Can decentralized social media address the ailments of the dominant, centralized model of social media today? What are the big debates in this nascent and rapidly evolving ecosystem?
We all know about all the things that are wrong with social media as we know it. It's a model where everything is centralized and the market dynamics foster very few competitors who just get bigger and have more power. Who builds? Who governs? Who owns? Where are the servers? Where do the applications sit? Where does the data live? It's largely in just a few companies that in many cases have walled gardens of content. This dominant model yields a tremendous centralization of power for tech executives who are incentivized by market dynamics, to grow and to extract and to capture more and more of our attention. So, how do we get ourselves out of this mess? Does the answer lie in decentralized social media?
Our guest for this episode is Evan Henshaw-Plath, veteran Silicon Valley engineer and CEO of Planetary, a decentralized social network.
In this episode we cover all things decentralized social media, including:
“If we relied on the government, be it Twitter, or Facebook or the nation state, they're always going to choose to default to people who have more social power. If we want a more social justice agenda, we need a system of governing anti-social behavior and anti-social speech that doesn't rely on the player's overall social power in the larger economy. But instead relies on the ability of the people in these systems to report, have transparency, set governance and enforce decisions amongst themselves." - Evan Henshaw-Plath
[00:00:40] JS: That's Evan Henshaw-Plath, Silicon Valley engineer who's on the front lines of Twitter when it was launched in 2006. And this is the Becoming Denizen podcast. I'm your host and curator, Jenny Stefanotti.
Evan's our guest today for an episode on decentralized social media. He's an entrepreneur and engineer who's worked in everything from small startups that he's founded to large corporations like Yahoo!. He has spent many years thinking about the things we discuss in this episode. He was actually one of the first employees, and lead developer and architect at a startup called Odeo, which is best known as the company who created Twitter. He was quite literally a pioneer in the social networking landscape, and is currently the CEO of Planetary, a decentralized social network.
In this episode we covered all things decentralized social media. This is a wide lens deep dive where we try to give you a full snapshot of the topic. We cover a framework for thinking about decentralization across the internet, as well as social media and social networks. The distinction between Web3 and decentralized web. What a protocol is and why it's a core element of understanding decentralized social media, and what the current landscape of protocols looks like, major players across the ecosystem, big debates in the decentralized social media space, and Evan's vision for decentralized social as well as what he's up to with his startup, Planetary.
These episodes are intended to help you walk away with a big picture understanding of an important topic like this one. There are additional resources on our website, www.becomingdenizen.com. There you can also sign up for our newsletter where we bring our weekly podcast to your inbox alongside other relevant Denizen announcements. Denizen has partnered with many amazing organizations at the forefront of this conversation. We also bring important announcements from them to our audience in our newsletter. And with that, I hope you enjoyed this episode.
[00:02:24] JS: We all know about all the things that are wrong with social media as we know it. It's a model where everything is centralized and the market dynamics foster very few competitors who just get bigger and have more power. Who builds? Who governs? Who owns? Where are the servers? Where do the applications sit? Where does the data live? It's largely in just a few companies that in many cases have walled gardens of content. And you have a tremendous power of tech executives who are incentivized by market dynamics, the stock market, to grow and to extract and to have more and more attention. So, how do we get ourselves out of this mess?
So many people want to get off Facebook, but there's nowhere to go. I have been saying for a long time, Evan, tell us what the answer is. And then we'll just tell everybody to jump.
It feels like now is really a moment for decentralized social media. In doing my research, there's a TechCrunch article over the summer that talked about the race to build the social network on blockchain pain. Twitter's been working quietly on their own decentralized social protocol under Bluesky. They just came out of hiding and launched a platform, or launched a space for developers to contribute to that. It feels like there's a lot happening in the market now.
But I want to go on the arc of your story, Evan. Because you were there at the dawn of Twitter, which is basically the dawn of social media as we know it. And this is what I find so fascinating. You helped brainstorm an earlier version of Twitter that could federate rivals into a decentralized system, which is precisely the conceptual architecture that we now want to move to today. I would love for you to tell us about those early – that moment. That moment when it could have gone in a different direction. And tell us why Twitter evolved into the version and we know it as instead.
[00:04:15] EHP: That's a fascinating story, because we think of the technology that we have now as in some ways preordained. But there's strong factors that made it the way it's structured now and don't think about the fact that, in many times, there's alternative paths in history that could have been very easy.
In the case of Twitter, we were very much in the space of Web 2.0 and the idea that the web should be driven by content that people create and that people need to be central to it as opposed to static web pages or webpages that are only produced by professional content creators. And that was the big innovation.
And O'Reilly media gathered together a Foo camp up in Sebastopol, and they called it the Social Graph Foo Camp. And we had people – we had the senior technical side from all of the social media companies that existed at the moment, Facebook and a number of others. And we spent a weekend talking about, "Okay what is the emergence of Web 2.0? What is the emergence of social media? What are we doing with all these things?"
And it was the first time that we had all of the technical leadership from every social media company in the same room. And I think this was in 2007. So, pretty early on in the days of these things.
And we basically had an idea saying, "Well, what if these things work together? What if you could connect between them and they weren't walled gardens?" We didn't want to go back to AOL, CompuServe. Who wanted to go back to that world?
And so, we actually stayed up all night and built integration between Jaiku, which was in the process of being acquired by Google. And Twitter – So, the thing, we hacked together a system over XMPP with a Jabber chat protocol, where you could literally follow people in real time across services. And it worked. It worked. As in, like, we made a prototype work. And we had it running live in production. For some Twitter accounts, you could follow users on Google's Jaiku and back and forth. And that was super exciting.
And Blaine Cook, who was the main Twitter person who had been pushing this and doing the development – he went to Fred Wilson over at Union Square Ventures and said, "Hey, Fred, this is where I think Twitter should go. We need to be a protocol, we need to be this very large player in a larger ocean of social media. And we need to play nice with a larger internet." And Fred, he was, I believe, chair of the board of Twitter at the time, signed off on it. He said ,"Yes, I like this vision. This is where we should be going." And Twitter set about working on it.
Now, remember, Twitter at this point has a dozen people at the company. And the site is crashing all the time. It's one of those moments where it's scaling like crazy. There was one moment where if the site crashed during the day, there was no way to reliably get it up until the middle of the night because that was the lowest traffic point.
What happened was there was never a direct decision at that moment to not do the federated thing. It's just that Twitter's engineering team was stretched very thin. And eventually, there ended up being a personal fallout between Jack Dorsey and Blaine Cook, where they got into an argument about how to structure the engineering team. And who was doing enough work? And everyone was stressed out.
And Jack fired Blaine because he wanted to run the engineering team in a different way. The Twitter website was essentially down for the next three months because Blaine was the only person who knew how it worked. And the effort to get it back up is when the whole effort on making Twitter federated was dropped.
[00:08:53] JS: Got it.
[00:08:54] EHP: It was like no one said, "Oh, we shouldn't be in this protocol." Twitter continued. And instead of being federated – this is from 2007, 2008, through 2011, 2012. Instead of being Federated, Twitter had a very open API, and said that anyone can make a Twitter app and they can make the app designed any way they want. Any app developer can make their own algorithm for what the timeline looks like. And there was a tremendous flourishing of Twitter apps those early days.
And people didn't use twitter.com. There were hundreds of Twitter apps all made by independent developers. And everything that we think of as Twitter was created by those developers and users. So, retweets, inline images, links, tweet streams, app replies, real-time search. It all comes from the fact that Twitter was this relatively small core that other people would build on.
And that existed until Twitter basically didn't figure out how to make enough money. And so, an entrepreneur named Bill Gross with a company called Idealab came in, and he was going to buy up all the Twitter clients and then get the sort of leading tweeters, like the people with lots of followers, to cross post to UberMedia. And then eventually have it so that when you used his apps, TweetDeck and others, that it would post UberMedia and not Twitter. And so, basically he was coming in and trying to use external capital to do a hospital takeover of the ecosystem.
And Twitter responded by panicking and basically shutting down the open API ecosystem, buying TweetDeck, which was going to be bought by UberMedia and rolling everything in-house. And so, Twitter at one point in 2007 was federated. But up until 2012, it was much more like a protocol than a centralized service. It wasn't until Jack returned as head of product and eventually his CEO that Twitter figured out how to be profitable and figured out how to survive.
And that's why Jack, and Prague and other folks sort of attempted to – Jack called it Twitter's original sin, that it went away from this open protocol system and became a centralized company. And so, that's where Twitter's Bluesky project comes from. And that's also where my company Planetary comes from. Blaine Cook and Biz Stone are both investors. And so, we basically took that vision of, "What is this open thing? What if it were an open protocol? What could you do that way? And how do we build it in that way?" As opposed to building it in a way where a single company or product team needs to make the decisions.
[00:12:13] JS: Yeah. I want to get there. And I'm excited to get there. But I want to take us on a little bit of a walk through decentralization and Web3.
[00:12:21] EHP: Absolutely.
[00:12:22] JS: And just paint a picture of how to think about decentralized social broadly. And then we're going to hone back in on more specifically what you're doing. Right now, the model is centralization of all the things. The team that is building the software, the servers, the applications itself, the data, the governance, all of that is central. And I think it's helpful to think about decentralization across that stack, right? Because in some cases, decentralization of the team can mean that it's open source and in the decentralized social networking space. So much of that is based on open source models of decentralized development, right?
Then there's also questions about where do the servers sit? And you see this. And we're familiar with this. And in crypto and blockchain, you've got a distributed ledger. There are servers. You can add a node to that. And you can host your own ledger, right? There's also questions about, in that infrastructure, where does the software live? What servers does the software live on, right? And it's interesting to see where there's some centralization of that, the decentralized web, Web3, space.
And then also questions of where does our data live? And there are different models of that. Sometimes it lives on the blockchain. Sometimes it lives on a centralized server. Sometimes you can opt into that. Or sometimes it lives on our phones.
And also, really important questions of governance. Who governs? Right? Is it that we actually have models of truly decentralized governance of some of these platforms? Or is it the case that we just have more competition to opt into different sets of content moderation practices or algorithms, et cetera?
And so, I think it is helpful because you can look at – as you as we look at the landscape we get into today, the landscape of decentralized social media, social networks, you get different answers for how much decentralization there is across these variables. And then I also think it's useful to think about a distinction between decentralizing versus decoupling, right?
In some cases, it's the coupling, for example, of data and applications that sometimes creates a lot of the issues. Or the coupling of governance and creation of the content. And so, in some cases, it doesn't necessarily have to be that some of those very – the variables are decentralized. But it might be just decoupling the entity that has control and makes decisions over it. I just found that framework to be useful when thinking that. And so, I just think holding that as we start to look at the landscape and all the things that are out there.
Before we get into decentralized social, I want to talk a little bit about the decentralized web. You sent me a really great article about the distinction between Web3 and the DWeb. Can we talk about that for a moment? I know there's not consensus here. But I think that teasing out the debate and the distinctions to the extent that people see them I think is quite useful.
[00:15:18] EHP: We're at a moment of a lot of experimentation. There's a lot of people trying a lot of different ideas and a lot of different ways of solving these problems. And so, in some ways, it gets grouped into this broad category of decentralized protocols.
[00:15:33] JS: Okay. We're going to get into protocols in a slightly geeky way in a minute. But for those who aren't technical, what's a protocol?
[00:15:40] EHP: That's a great question. A protocol is a set of agreed-upon standards by which two or more pieces of software can interact between each other. When your web browser talks to a web server, that's speaking over the HTTP protocol. When you send an email from your computer to another person's computer, it goes through a set of email protocols that each of the intermediary servers knows how to transfer the data back and forth. It's basically agreed upon set of standards that let different pieces of software interact.
When we tried to build federator to decentralized Twitter, or microblogging as we called, back 15 years ago, we didn't have a bunch of underlying tech that comes out of the blockchain world. There was some peer-to-peer technology that came out of the work on BitTorrent. And there was some distributed programming techniques and things like that. And there was a bunch of cryptography mostly based on PGP.
And what happened was a bunch of people doing cryptocurrency research and decentralized tech developed emerging technologies, the most known of it is Bitcoin, and then Ethereum and a bunch of other blockchain tech. And that pushed forward both the research and the technology that we had around these things and made a whole bunch of stuff possible.
Now, in the space, we have two sort of broad communities of people. We have people who call themselves Web3 and people who call themselves DWeb. And the Web3 folks tend to believe that the underlying technology should be built on blockchain, which should use a consensus protocol, and which should include extrinsic financial transactions or extrinsic motivation into most of the interactions. And there's debates on which things should or shouldn't be paid for.
[00:17:40] JS: Can I ask a quick question? A clarification. Can you double click on what you mean when you say consensus protocol?
[00:17:46] EHP: Oh, consensus protocol is the way in which – say, in Bitcoin, if I'm going to send you money, send you some Bitcoin, a way of verifying everybody else, I no longer possess that Bitcoin. And now, you, Jenny, possess that BitCoin.
[00:18:04] JS: Got it. Got it. Okay. Consensus across the different instantiations. Got it.
[00:18:08] EHP: Yeah. Everybody agrees without there being a central server that does it. Everybody agrees to it. And the way Bitcoin uses is it used as this thing called mining, which we've all heard about, which does a bunch of mathematical verifications to verify the next block.
And the actual computer science behind it is relatively complicated and hard. But the point is you have a unified, very slow uncentralized database that you can control and then each person can contribute to. That's the Bitcoin world. And the Bitcoin world ideologically has – the cryptocurrency world it logically has roots in libertarian politics. And it's about who controls the financial system and how it works. And is there a central bank? Is there a centralized control of things? Is there fiat currency? And some of the people in that community don't want to pay any taxes or they don't want there to exist a government. But not everyone.
And then on the other side of this space is what people call the DWeb, or the decentralized web. And the decentralized web is sort of the – if there's a yin-yang of these communities, it's the socially progressive end of the same sets of technologies, where each transaction isn't mediated by a set of financial agreements or financial transactions. Instead use the same underlying sets of technologies but without consensus and where people just verify that their other peers are doing things in order to balance things out.
This gets back to BitTorrent. Remember, when you would download a movie from Bitcoin and it would show – Or an MP3, or show how much you downloaded and then how much you contributed back of uploading to others? The DWeb works that way, which is sort of in a protocol way of saying we're not going to use financial transactions to organize this. We're going to build this decentralized social software around the idea of mutual hosting basically.
And that ideological roots and those technological routes then fall out to a whole bunch of other things. In the the Web3 world, financial transactions are easy. And so, people are like, "Well, I have a hammer," which is financial transactions. And so, every problem will be solved, like spam or anything else, with financial transactions.
And in the DWeb world, it says, "Wow! Look at all this what we've got. It looks like an economic commons. Let's look at Elinor Ostrom's work on governing the commons and say that we need to build into the structure the tools for these communities to have a commons that we can't own, but we all get economic and social value out of."
[00:21:09] JS: And it's worth noting. I think after quite a long period of collective deliberation, recently the DWeb principles were launched, which really gets into a vision for the internet and the role that technology plays in society that gets into in the primacy of human agency of holding security and privacy and self-determinization, interoperability versus walled gardens. Talks about who benefits from it being distributed to the people who contribute to a success or society most largely, writ large or high concentrations of organizational control is antithetical to the decentralized web. Talks about ecological awareness. It's just worth noting that that is also out there. This is really fascinating. And thank you so much for helping us to outline and understand these distinctions because they're so critical.
And so, we have now decentralized social network that almost like it's the wild west of decentralization that works right now, which is playing out on top of these two distinct really paradigms of what the decentralized web should look like that are built in parallel.
Decentralized social addresses these issues about who owns the data and how algorithms are created. I think really critically is that it addresses the fundamental market failure that has yielded the outcomes that we see, which is that there's a network effect associated with social networking. I.E., the product becomes more valuable as more people are on it. So, then the switching costs are very high. There are significant competitive barriers to entry, right?
If you have this interoperability and you essentially have a federation of social networks where you still have access to your social graph, or they talk to one another, then you enable more competition, more choice around just, again, going with the one that's more aligned with your values. And you see – we'll get into some of the examples. You see some of the examples today where you have a platform. And then on top of that there are many different applications of social networking that are interoperating.
And then there's just these different protocols. And I think you touched on it a little bit when you were talking about DWeb versus Web3, right? We talked about what protocols were conceptually. And there's not going to be one thing that necessarily wins out, but you still have impermeable walls between the protocols as we're looking at federations of social networks, right?
[00:23:36] EHP: Yep.
[00:23:37] JS: And so, there are a couple of primary – as I understand it. And you can correct me. There are a couple primary protocols on blockchain. DSO seems to be a really important one. Steem seemed to be a really early one. Steem was an early blockchain social network that had a peak market value of 2 billion, which fell to 100 million today. And then there's what Twitter just announced, which is the AT protocol, Authenticated Transfer Protocol. Mastodon, which has about 4 million users, is built on something called ActivityPub.
Mastodon is a platform where you have different instances of Mastodon that sit on top of it that talk to each other. But then Mastodon as a whole talks to any other set of social networks that are built on ActivityPub. And then there's Scuttlebutt, which is the one that you – the protocol that you build on top of. Not all of them are client-based at the user level in terms of where the data is stored like Scuttlebutt. But we talked about BitTorrent model, right? Versus blockchain. The AT protocol and Mastodon, don't those operate at more of a kind of less fully decentralized at the user level model, but something in between? And are there any major protocols that I'm missing worth talking about?
[00:24:49] EHP: I was looking at – my list right now of protocols is at 74 distinct protocols. And there are many, many protocols in the space. But the protocols, often, they overlap. So, everybody is using cryptographic keys. They use public/ private keys. The public key is your root identifier. Usually you have a name that maps that. Everybody uses a signed blog where you cryptographically sign to verify your content. And then there's different data structures where you point to what the previous and next content is.
On some level, across all of them, we're using a bunch of the same techniques and technologies. Now, the data itself and the where the code runs can be run entirely on device and have it directly be a mesh peer-to-peer networking. And that's the way Planetary works. Or it can run on a server, and that the servers talk to each other. And so, your account exists on the server. And that's the way a federated protocol like Mastodon works.
Twitter's Bluesky, the AT protocol is also a federated protocol. With the exception that Twitter uses a set of keys which the user controls, they can move from one Bluesky's server to another. And so, if you don't like the way Twitter's Bluesky server is using, what the rules are – everything else, you could go choose another one.
[00:26:19] JS: Right. Yeah. Okay. So, you still have some degree of centralization, but there's just more competition that allows you to move around.
[00:26:29] EHP: And other protocols in the space, like Forecaster and Lens are very similar in how they work like that. Another one that exists is Minds, which records the financial transaction, the identity, on a blockchain, so anyone can verify it. But Minds, the company, is the only one who actually makes an app that uses it. So, it's not open in that way. You're dependent on what Minds, the company, wants to build.
And then the last category of stuff is there's a company called Protocol Labs that has a technology called IPFS. And IPFS and the Protocol Labs sits halfway in between the Web3 blockchain world and the decentralized web. And so, they do a bunch of work on the commons. They've built a bunch of important technology around distributed file systems and distributed data, distributed databases. And there's companies like Capsule and Textile who have built on top of this IPFS ecosystem, which is sort of an intermediary between the DWeb and web –
[00:27:36] JS: I know this protocol thing is a little bit into the weeds. But I just felt like it's very important to touch on it and have our community and audience have an understanding, because this is the layer where a lot of this is interestingly playing out right now.
Now, we met at Unfinished. One of the big things happening under the banner of this organization is Project Liberty. And so, they're really seeking to address this. They're building something called the decentralized social networking protocol. It's another one. I know it's a lot of talk and not a lot of actual code at this point. But it's a protocol for writing and reading social network data on the public blockchain using smart contracts. And it defines how identity, social graph and messaging are represented to create a decentralized social network. This is new. I talked earlier about decoupling versus decentralization. This is decoupling your data from the applications that might sit on top of it for social networking. That's an architecture question that is sometimes tacked to the technical protocol itself. Is that correct?
[00:28:39] EHP: Sure. And the architecture matters, because how we design it, do we design it so that you can see who everybody's friends are or not? Do we design it in a way that data is encrypted and you hand the keys over who can read it? Or if it's verified? A bunch of these systems do not support delete. And there's no way to get – and some people see the lack of delete as a feature. Other people see the lack of delete as a bug. Other people allow you to delete data. And there are some protocols where all data disappears after six months. All these sort of fundamental decisions of, "Is there a single name space? Or just IDs? Or does it piggyback on the domain name system?"
[00:29:29] JS: Or do you let the apps that build on top of it make those calls?
[00:29:33] EHP: Exactly. Can I move between one app and another? In Mastodon, I can't move from one app to another. In Bluesky, and in Planetary, Scuttlebutt, you can. And in others, like DSNP, or Lens, or Mind, or Capsule, there's only one app in the ecosystem. So, there's nowhere to move to.
And the ability to – the autonomy to leave is very important. Even if you don't take your toys and go play someplace else, if you can say, "I don't like the way this app is working. I don't like the people who are making these apps. I don't like the way they're designing it. I don't like the algorithms that let me choose or the algorithms they choose for me." You can always download your data. You can have your data. But the data isn't important.
What's important is the social connections. Do I get to take my name with me? And when I take my name, do the people who are connected to me on this network get to see me – Like, continue to connect to me? And if you can't do that, then it empowers the platform versus the users.
And so, in this way, Bluesky is done a lot right, because they have a balance between running everything on your device. The way Planetary does is harder to build. And if you're doing things like end-to-end encrypted messages in groups, it's great. But if you want something that's super quick to sign up, or you want to be able to connect and view the entire world versus an intimate scopes network, it's harder to make an intimate scope networking device. Whereas Bluesky's model of federated where you can choose which of these cloud providers do it let you have hundreds of millions of –
[00:31:31] JS: Helpful. Thank you. Worth noting that most of these protocols are actually open sourced. Free. Any developers can contribute to it. And I think that's a really important thing.
Okay, moving on. What are some of the big debates in the decentralized social space in this Wild West moment? I know we can't drill into them. But I'd love to hear the top line.
[00:31:54] EHP: The top one that everybody asks whenever anyone does anything in this space is why didn't you just use Mastodon's ActivityPub and Activity Streams? Because people look at it and they're like, "There's this thing that exists. And it has a few million users. And it does things more or less right."
And the answer is that Mastodon never addressed the issue of user control, and privacy, and encryption and the ability to build new kinds of apps. And then the other debates are, "Is the social graph public or private? Who can see it? What's the way we work? What kind of zero knowledge protocols do we use for exposing user data?"
[00:32:38] JS: There's an assumption, I have different social graphs on different applications for various purposes, right? I even have different graphs on the same app. I have my private Jenny for friends Instagram and my public Instagram with followers. Is it the case that there's an assumption that there's some universal social graph of everyone that you've ever touched points with that you ever might have connected with on any app and then the relevant subset the social graph for you for that particular application that exists above that?
[00:33:13] EHP: Some of the implementations say that we should just get everybody in a universal social lab and we should expose everybody to everything. And there should be no privacy. And you should just be able to find everybody. And that is the opinion of relatively privileged straight white men who often don't face blowback from things.
What most people do is they actually scope their community based on either accounts or social networks themselves. And so, you need to be able to have many different accounts with different social graphs with different kinds of personas that you present yourself to. And you need to be able to either choose to connect them or disconnect them.
And in Planetary, you can do that. In Bluesky, you can do that. In some of the others, it's not really designed to do that. For example, DSNP, this Project Liberty that comes out of Unfinished Labs, that you write your follow messages in an immutable blockchain that everybody can read. And so, there's no idea of having – maybe you're interested in hentai porn. You might not want your work colleagues to know that. Or they might not want to know that you're a furry or someone who is coming out as queer wouldn't want all of their family members to immediately notice that. We need those divisions. Those divisions are what help us make society work. And so, we need to design systems that allow that to happen.
And at the moment, most users – sometimes people users have like public and private accounts like you do with Instagram. But most users actually use a different app for each different community. LinkedIn for your work friends. And you'll use Instagram for your party crew. And you'll use Twitter for politics and Facebook for college friends. All those kind of things.
[00:35:24] JS: Yeah, yeah, yeah. Okay. Other big debates.
[00:35:26] EHP: Yeah, other big debates. Well, the fundamental one is how do you solve spam and bad behavior? Some people think that bad behavior is simply a financial thing. And so, if you charge for each post, or you charge to spread the audience, or you reward people with financial rewards if they make posts that other people like, that solves the problem.
[00:35:52] JS: That sounds like a bad idea. That sounds like a race to the bottom of the brain stem driven by not just our need for attention and likes, but by financial incentives as well.
[00:36:05] EHP: And I have an opinion, which is that I think that most social interactions shouldn't be mediated by extrinsic financial motivation. I mean, we know that even in the world of work, which you're doing for money, you're motivated to do good work not by getting more money, but by getting other forms of reward from the work itself.
[00:36:28] JS: Yeah, this was actually my next question. You originally were very interested in blockchain and spent some time working in the space and became disillusioned. And this kind of points to the big reason. Can you say more about it? Or is that the straightforward answer? Because everything becomes tokenized in order to incentivize engagement and behavior in the blockchain universe.
[00:36:50] EHP: I wanted to learn more about blockchain. And I joined the company just as they were getting ready to start Y Combinator. They're a company called Quanstamp. I led their engineering team. And we built a test net and launched it. We did security verification on Ethereum. And I learned a ton about how the ecosystem works. Some of which is amazing. Some of it is super interesting. But it also is a space where there's some incorrect assumptions.
The incorrect assumption is that most social cohesion exists through financial transactions. The other incorrect assumption is that financial regulation we have today exists just to reinforce the existing power balances. And it does exist in some ways to do that. But most of the financial regulation we have is because someone was doing something that was unethical or scammy, and we had an economic crisis like 1929 or the 2008 credit default swap transaction, and we created these regulations around our financial system to account for the way the scams are happening.
And what we see in the blockchain world is everything going back to tulip mania in Holland four or five hundred years ago, every single financial scam that anyone has ever come up with in the history of capitalism and markets happens all at once. And because people are like, "Oh, it's an unregulated new space where we can build all these things." There's interesting technology there. There's a ton of interesting projects there. There's also a ton of scammy stuff. And the unregulated nature of it lets you try interesting things, but it also lets everyone try every single scam that's ever existed.
But I found that the immutable stuff and the connecting financial transactions, everything, was a problem. And we could reuse a bunch of the tech without those fundamental political and economic assumptions.
[00:38:56] JS: I think this brings us to your vision for decentralized social networks. Obviously, we want to hear about Planetary. But let's start with just that wide lens of what's your vision for how this would go down to yield the optimal social outcomes? And then let's talk about Planetary more specifically.
[00:39:15] EHP: Sure. As Twitter grew, it became more substantial for us and clearly was having a huge effect on society and other social networks. I don't know if people realize it, but Kevin Systrom, who founded Instagram, was our Twitter – was the intern at Twitter when Twitter was founded. The world of these folks is very small.
As that was happening, I actually went and spent a stint at the MIT Media Lab as a researcher in the Center for Civic Media saying, "Okay, what's going on here? What do academics say? What's the research? How can we understand what's going on?"
And through that, I found a few sort of useful metaphors for describing what's going on. One, what we see existing on social media in these apps is an analog to physical space. Some apps are like a dive bar. Twitter is like the hippest dive bar with everybody you can think of, it's loud, it's culturally relevant, there's tons of interesting conversations. There's also fights. There's all these things.
And so, if we think about it that way, we think, "Okay, you can take a bar and change the dynamics." You can turn down the music. You can turn up the lights. You can get a bunch of burner fire dancers there. All these things will change the vibe. You can put football games on the screens.
And Instagram is a super, super hip art gallery. Where everybody is, being seen and seeing each other. And there's lots of social kudos being given around and judgment around stuff. And Facebook, at this point, has become – it went from the frat house, to the ivy league, to now the community center, the senior center you rent for a family reunion.
But when we think about these things, all of these things then reflect physical spaces. And you can use metaphors of architecture and urban design for what makes the space safe.
[00:41:27] JS: Which is what Eli Pariser talked about in his TEDx Talk.
[00:41:31] EHP: Exactly. Eli and I found the same set of research and have talked to each other a lot over the years about this stuff.
[00:41:37] JS: We'll include a link to that in the show notes.
[00:41:39] EHP: What we need is the same things we need for healthy cities. A street is safe when it has street lights. A city is well connected when there's good public transportation. A city is healthier when you can walk from place to place as opposed to having to get into your car and go to these giant parking lots. Street fronts are safer when there are shopkeepers out front managing this stuff.
And so, what we need to design into the software is the affordances of little things that make the communication healthier and safer. And one of the things that makes it safer is an understanding of context and space and boundaries.
I interviewed a bunch of people who were Maori language speakers here in New Zealand who were using Twitter. And they said they have no problem with white supremacists on Maori language Twitter. Because if you're a white supremacist, you don't go through the process of learning the Maori language to go disarm people. That's a barrier.
And what fails with Twitter in some ways is because it's connecting and Facebook's mission of connecting the world is you get everybody together. And you get these conflicting collapsing contexts. And you don't have the ability to have ecological space and distance between them.
In Facebook's case, their recommendation algorithm drives you into these weird crazy spaces. If you believe in vaccines, you don't need to have long bouts of discussion about it. But if you don't believe in vaccines, you're going to be constantly posting in these Facebook groups about it.
And so, the Facebook algorithm says, "Well, that's engagement. We want engagement. So, let's bring people into these weird conspiracy theories that keep them excited." We need space by which people can understand where they are. And what the limits of the space are and what the rules of the space are. Subreddits do a particularly good job of this.
And we need to realize that what we're building isn't the central plaza of a town or the Central Square. And it isn't a shopping mall, which is owned and governed by the mall cops of Twitter, Facebook, TikTok. But it's actually a commons. Meaning it is a thing of economic value that's held together that we do social and business things to. We need to set rules for it. And none of us can own it because it gets created by all of us.
[00:44:08] JS: A lot of the issues with social media as we know it were just externalities associated with optimizing for something else, right? The inherent incentives at the advertising model, right? This sounds like it's calling for a set of design principles. This feels like some of the work that Tristan Harris is doing out of the Center for Humane Technology. Or a while ago, we talked to Tracy Maeres who's a Facebook veteran who started the lab at Yale that looks at self-reinforcing pro-social behaviors on these types of applications. We have the analogy of the sort of physical space. And we understand that. But like how do you actually get to broader development practices that yield those outcomes?
[00:44:53] EHP: The way we do that is we look at Elinor Ostrom. She's the first woman to win the Nobel Prize for economics. And she studied economic commons. And her work and the work that's come out of it, because she's passed away, says these eight principles of a well-governed commons. These are things that a commons needs in order to be able to govern itself.
The actual rules and structures of these commons vary a lot. But we need to look at these principles and say, "Are the things we're building reflected in things that – in these principles? And are we building technology that pushes people to work in that way?"
And so, I think that the DWeb principles are important. The principles of a well-governed commons are important. And then we need to evaluate and see where it is. The Metagov Project that comes out of Harvard is doing a ton of work on this stuff. They do work both in DAOss and in DWeb projects. I think their work is super interesting. Govlab is interesting. There's a lot of governance questions. And there's just also a lot of design questions, you know? Sharing content and retweeting is super powerful, and being able to repost stuff.
In Instagram, they make those re-sharings disappear after 24 hours unless you copy and paste and repost it. They distinguish between original content and sharing content. And that reduces the virality in some ways of that shared content. That's powerful.
Twitter doesn't do that. You can retweet and retweet and read it and you get things that go viral. WhatsApp has this network, they limit private WhatsApp groups to 256 people. And they're encrypted, so you can't see it. What, in Brazil, for example, the Right-wing does, is they create these like phone tree things of WhatsApp groups. So, they have a central one where the social media crew and other people are creating stuff. And then those go out to 256 people. And each of those people have groups of 256 people and those have – And so, they actually create intentionally deceptive memes and propaganda and send it out through this sort of dark network of WhatsApp groups.
WhatApp's limited how many times you can share. But they could do a lot more. I mean, by virtue of the fact that there's a Wikipedia page on WhatsApp lynchings. Clearly, there's a problem. So, we need ways in which you could say, "Oh, this is probably where this is coming from. Or this is how far you can share it. Or this is where the data came from. Or this was within this group. Or you're violating the rules of this community you're in. We're going to be able to see what you've done, tell you about it, give you options to change your behavior and then exclude you if you don't change your behavior."
And one of the reasons that Jack is so excited about Bluesky is because Twitter has an impossible problem. It's to set a set of rules for moderation and social behavior that applies in hundreds of languages in 200 countries in hundreds of thousands of different cultural contexts and communities and there is no one set of rules. Like, is a beheading video important news because it happens in Syria versus Mexico?
[00:48:19] JS: Yeah, it's very hard. It's a very, very hard problem.
[00:48:22] EHP: There's no right answer to that. One group of people, however well-funded, can't solve it. We need to push that out. We need to give people who are users of these systems more empowered tools to make those decisions, set those rules and enforce those rules.
[00:48:40] JS: Content moderation is such a complex one.
[00:48:43] EHP: Oh, it's a nightmare.
[00:48:44] JS: I mean, if you've got the data sitting on devices, how are you able to moderate the content? It's not sitting encrypted in devices, so you don't know what it is and you never see it.
[00:48:56] EHP: On Planetary, the content can be encrypted or non-encrypted. If it's encrypted, we, the platform creator, can't see it. But the devices directly talk to each other. So, there's plenty of non-encrypted data that we don't see as well. And what we do is we do a few things. One, architecturally, what you see on your device is based on the feeds that you pull and then who they follow. There's a network effect. So, if I don't see everybody's content everywhere, I have to pull in these feeds.
And then my relationship to the people I'm following creates something called a trust net. We have this trust net algorithm that says, "I like Jenny. So, therefore – and six other people that Jenny also follows. Therefore, this other person who I don't follow, maybe I will replicate their content and be able to see it in my network." But if a bunch of my people either are neutral or block someone else, then it alerts me and maybe I don't replicate that person's data.
And so, instead of having a flat blockchain where everybody has access to everybody's content, you have socially specific content. For example, it's like being in a forest. You know the forest is very big, but you only have a light that shows you your portion of the forest. And you can make the light brighter and see farther out. Or you can make it dimmer and you want a more intimate space. But if something is going on five miles away on the other side of the forest, you don't see it, and it doesn't affect you. But you could say, "This behavior over there is something that people in that community need to govern."
And then we have a backup, which is if you're in these communities, even if they're encrypted, you can choose to report to ourselves a third-party set of communities that provides judgment blocking services. And you can infiltrate communities, decrypt their content and report it out to the outside world. In many ways, it's almost dealing with hate speech using sort of a classic antifa model, which is you need people to go in and do the social work of infiltrating disposium. That's how the Southern Law Poverty Project works, infiltrating white supremacists and hate groups. It's a model that works.
If we relied on the government, be it Twitter, Facebook or the nation state, they're always going to choose to default to people who have more social power. If we want a more social justice agenda, we need a system of governing anti-social behavior and anti-social speech that doesn't rely on the player's overall social power in the larger economy. But instead relies on the ability of the people in these systems to report, have transparency, set governance and enforce decisions amongst themselves.
[00:52:08] JS: Right. And on Planetary, although your data sits on your phone – Well, you have a choice. You can put your data on a server or on your phone. But it's not just sitting on your phone. It's also sitting on some other people's phones encrypted. I'm asking this because I made the mistake of deleting Signal from my phone "Oh, I can just look at it on my desktop. And I just want to be on my phone less. And I lost everything. Because all of my data was only –" I mean, the data is on my desktop. But you can't recover that data once you've deleted it from your phone. That's the extreme variation of encryption and privacy that can lend itself to those sorts of things happening. You use your phone, you lose all your data. Particularly if you're monetizing your data, which some people might choose to do.
[00:52:58] EHP: Or data exists on your phone, but you host your friends' data and they host your data. And so, as long as you have your keys, your cryptographic keys, which is how Signal works, too. Then you can recover it. And in Planetary's case, for iOS and for Mac, we actually just rely on Apple's keychain. We take your keys and we put it in Apple's keychain and then Apple backs it up to iCloud. Or when your phone syncs to your laptop, Apple backs it up.
As long as – And even if you delete the app, because the keys are small, the keys stay around. We have users who install Planetary. They use it for a while. They're like, "Oh –"
[00:53:40] JS: I know your business model, as I understand it, the users will pay for it? Users will subscribe?
[00:53:45] EHP: Yep.
[00:53:47] JS: Because you did raise some money.
[00:53:49] EHP: Yeah, we raised some money from a number of angel investors and a few early stage VCs. And our business model is several fold. One, because you can have private groups, you can charge access to those private groups. Similar to Substack, or OnlyFans, or Patreon, there's the ability to charge us for these things. And we, as the application provider, as the platform, don't need to know what it is. Just like if we made a web browser, we don't care what websites you look at.
But we can provide the ability for people to charge those transactions and we take a cut of that. We provide some cloud hosting services. So, you can register ravel.planetary.name that is a personal website that provides a link tree type thing. And you can host that, or we can host that. And if we host it, we can charge a small amount of money for those additional features.
[00:54:45] JS: Got it.
[00:54:46] EHP: And because we're not running a lot of cloud infrastructure beyond that, it's relatively cash efficient. And inside our protocol, we do have something that's almost like a cryptocurrency thing. We call it a local coin, which is essentially the digital version of a punch card that you would have at a cafe. If you go to a coffee shop, they give you a punch card. Each time you go to the coffee shop, they give you a new punch. When you get 10 of them, you get a free coffee. They're essentially issuing their own currency where they keep track of the transactions with little stamps or punches in the punch card. You can take two of these punch cards, each of which have five punches, combine them together and then take it back to the cafe and say, "I'd like a cup of coffee."
And so, users can create these sort of thank you reward tokens back and forth and trade them between them. And each person's is their own little teeny cryptocurrency that they control an issue. But because of the way they're tracked, you can exchange between them and even exchange those in and out of other cryptocurrencies or fiat currency.
Basically, it's creating this almost thank you marketplace between people. And the conversion in and out of these sort of reward points that people can issue is something we can take a financial cut off and then no one's having a problem with it. Those are the main ways that we've thought about making money.
The other is if someone builds a large audience, they may want to run ads in their own feed. And so, right now, between Instagram creators and Instagram itself. Where the Instagram creators say, "I want to do this sponsored content." And Instagram doesn't want them because they want the ads to be going through Instagram.
And so, what we want to do is if someone has an audience like that, we provide a marketplace and a space by which we can connect the creators with the people who want to do sponsored content.
[00:56:50] JS: Is there anything about Planetary that we need to know that we haven't heard yet?
[00:56:54] EHP: I would say the last thing about Planetary that we haven't talked about is the algorithms itself. It's easy to talk when you talk about social media, about people's accounts being banned, or deleted, or kicked off platforms. But that's actually not the biggest problem that most users face. Most users face the fact that they don't have any transparency in the algorithms and the algorithms change on them all the time.
You make a post. You want your followers to be able to see that post. You want them to engage and see it. But all of a sudden, Instagram says, "We don't show it unless the first few people that see it like it." That's why women's selfies get valued more than others is because Instagram sees a photo, shows it to your close friends. They see a photo –
[00:57:37] JS: Yeah, they try to find that signal of how valuable it is to then propagate it.
[00:57:42] EHP: Yeah. You see a photo of your friend, you want to give them social Kudos and support so you like it more quickly. Therefore, when women post photos of things that aren't themselves, just artistic photos or other things going on, their friends don't feel the social obligation to like it as quickly. Therefore, Instagram shows that to less people. Therefore, everyone is trained that selfies are the thing that work.
[00:58:10] JS: This is called a reinforcing feedback loop. In addition to Elinor Ostrom, Donella Meadows is a really critical thinker of the last generation to inform design of this space.
[00:58:24] EHP: Exactly. What we do in Planetary is we actually just have a setting where we're like here are the algorithms you could choose from. Here is the source code for how they work. You can choose that. We don't go in and revert your changes after 30 days because we think we have a better idea of what the right things are.
And we have a space by which third-party developers can build new algorithms and talk about what the right algorithms are. And I think that algorithmic transparency, the ability to see and understand what the algorithms do, and choose which algorithms you want is critically important to this whole system.
[00:59:09] JS: Yeah, fascinating. I mean, also, interestingly. Because when I was doing this, it's like read the code. What fraction of your users are going to be able to read the code? I wonder if there's also something interesting here around taking a page out of liquid democracy. Or you might delegate some of that to people who have that understanding versus just how you explain what the algorithms do or how they're distinct from one another.
[00:59:32] EHP: There's a whole area of academic research on algorithmic transparency from transparency about criminal sense, descending algorithms to social media stuff. There's actually a lot of work on how do you show to non-technical people the algorithms work? Because some of the algorithms are just – they're just SQL code. You can look at it. And some of them are, like, "We took this data and then we ran this machine learning algorithm. And we made this training set. And then we apply it." You can't even use a programmer to walk through and understand it all. So, you have to look at the effects of it.
And so, part of that is documentation. Part of it is developing in reality the ability to have transparency in this and having people be able to document and say, "This algorithm works this way." And maybe different people have opinions about it. But we only have a certain number of experts say, "This is my analysis of this algorithm. This algorithm shows you much more stuff that's hot and being discussed. Or this algorithm shows you new topics of other people. Or this algorithm focuses on your type group of people who are most closely connected to." All those kind of things. And we need to just be able to say, "This is what it is. And these are people's opinions of it. And here's the analysis." Just like when you go to vote, you get an electoral analysis of each ballot initiative from a bunch of different perspectives.
[01:00:55] JS: Overarchingly, how's it going? It's 2019 that you raised some pre-seed funding. So, you've been at it for a couple of years now, a few years now. How's it going overall?
[01:01:05] EHP: It's going well. But it doesn't feel like none of these systems have taken off yet. It feels like we're still in the very much early adopter community where we're not close to crossing the chasm, as it were. And in part because they're still hard problems to be solved. Making all this stuff work efficiently on a mobile phone is hard. Being able to discover people, figure out the social graph stuff is hard.
And our protocol is secure Scuttlebutt. We have been having to do work that allowed us to do delete, and allowed us to do private groups and allowed us to do partial feed replication. And so, there are things in it that are obvious ways in which it's held people back.
The app is like – to compete against Instagram, Twitter, TikTok, all these others, you have to be very good right now. People aren't aware about how fast the software has to work? How good the recommendation algorithms have to work? How smooth it is? And so, with a small team, that's hard.
[01:02:21] JS: Yeah, amazing. This has been a little bit of a longer conversation than usual. But we had so much ground to cover to just give everyone a picture of this landscape from a broad lane. And to talk about your work, it was amazing hearing that story about that moment when the original sin, you called it, at Twitter, when the fail whale turned us in a different direction.
Thank you so much for joining us and for helping us understand this topic, Evan. It's been a real pleasure. I'm so grateful people like you are in the community and working on the forefront of so many really important problems.
[01:02:57] EHP: Thank you so much.
[01:02:59] JS: Thank you so much for listening. In addition to this podcast, you can find additional resources on our website, www.becomingdenizen.com, including transcripts and links to additional background materials for each conversation. For our most essential topics like universal basic income, decentralized social media and stakeholder capitalism, we also have outlines summarizing our research, which make it easy for listeners to very quickly get an overview of these really essential and important subjects.
On our website, you can also sign up for our newsletter where we bring our weekly podcast to your inbox alongside other relevant Denizen announcements. We're partnered with a lot of really incredible organizations on the forefront of the change that we talk about. And so, we're often bringing you announcements from them.
And finally, this podcast is made feasible by the Denizen community and listeners like you through our gift model. Denizen's content will always be free and accessible. We believe that it's just much more aligned with the future that we seek to embrace this gift model. It's through the reciprocity of listeners like you that we are able to continue producing this content. So, you can support us or learn more about our gift model on our website. Again, that's www.becomingdenizen.com.
Thanks again for listening. And we hope you'll join us again next time.
Our weekly newsletter is the best way to stay abreast of our inquiry and ongoing events.
Denizen operates a gift model to foster generous, reciprocating relationships.