it actually has the incentive to keep you off the system as much as possible so long as you got the maximum amount of value. Isn’t that the correct incentive?
Peter Wang: Well, I think, again, in terms of leverage of software, we think now very much on the internet. One of the legacy things that we’ve heard from the success of the frozen accident of the internet is this thinking that these must intrinsically be client server, and they don’t have to be. We can create a very powerful peer-to-peer communication networks. Now, we still need some level of transport level centralization for each optimized for low latency and resilience and things like that. But at the end of the day, people are talking to each other. People don’t have to talk to a server. So a lot of the ideas behind the decentralized web are really around creating these tools that don’t, they don’t impose a client server model. And the reason that’s so important is because the instant you have a client server software architecture, obviously as the app maker and the developer, you’re going to run the server for everyone.
Peter Wang: Now as you scale, you naturally carry this drag with you. Every user you get is another liability, and you’ve got to keep a session alive. You’ve got to back up their data, you got to delete it if they ask you to, if you’re compliant with GDPR, all this nonsense. So the pernicious thing about client server architectures is that they force successful businesses in that environment must have scaling problems. They must need it and monetize every user’s use because they’re taxed by bandwidth costs. Then you’re looking at putting either a monthly charge in or selling ads or something like that. But what if you actually just created a system that didn’t do it that way? What if you created a system that just had users have their own data? If they want to back up their data somewhere else they can, and to share their data at one time with someone else, the data moves over there. That’s a very natural system.
Peter Wang: No server needs to be up. If I’m sitting next to you at dinner and I share a picture with you, it doesn’t have to roundtrip to Cupertino or Mountain View. That kind of model dramatically increases the amount of bandwidth available. It dramatically decreases the cost to build new innovative types of software. So I think that is why I’m so gung-ho on pursuing the decentralized web technologies as the next generation of essentially a portal to a new land where we can create new kinds of information systems.
Peter Wang: It’s not just about decentralized Facebook, it’s about creating better Wikipedias, creating better shared Jupiter notebooks. It’s better everything when we don’t force these centralized surveillance where ad supported kinds of companies. And I don’t believe because these companies are now hugely public, worth 50 hundreds of billions of dollars, they essentially are bound as fiduciaries. Their directors cannot have them pursue models that destroy their market capitalization. I have basically deeply pessimistic outlook on any of them ever reforming their business models. I just don’t think it will ever happen. So criticized by creating, that’s why I’m putting my efforts and some of my own dollars even behind the decentralized web.
Jim Rutt: We’ll turn to that next, but I want to make a little sidebar here. Maybe it is feasible to have subscription based services at scale these days. I went and did the numbers and Facebook’s revenue per month per user is 1.99. And built into that 1.99 is a bunch of profit. They have a very high profit margin and a large cost to operate their advertising infrastructure. So they could operate probably as a profitable business at 1.25 per month per user. That seems to me a number that is so low that maybe just maybe particularly if the payment infrastructure were painless, we could finally get people willing to actually pay for a service so that the service job is to make them happy, not to exploit them.
Peter Wang: A couple of things. Number one, that number is incorrectly low in North America, Canada, basically in Canada, the US I think it’s something like $15 to $25 per quarter, so it’s much more than that actually. Much more than the number you quoted. Furthermore, that’s on average. Keep in mind, Facebook is a targeted advertising platform, not all users are created equal. The users who can pay $25 a quarter or a hundred dollars a year for Facebook, they’re probably worth $500 a year to Facebook because they’re the ones with money. The poor users who don’t have that kind of money, who are at the low end of that, they’re worth less. So it’s not all equal. I had this explained to me because I did the same math. I was like, crap, what if we just had a big old campaign to get Facebook just to charge us?
Peter Wang: And I was like, no, no, it’s not going to work out. And the reason is because of that dynamic. And second of all, because you might be willing to pay five bucks a month for Facebook. I bet you, you’re not going to pay 500 bucks a year for Facebook. It isn’t that valuable. That’s the problem there. And tied to that then is their market cap. At the end of the day, they are worth a certain amount of money on the market. If the directors of the company go and sabotage the company’s business model because they want to be nice people, they’re going to have shareholder lawsuits for the next 10,000 years. They cannot do that. Not unless there’s a serious antitrust action or something.
Peter Wang: Can they then say, we’re going to pivot our model. We’re doing this, we’re going to banking. We have all this creepy information on people, we’ll use it to score their credit instead of selling ads to them. So that’s why they’re doing this all Libra stuff, is because they’re trying to pivot and to see if maybe that’s another area to do this stuff. But at the end of the day, I think it’s fundamentally philosophically inappropriate to be the central entity holding all of this information on people. So any business model tied to monetizing that privileged information is going to be immoral and problematic in its own way.
Jim Rutt: And I agree with you that the incumbents can’t do it, but I would just hold out to you, young entrepreneur wannabes that maybe the time is right. Because if you can get two dollars a month from people, and you don’t care if they’re poor or rich, and then you can support a service that has the level of complexity of Facebook at least at scale where the scale transform is, I don’t know, but I bet it’s a shitload lower than it used to be. Now that you can actually build your backend on cloud services and don’t have all the high fixed costs or running your own data center, it might just work.
Jim Rutt: The attack with a different business model might bring them down, particularly because it would preferentially, I would pause it, strip away their most valuable customers. It’s the people that are willing to pay two dollars or five dollars a month that as you point out, might be worth $50 a month. As those people migrate to the cheap paid platform that has a pro-consumer rather than an exploit of business model, that could be very destabilizing. Just a thought for you, young entrepreneurs out there, go get them.
Peter Wang: Well, the other thing to keep in mind is, there is a [inaudible 01:16:44] to this or a path dependence. It took Facebook a lot of effort to get everyone onto Facebook. Now that everyone’s on Facebook, when the next platform comes along, all it has to do is, hey, click here to invite your Facebook friends, button. Or click this, share this link on Facebook, have your friends join you on this new network. That’s doable. That kind of thing is doable. So people can bootstrap the next network off of Facebook the way that Facebook couldn’t have bootstrapped of Myspace or classmates.com, God forbid. So I think there’s definitely the opportunity there. Again, I wouldn’t be investing in this space if I didn’t think it was possible to change it, but it is going to be a process.
Jim Rutt: Let’s move on to the next item. And that is, as you hinted at, you’ve been a significant player in thinking about an alternative distributed network architecture and how you and I’ve talked about some really radical stuff, ideas of yours. So go riff on the idea of a distributed internet and distributed web and all the cool things that can be done on it.
Peter Wang: Okay. So the basic idea actually came to me, not because I’m a crypto anarchists from the nineties or anything, but actually it was a few years ago that I had this realization that almost all of the data that I get in the space of a year, that I download, read, consume, whatever, all of the content that I get could fit on a flash shift. The size of my pinky nail. That would pretty much hold, maybe barring like streaming 4K movies and things like that. But most everything else, the news, the fake news, everything, all of it could fit on a physical device the size of my pinky nail. So I thought about, wow, now that’s of course, getting it all at once, it’s different than getting it on demand. But still, it got me thinking about, we put a lot of infrastructure in place to do this client server stuff when really the content sharing piece could be done in a way that thought at all to lots of these other things. That went down this rat hole of like looking at things like IPFS and looking all this other stuff.
Peter Wang: I emerged at the other end realizing that there’s a few pieces that we need to actually create. We create a radically different kind of approach, which is basically to say that the client server approach of HTTP is busted. It inherits all of the problems of TCPIP, [inaudible 01:18:55] IP itself, and a client server architecture fundamentally creates asymmetry. The user does not have control and access over their own data. So then I also realized that the lack of identity system on the internet has been a problem since day one. Since we started getting into spam problems. So we use email, we use phone numbers, which all these are identities that we rent from various centralized services. Very few people can run their own email server anymore.
Peter Wang: It got me thinking, I came to this realization that we can create a radically different architecture that’s actually very powerful, that maintains all the modern things that we’d like to see in our web apps and stuff. But it separates out the interaction of an application with the data and then the transfer of that data to a sharing it with other people. So we have a point to point system for people to share their data with each other. We then use some modern techniques like CRDTs, and some other hash blockchain Merkle trees base things. We can create decentralized infrastructure for running apps that are eventually consistent. That’s almost like, imagine if this is, maybe the best way to put it is this. Imagine if visiting a website to view, going to someone else’s server to be a webpage. Instead, what we did is we made a two step so that we have a Dropbox and I share my Dropbox folder with you.
Peter Wang: So they’re synchronized whenever they can be. But I then point my web browser to just look at my local Dropbox. If I want to edit a file and save, it saves it to local storage. It saves in the Dropbox, eventually that gets synchronized to your Dropbox and then you open up and look at it. And just by separating out that application, interaction with the state and then the transfer and merge of state down the road, we can create a fundamentally different architecture that then doesn’t put a Google or Facebook in the middle of things. Furthermore, if we separate transport of data from the application communication to data, we can create mesh networks and local things. For instance, I had this incredibly frustrating experience when I was on a flight, an international flight with my family and I had a video file on my phone that my kids wanted to watch.
Peter Wang: I was on an Android device, they’re on an iPad. And between these two devices, I could not without, there was no wifi, there was no internet. I could not transfer, actually it wasn’t a video file, it was like 30 megs of MP3s. I couldn’t transfer these songs from this one thing to another without being connected to some centralized data center. And that was just absurd. But if we were to use local mesh based technologies, my phone and the iPad could communicate at gigabit speeds and I could just transfer that to the device there and they could just play it. So it’s a completely different way of thinking about applications, about thinking about the relationships between application authors and the data and the users and their data. And that’s why I support funding the Beaker and the Dat projects and it’s an alternative way of building a browser and application structures and user authentication and all these things. It’s an alternative way of building essentially an internet infrastructure.
Jim Rutt: Let’s tell the people where they can find out more about Beaker and what was it, the other one was?
Peter Wang: Dat. Yeah. So it’s the datproject.org, I believe, and beakerbrowser.com.
Jim Rutt: Dat, D-A-T?
Peter Wang: D-A-T.
Jim Rutt: Tell us a little bit about those projects. What do they do?
Peter Wang: Well, so Dat is, you can think about it as essentially like a peer to peer [inaudible 01:22:16] like protocol for doing decentralized Dropbox. So it’s a blockchain based system, and you can create a Dat archive of anything on your drive. It’s basically a shared folder that automatically syncs with other people who have a copy of your shared folder. That’s the Dat project. It works like IPFS, but there’s not an attempt to monetize. There’s no file coin monetization model built onto it.
Peter Wang: Then the second thing, the Beaker Browser, it’s a browser for the decentralized web. It looks at your local Dat archive, it runs a full, I mean it’s a full web browser, so it can speak to Htp://standard internet sites. You can load up Facebook, you can go to Twitter. Everything works great. It’s just a web browser, but it can also load Dat://URLs, which then are 32 byte hashes or 64 byte hashes that basically go and create, or they create on your local machine, a mirror of a particular shared archive and then it opens it up on the local machine. And you can make your own copy of it, you can edit the code, you can edit the source code of the application that’s being served up.
Peter Wang: They’ve built chat apps on top of this. They built like a Slack thing, they’ve built decentralized Twitter. Even a basic decentralized Facebook like thing. And the whole idea here is that websites now become central. The websites again becomes central to the experience. That data itself becomes central to the experience and applications are just lightweight things that sit on top and give you different kinds of UIs. They’re no longer walled garden platforms that you fork all your activity and data into. Your data is still your data and apps are just lightweight things that manipulate it and show it to you in different kinds of ways. It’s a really wonderful system.
Jim Rutt: Have there been some winners yet? Again, as we know, each one of these new platforms that take off, usually there’s a single winner. The Apple took off with VisiCalc and the IBM PC took off with the Lotus 1-2-3. The Mac took off with the various graphics tools that were created for us. In your mind, a application that has done so well on these new platforms that it’s compelling?
Peter Wang: No, not yet. It’s still early stages on this stuff. It’s still early days.
Jim Rutt: Take a look at it, I would encourage other people who are thinking about something new. I mean, I remember how exciting it was, the web in 1992. I can’t even make myself think about the web anymore. It’s so old and horrifying. So it’d be fun to actually spend a few cycles looking at something entirely fresh. I’m sure there are a lot of other people out there who feel the same way.
Peter Wang: Well, it is nice to have had a privilege. It has been a privilege to have ringside seats watching a generative open collaborative communication space get punched, beaten and cut down into a mere broadcast system for copyright materials. It’s really been something fascinating to watch. But that’s a long conversation to talk about the OS technical debt between the operating systems that led us to the browser winning as essentially the winning middleware, which then led to this kind of thing happening. But that’s maybe a different conversation for a different time.
Jim Rutt: Yep. I think, in fact, we’ve come near to the end of our time. Do you have any final thoughts?
Peter Wang: I just want to say it’s a privilege to have this conversation with you and thank you very much for having me on the show.
Jim Rutt: And I have to say, I had high expectations and you exceeded them. This has been an amazingly both broad and deep conversation, which is the goal of The Jim Rutt Show. Which is, we say is real thinking about deep ideas. So thanks very much, Peter, and I hope to talk to you again soon.
Peter Wang: Thank you, Jim.
Jim Rutt: Production services and audio editing by Staunton Media Lab, music by Tom Mueller at modernspacemusic.com.