Igalia's Brian Kardell and Eric Meyer chat with Robin Berjon about a piece he recently co-authored with Maria Farrell titled 'We Need To Rewild The Internet' about how the internet has become unhealthy and extractive, and how we can revitalize it using lessons learned by ecologists.
Brian Kardell: Okay. Hi. I am Brian Kardell. I am a developer advocate at Igalia.
Eric Meyer: And I'm Eric Meyer, also a developer advocate at Igalia.
Brian Kardell: And today we have a guest. Do you want to introduce yourself?
Robin Berjon: Sure. Hi to both of you. It's great to be here with you. My name's Robin Berjon. I work on standard set and governance at Protocol Labs, and I also sit on the board of directors of W3C.
Brian Kardell: Yeah. Great. And the reason that we asked Robin to come on today is we have been doing this series. I think we're up to 12 or 13 episodes already. That started in 2020. We began talking about web engine diversity and web ecosystem health. There's a whole series of chats that we've done and that same year in 2020 that we started that I also was supposed to give a talk at Web Directions that was originally called Web Browsers: An Inconvenient Truth, Ecosystem in Crisis. And I never wound up giving it. But since then I've converted some of those into articles. I noticed that Robin especially writes about some very similar things and he and Maria Farrell recently wrote an excellent piece called We Need to Rewild the Internet. Could you think that you could break it down for us, Robin? Explain the metaphor at least?
Robin Berjon: Sure. The first thing I would say is that rewilding the internet is not meant to be a metaphor. The core idea really is that when you're dealing with a highly complex environment such as the web where nowadays we're talking about five billion users, thousands of companies and entities interacting with one another, millions of websites. When you're dealing with a really complex environment, the rules that prevail in an ecosystem, in a natural ecosystem are very similar to the rules that apply to that kind of environment. And so when we look towards rewilding as an ecological practice, it really is to find tools to help us bring back essentially a healthier and livelier web. And so it's not a nostalgic approach. It really is a, Hey, given that this is a really complex environment, what kind of approach would make it sustainable, would make it healthy, would make it pleasant to be in. And so in order to do that, we learn from conservation biologists and this entire universe of rewilding where people have very specific approaches that differ from what we're seeing in general from either governments or corporations.
Eric Meyer: Rewilding in the ecological sense since you brought it up, how does one rewild ecologically?
Robin Berjon: The thinking around ecological rewilding is to depart from previous approaches to conservation biology that are characterized by what is generally called the pathology of command and control. What initially happened when people started to care about ecology is that a number of projects started by looking at what they thought were pristine environments and decided that basically everything had to be preserved to be as similar to what they thought of as pristine environments. The rewilding approach is much more about realizing that in an ecosystem, every species has basically a functional role, and that functional role is in part to control other participants in the ecosystem. And basically this mutual collective form of control creates a form of balance that is dynamic and that yields greater complexity. And so what rewilders have generally done is they look for mechanisms that in part let nature run its course because you can't control the complexity, but also where needed seek very specific functional interventions. So one of the famous cases, especially here in the US, was the reintroduction of wolves in several habitats, notably the Yellowstone area. And when wolves were reintroduced, they started preying on deer, which means that deer ate less of certain types of plants that had been seriously damaged. And because of that, all kinds of erosion processes were halted. And so just this single reintroduction of one species that served a function that had disappeared was successful in bringing back balance but also liveliness to the ecosystem. And there are plenty of other examples. Not all of them successful. But really the idea is to approach this class of complex problems, not with a very controlling mindset, but not either with a completely laissez-faire mindset. The idea is how do you find something that is dynamically balanced so that no species has too much power, no participant has too much power. And by bringing about this equality, you get something that is much more lively.
Eric Meyer: Yeah, that's interesting. Especially since it seems to fly in the face of a lot of how our industry operates in a way. In a lot of places there's a lot set on how do we measure the outcomes? What are the OKRs or whatever the latest term is. And this is more of a, Well, we have a general goal and we're going to try to get there through a chaotic method. Am I mischaracterizing that?
Robin Berjon: I don't think you're mischaracterizing it. I think maybe chaotic ... Chaotic is certainly one way to look at it because nothing is in control and therefore you don't want a completely ordered regime. But it's also not complete chaos in the sense that in a functioning healthy environment you also get a lot of structure. You get a lot of order at all kinds of scales. It's just not a top-down, single centered ordering and control system. And so one way to think of it is instead of having one set of OKRs, one set of KPIs for the entire environment, you let a diversity of different actors have their own OKRs, have their own KPIs that are distinct from one another and then enter into a form of tussle so that they make things work as best they can for themselves, but they don't impose excessively on others. And one great example of how having a single set of OKRs for the entirety of the web can go wrong is that there was a very good article in the Verge about how Google's SEO rules have basically shaped the entirety of the web or very near the entirety of the web. And there's a good reason for that. If you're not on Google today, given Google's dominance, essentially you are invisible. It's essentially the same and perhaps even worse than not being registered with DNS. These are basically the two primary information-seeking systems on the web. So because everyone's abiding by the same OKRs, the entire web now looks like it was designed to match the KPIs of Google Middle management, which is not what it was supposed to be. And so instead of having a completely chaotic system where nothing is findable, nothing is structured, we can imagine a richer set of structures where a great diversity of systems get to shape and order corners, environments overlapping parts of the web instead of having either just one or a complete mess.
Brian Kardell: So the thing I would like to actually just read from this article, it says, our online spaces are not ecosystems, though tech firms love that word. There are plantations. Highly concentrated and controlled environments closer akin to the industrial farming of cattle feedlot or battery chicken farms that madden the creatures trapped within. It's a good piece. I really enjoyed it.
Robin Berjon: Glad you like it.
Brian Kardell: I guess there's a big question here that's also like, so what and how did it get this way and what can we do about it? So what is it that makes it this way and keep it this way. You have some other pieces that you also have written that are a little bit more specific. I also like this quote from, you're going to need a bigger browser. It says we should move away from the assumptions that browsers are free and therefore cannot make money. It's a lie. There's money to fund a much better web and much more powerful user agents that support user agency. That money is currently locked up because defaults have been set and literally nothing else matters. Together the three of us actually led a session at W3C breakout session this year about how we fund the web. I think that's very related. But yeah, I don't know if you would like to present any of the things that we talked about or-
Robin Berjon: Happy to. I think that the money question is really central. The reason we're in this bad situation today is we have an interlocking set of constraints that reinforce one another and prevent money from flowing to places where it could help bring about change. And so basically people get an operating system on their mobile devices. It's one of two operating systems. That operating system is going to have a default browser and possibly a default search app. And both of those default browsers and any default search app will point back to the same default search engine. In turn, that default search engine makes money from being the default, which allows it to buy its position as the default search engine in all browsers in all operating systems. And so because defaults and because switching friction are so effective, this creates a system in which basically one company can own everything and keep financing everything and finances every single browser and pays for its continued dominance. And there's no way to break into that. You could make the best search engine ever and you would never have a single chance of entering that market. And if you make a browser, it is basically practically your only possible source of revenue. There's a few exceptions. There's edge, but it uses the same model from Microsoft and that's it. The amount of money that we're talking about is huge. Google pays over $35 billion a year to other companies just to be the default. That's not counting how much money they would value being the default in Chrome as well, or Android as well. So we're talking 10s of billions of dollars that are taken from the web because search is a concentration of the value of the web. It is where all the content comes, is distilled into the single place. And without that content, the search would be without value whatsoever. And so it takes money from the web and uses that to pay for market dominance. And that money for the most part does not go back into supporting the web. Most of the money that Google makes on the web goes into supporting its other activities in completely unrelated fields. The $20 billion that Apple gets from that system of default search engine and search engine royalties every year, those $20 billion are not reinvested by Apple into supporting the web. So all this money is being taken from the web, from value created by the web and invested in non-web things and invested basically in locking down market dominance.
Brian Kardell: Well and shareholder profit.
Robin Berjon: And shareholder profit. Well, yeah, of course. But I'm not even sure they give that much back in profit. A lot of it is being reinvested in dominating the next slice of industry, the next slice of the computing sphere. And so as people who would like a healthy web, our only chance of changing this is to change the system. There's no way that we can out compete people who can put 10s of billions of dollars into dominating a space. The only way we can shift that is by shifting this funding model. And essentially the strength of that model is also what makes intervention possible. The strength of that model is that it's entirely consolidated around a very narrow intervention point, which is search royalties. And so if we can take control over search royalties and basically assert the right of the web to that money via a number of policy interventions, then we can gain access to that money and use it to fund actual web infrastructure, which would be a huge change. I'll get off this soapbox for a minute. And I'd love to hear what you two have to say about that as well.
Eric Meyer: So you say policy intervention, you really mean governmental policy intervention? Yes?
Robin Berjon: I mean it's the most likely path I think for two reasons. One, I doubt that this will be done voluntarily in industry and that we can find the pressure points to make it happen otherwise. The other reason for which I think it's likely is because it's actually a relatively simple privacy intervention to make that possible. You basically need to tie the ability to set a default search engine or to set search royalties with participation into this taxation system. And I think there are ways of making that attractive to a sufficiently impactful regulatory jurisdiction that we can actually make it happen.
Brian Kardell: Yeah. This isn't a concrete, this must be the way. If you imagine where when you set up your operating system, it let you choose the defaults, Microsoft was forced to in courts with regard to the browser way back. One of the things that you could correct in that model would be that you could enforce that in order to be eligible to be chosen as default, you must participate in this, you must pay back. So if you're chosen however you make your money, it has to be paid back in some kind of ... Let's call it a levy or tax or whatever that would help fund things. It would allow things to get traction in the first place. So as I said, there was a bit of a historical accident where Mozilla was the first real open source thing in the modern open source sense trying to do a browser and they got some early funding, but how was that going to work exactly wasn't clear. They still, because of the inertia of being previously Netscape and this interest in this new idea of open source, they still had 3% of the market already before they ever launched. Just their beta had like 3% of the market. And that 3% was worth eyeballs to Google who was also trying to get search. They were trying to launch search. So when they struck that deal, that was a moment in history that's really hard to recreate. Brave, it's not maybe in the top echelon, but it's pretty well known. It has 10s of millions of users monthly and it doesn't even show up on the graphs how big the web is. So if you wanted to break in, you need a way to fight defaults problem. And that would be one way to battle the defaults problem where you could amass enough market share to matter enough to get a deal and then that deal would have to play back into this levy system.
Robin Berjon: I think we're all struggling because all of us grew up in a world with a very simplified idea of how things like this get organized. You have companies, they exist in country, that country is governed by a state. All the states have a very nested, hierarchical structured that is pretty simple and legible. And then internationally, while those companies might have some difficulties because they need to comply with multiple laws, but there's no real transnational thing except for the UN but that's vague and they deal with things like human rights that seem pretty far away from JavaScript to most people. But what we actually need to think about is the fact that the web, the digital sphere in general has created this entire new space. That space is completely transnational. It really doesn't align with territorial boundaries and it is becoming the infrastructure not just for a few games and unimportant things, but it's the infrastructure of all infrastructure. It's the infrastructure of everything. And therefore we need to ramp up our thinking to match the complexity of the task. You mentioned the difficulty in getting to 1% on any kind of diagram. Yeah, 1% of five billion users is still 50 million and getting to 50 million users is a pretty big deal. And so a lot of the tools we have in terms of thinking about standards, in terms of thinking about open source are really not scaled to the size of the problem that we're dealing. And that's why I'm using too many words to describe this. You're using too many words to describe this because we're still basically putting together the pieces of how the hell do we think about this massive thing that we need to govern if we want to keep it. But we're still struggling to figure out is it a tax if you do a thing without a government, but it's prodded by a government and helped, but it's actually transnational so that everyone benefits. These are slightly mind-bending ideas, but they're also I think what the situation calls for. But again, this is why we have to maybe look for ways of explaining them that are better than what certainly I've come up with so far so that we can keep imagining this future and thinking of a world that actually works.
Brian Kardell: This talk that I almost gave a thing that I missed from it is explaining ... I think maybe I did this in a blog post explaining world that the web was born into. It very nearly was born with web browser being like a shrink-wrapped piece of software that you would just go and buy at the store. Because there were actual stores back then that you would go and buy software at. It's interesting to think about all the ways that companies exist and things exist to produce software and what is the funding model and all that. It is currently ads on search and as you said, a lot of the value is extracted and only a tiny portion of that goes back into actual funding. So I think it's informed, educated, back of the napkin estimates, but I think $2 billion a year would cover the entire ecosystem, not even just one browser. When you look at it that way and you say, well, but there's five billion users or six billion users, that's not a lot of money that we really need. So yeah. I don't know. I would like to see us explore lots of ways. I think about this with television all the time. That also is going through this terrible phase right now where if you watch sports, you have to pay to get all of the different streaming packages because they're not just on a single channel anymore and it costs you $114 a month to watch football or something like that. And that still has a million commercials. I don't know. It is a similar thing where there's a lot of value being extracted out of that in multiple ways. It would be nice to think about the ways that we want, what is the future that we want of the internet? I think ads have a huge role to play. It's a really good way to fund a lot of things, but I think that needs to also be somehow rewilded. Everything is pumped through two specific channels and it's the only way to monetize right now, which is not great.
Robin Berjon: I certainly can't disagree with that. It's true that we could imagine, you mentioned television for the longest time in many countries, television was publicly funded. If you think of a model like the BBC, there's basically a small ... I forget what they call it. I believe they have a very specific British name for it. But there's a specific tax that goes towards the BBC for everyone who owns a TV or perhaps even a radio set as well because there's also public radio channels. And sure enough, with five billion users, you really don't need that big of a tax to finance $2 billion, which is the high end of the fork I think, to support all browsers. That being said, I don't think that browsers are the entirety of what we need to support, and that is also why I think we need to be a bit more ambitious. If we think in terms of what makes up web infrastructure, there's a number of other things that come into play. Search is one. We need to be thinking actually of a model that finances search because as we've seen, whenever search becomes under the control of a single player, it gets into a terrible state. People have been joking for months about how bad Google search had become, and now we have the AI version that's even worse and people keep posting these hilarious things. Funny as they are, this is critical infrastructure that is being stripped mined into uselessness for pure profit. And so on proposal I had made before for the advertising ecosystem that you mentioned is that instead of having the rules for advertising be set by the people who run advertising infrastructure, it will be good to have the rules for advertising set by the people who use advertising. And that is publishers, advertisers and people. These are the three constituencies that, they all benefit in different ways for advertising. And yet it's hard to think of people benefiting from advertising. But were it done well, it could be done without violations of privacy and to grant access to quality content to the greatest number. These are valuable things. So I think yes, looking for $2 billion a year for browsers is definitely part of the problem, but we should also be thinking about ways of governing every single one of these pieces of infrastructure so that they work for us. And then you can have an absolutely wild competitive capitalist market landscape sitting on top of it so long as the infrastructure itself is not controlled by individual players. And so long as it works in support of public interest.
Eric Meyer: I will say that I had a little bit of maybe a tremor when I read in your article. You're saying about rewilding. It targets entire ecosystems to make space for complex food webs in the emergence of unexpected inter-species relations. It's less interested in saving specific endangered species. Individual species are just ecosystem components and focusing on components loses sight of the whole. When you apply that sort of thing to social and human systems. That sounds like a lot of dystopias I've read over the years. So how would we avoid the dystopic outcomes here?
Robin Berjon: That's a good point. I think it's important to understand that this is not something that you apply to individuals, for instance. So what this sentence is talking about is there have been rewilding projects where the species that disappeared and caused ecosystem upheaval either was completely lost, it went completely extinct, or did not reproduce well in captivity such that reintroduction was very perilous if possible at all. And so one example is ... I forget exactly where. I think it was in Madagascar where a giant turtle was performing an important ecosystem function and it became very highly endangered and it was extremely hard to reintroduce it. And instead people worked on reintroducing a different giant turtle. And apparently this has been having a positive impact. Because even though it's a different species, it is having the same effect on the environment that is the tinkering that people have historically been afraid of because that's the bioengineering that can really backfire. But if you really think in terms of functions it can work. Similarly, in Europe, there's a project, I forget exactly what it's called. But that is basically trying to breed horses back into their wild equivalent aurochs that disappeared. And that has not led to reintroductions, not horses actually. Sorry. They're cows. It's called Project Tauros. And the idea is really bringing back the wilderness to something that is functionally equivalent, even though we will never get the same species back. So I think when you read that, you're thinking more about, hey, people are interdependent or groups of people are interchangeable and let's get rid of them and focus on the whole. That's really not what we're saying here. What we're saying here is much more, hey, perhaps specific companies or specific software projects or specific protocols might be interchangeable, might be replaceable. It's okay if we lose some so long as we maintain that overall structure. The important thing here is to not focus on nostalgia. So the idea is that hey, chat systems are too scattered today. The idea is not hey, let's bring back IRC, but much more how can we make them interoperate or at least how can we integrate them better if interoperation isn't possible? I hope that clarifies it.
Eric Meyer: Yeah. It clarifies that what you're talking about the individual technologies or the individual protocols or the individual whatever. The individual technological components, not the actual people who are using the web.
Robin Berjon: Yeah. People matter.
Eric Meyer: Which is good because unfortunately, and also by contrast, the system that we have now does not always value people.
Robin Berjon: I think the system we have now rarely values people, I would say.
Brian Kardell: Yeah. So today we have Google, Apple, and Mozilla. Definitely champions of the web, but single org stewards, important thing to maintain any one of those or for it to be a single steward. I think it could be, for example, really good to improve that because there are multiple stakeholders who spend a lot of money to make products around things and participate in standards. Ultimately they're all begging for those three to please pay attention and help with their thing. And maybe that thing is like SVG or MathML that gets just absolutely no love from those orgs. But there's an awful lot of people who care about that. In the recent state of HTML survey, SVG was very, very high on stuff that people cared about, and yet it doesn't really get a lot of attention. So I think what you're saying is that what's important here is to maintain the qualities of this that we think are good, that we think are important, how they interplay that there is some diversity, but there's other things that maybe we would want to change about how investment came into the platform and how we manage where it goes and how it's responsible to the different constituencies. Is that way off or that's kind of right?
Robin Berjon: No, I think that's kind of right. You mentioned SVG. I remember for instance one pretty intense debate that happened probably ... I don't know. 12, 15 years ago that was about how SVG would work in HTML proper with the HTML parser. SVG was initially developed purely in an XML context and it had a pretty well-defined, or at least decently well-defined integration into XHTML, but it had no defined way of being integrated into HTML. And at some point people started saying, you know what? These are just pretty much tags and if you don't do the XML self-closing thing, but you close them as separate tags, separate closing tags, then it will just work in an HTML context and maybe the parser could simply map them to the proper SVG classes. And voilà you have SVG in HTML. A lot of people started screaming against that because they were like, 'That's not how SVG was defined. It was supposed to be XML. We really care about that very specific syntax. That's how everything works. You're going to lose namespaces, you're going to lose a whole bunch of other things.' But really at the end of the day, what is it that you want? Do you want specifically one syntax for vector graphics or do you want an awesome vector graphics system that actually happens to work in the browser? And I think any one reasonable would answer the latter. And so yeah. Maybe it causes friction when you have to switch syntax and switch technologies, but really focusing on the outcome and on the functional role of that specific technology is what works there. And of course, that's a small example. Most people will never be impacted in any useful way by some pretty minor differences in syntax for vector graphics. But I think this is the kind of thinking that we can apply across the board.
Eric Meyer: So I'm curious how much hope ... Or maybe hope's not the right word. Do you feel when you look at, for example, the Fediverse Mastodon and that Threads and all that work that's happening around ... I think we're still calling it microblogging.
Brian Kardell: Can you submit your reply in the form of a Bluesky skeet?
Robin Berjon: Well first, one of the important components of rewilding is hope itself. Working in ecology is really hard. People talk a lot about the damage of the ongoing climate disaster, and that is absolutely true. But it's only one of the ecological disasters that are currently ongoing. There's also a massive collapse in biodiversity that is pretty much just as threatening as the climate emergency. And so when you work in rewilding, one of the core tenets is you have to approach the problems with hope because approaching them with doom and gloom is really not going to get you anywhere and it's not going to help you make change in the world. And I really think that is something that we need to bring back to the tech space. There's been a lot of very justified work on criticizing tech. But it has created this mentality of powerlessness. We are here and we really can't do much to change things. Why bother imagining a world in which things could be better? I think that's not the case. And I do have hope. And I think yes, the Fediverse is one area in which that gives me hope because people are using it, but also because people are reinventing new ways of doing things. I think other approaches to social networking that are being invented these days that are not necessarily based on activity pub and other Fediverse technologies are also very interesting. I think Apporto and Bluesky have some very interesting architectural approaches. I wish people were less interested in picking sides and more interested in looking at architectural choices. I've been paying pretty close attention to Nostr recently. I haven't had a wonderful experience using the social network, but the technical decisions made in terms of the architecture and not necessarily the details. The details are very much a mixed bag, but the architecture is itself very simple. I was able to implement a useful Nostr server in under a day, and it is leading to all kinds of very interesting experiments with people, not just using it for microblogging, but using it for longer blogging, for file sharing. People are creating a complete marketplace protocol on top of the Nostr protocol. And basically this whole world of experimentation I think is very promising. It's not enough experimentation. On its own is not going to displace the overbearing power of incumbents. But we need both. We need an approach that targets that excessive authority, and we need an approach that creates new imagination, new dreams, new thoughts about what the future could look like. And bringing those two together I think can be really powerful.
Brian Kardell: So just on the off chance that some of our listeners might not know what Nostr is, give us a brief, what is that?
Robin Berjon: Sure. So Nostr is a decentralized social media protocol. It works in a very simple way. You post messages by signing them. Your identity is a private key and you sign messages and you post them to relays. And anyone can run a relay. You normally listen to multiple relays and it's decentralized in that it's very hard to censor because you'd have to censor all the relays to prevent one person from posting. And it's also easier to run than most peer-to-peer systems because peer-to-peer protocols tend to be difficult and painful to work with. And so this basically server-based, but using the fact that that service is cheap and anyone can spin up a new one is really what's powering it. So yeah. You post sign messages to a bunch of relays people, you listen to a bunch of different relays and you follow the people's public keys and you get updates. That's it.
Brian Kardell: Interesting.
Eric Meyer: I do feel a little bit like as developers, maybe we can get a little bit more of a handle on all of this. One of the things you talked about in the article was shifting baselines and how if a baseline shifts over a long enough period of time, it's very difficult for humans to notice. This is one of the problems we're having with the climate crisis, for example. Things like insect diversity and those ecosystem problems. I do feel like as developers, we're a little bit used to baseline shifting on scales that we can see even if we don't immediately react to them. And I use react there as a pun because JavaScript frameworks are an example in our field, in the web field of how a normal baseline changes and how shifting from an older situation or an older context to a newer context can be difficult. It can be very hard to keep up and it can be very difficult to switch from jQuery to MooTools or MooTools to Angular or whatever. It does seem like we've also had that on a longer time scale that people don't notice as much with the web itself. It started out wild and it became managed. Plantationed, to use that term. And a lot of people now, probably a lot of developers now might think, well, of course this is how the web is. Those of us who've been around a really long time and are old can remember a different time. But I wonder if maybe it's possible to make the argument to the people who developed this stuff and present to them, Hey, the way that JavaScript framework baselines have shifted, that has actually happened on this larger scale. And it's possible to shift it again. It's possible to have a situation different than the one we have now. Am I on anything like the right track here or have overlooked critical differences between those two kinds of scales?
Robin Berjon: I completely agree with you. I think, well, apart from the fact that you're really dating us when you talk about switching from jQuery to Mootools. I like everything else that you said. I think there's a form of defeatism in a people who use ... A lot of developers are just like, 'Oh, well, that's the tool that's popular this week and next week it'll be another one and I'll just have to learn that and that'll be it.' Without getting involved in the decisions that shape these tools and these decisions matter. And I think there's several things that people that as developers we can pay more attention to that can really help shape our collective systems in ways that will have also good effects for people who are not developers. One of them is to pay attention to the governance of things. A lot of the time we tend to not want to deal with governance or not think too much about it, but really the defaults that we set in the systems we create, they matter a lot. There's this really good book from Nathan Schneider called Governable Spaces, in which he points out the fact that every single collective system you install starts off with one super admin with total power over everyone else, and then you can add some structure to delegate a bit of power to other people. But there's really this futile system that's completely built into it. Well, we can think of other ways. We can think of systems that are just open by default and then you have to install a political regime for it. They've done experiments with that in some games where it starts, where everyone's an admin, which of course has security implications. You do want to install some kind of political regime pretty quickly. But you don't have to have a super admin. You can have some kind of voting system, liquid democracy, et cetera, et cetera. And so I think it's important to think of those defaults. Another thing that people that developers I find are often insufficiently curious about is to understand what's going on under the hood. It's always instructive at least once when you are using a library to pop it open and go like, okay, when I call this, which is something that ... It's a method that I call every day. What does it actually do? And there's several things that come out of that. A lot of time you realize that some things that happen under the hood are way more complicated than you anticipated, which will help you understand other parts of the system or why there are constraints on what it can do. But also you might realize that it might be doing things that you would rather it didn't or that don't feel necessary and that you could build an alternative. Think of alternatives. And so really digging a little bit into what's happening at the lower layers can help. And a final thing that I think is worth reflecting upon, especially for people who are curious about open source and open standards and all that, is to really take the time to read up on good protocol design. There's a number of RSCs about that. There's people who blog about it, like Mark Nottingham regularly talks about things like that. Martin Thompson has written some good things. And the reason I bring that up is because I'm noticing in this effervescence of people creating new Fediverse systems, new social protocols, et cetera, that a lot of people have their heart in the right place. They are outstanding engineers in terms of implementation, but they really don't have the slightest idea of what it takes to create an interoperable protocol that actually works at scale. I see the same mistakes being repeated over and over again and things that the ITF or the W3C figured out 20, 30 years ago are being repeated. And again, if we want to make the world better, if we want to make this tech stack work better, let's learn from the people who shot themselves in both feet and the knee 15 years ago. It will really shorten the painful period.
Eric Meyer: So paying attention to those things is good. Paying attention to governance, paying attention to protocol design. Let's say someone pays attention to governance and they want to do more. They want to help however they can to push things in the direction of better governance to rewild the internet. What are some things they can do?
Robin Berjon: Inventing the future. I would say it really depends on what your focus is. There's no point in trying to become good at something that you're not personally interested in. You will hit a wall. I think of the entire universe of things that you feel need to be rewilded ... And there's a choice. Search, social, co-chairing, chat. There's a whole universe. Advertising if you want. Browsers. There's a whole universe of things that need to be rewilded. Pick the area that bothers you the most or that you're the most interested in, and really try to understand what alternative is possible. And the alternative is never going to be something superficial like, oh, wouldn't it be nice if X is? It's how do you change the dynamics of that system such that it works in and produces the outcomes that you think are beneficial? And a lot of the time this is going to be, it's not something that you can do in half an hour while you're waiting for the bus. It's going to become a quest because once you start pulling on one thread, you're going to realize that you can't just change that system. You're going to need something else. But really following that interest in making one specific area better to what proposals, what structure would make the great step for developers who are the people who will be able to dig into things and get their hands dirty and look for the next thing.
Brian Kardell: So you talked a lot about the protocols and everything, and I just wanted to mention that you're working for Protocol Labs and on distributed web ideas. Introducing that, I see it as a little bit similar to the SVG thing that you were talking about earlier. How do we marry that into the web that we have today so that you get it to a broad audience and everything? Do you see any correlation there, or not really?
Robin Berjon: No. I absolutely see a correlation there. I've been working on that quite a fair bit actually, especially in recent months. It certainly interesting to think of all these new ideas in the DWEB space, in the Web3 space, et cetera, as having potential better pathways to blend into the web. I think a lot of the time these new technologies were invented in ways that don't necessarily mesh well or integrate well into the web or work well with JavaScript or work in browsers or with HTTP. And in order to bring the best of their value out in order to really deliver them to the greatest number of people, but also in ways that have the greatest impact, it's really valuable to look for ways to integrate them better into existing web stacks. And so one of the things I've been working on quite a fair bit recently has been to poke at IPFS, the interplanetary file system and basically break components out in a way that I think could help them work better with the web. So for instance, IPFS is content addressed, which means that you find content based on the hash of that content, which has all sorts of nice properties in terms of verifiability, knowing that the content you have is authentic, caching, getting content from untrustworthy sources, but still knowing that you're getting the right thing. These are all nice properties. But so IPFS relies on content IDs that encode these hashes. And content IDs are very flexible. They have many, many options, but that also makes them pretty unwieldy to manipulate and deploy. And so I'm working on a subset of CIDs called LUCIDs, lightweight, universal CIDs that is very easy to manipulate with JavaScript, very easy to integrate into existing web systems and has practically no options. And I think by doing that repeatedly, we can find increasingly better ways of integrating these new technologies into the web and bring the benefits there. And so I really think that's the most fruitful approach that we can take.
Brian Kardell: I like that you took a page from Tim and put Universal right in the name.
Robin Berjon: Tim went back and forth over that one, but he always regretted yielding on the universal at some point. So let's keep it.
Brian Kardell: Yeah. Yeah. I think that's really cool and I'm looking forward to seeing how that comes about. I think we've had some discussions on this before as well. We have had other protocols, like the tell protocol that you can put into your page and it makes things a link. But actually turns out in a lot of cases that the browser just smart, educated, guessing what things are, phone numbers, and then making them dialable is a way more practical thing in a lot of cases. It doesn't become a link unless it can become a link. It doesn't become a link on your desktop that won't go anywhere. Might launch a Skype or something, which you don't even have an account for anymore. So yeah, it's interesting to think about how you integrate these things and what approaches we do and what things in the past we learned from. I think there are a lot of really neat ideas there that I would like to see us have a bunch more experiments with and how we can bring those things to more users because that's how we can find out if there's some there there that can grab on and get somewhere. I think a new protocol is a little bit like a new browser where it's like the power of defaults. All these things that are the default because they're so widely used, you're trying to break into that space. It's hard to get the inertia that you need to break into the real mainstream of things. So I don't know. I really like the space and the things that I see happening there, so I hope that it continues and I would like to find more ways to play more of that stuff myself. Do you have some ideas about ways that some people who are listening to this show that don't know a lot about IPFS ... How can they get their hands dirty and experience it a little bit and a low friction way to get introduced to it and have something they can play around with?
Robin Berjon: Again, it all depends on what use cases you are interested in personally. But one easy way to jump into IPFS is to just head to ipfs.tech and follow one of the tutorials. That will get you on your feet pretty quickly in terms of experimenting with ways of integrating IPFS into the website, there's been work in Brave, there's a build of Chromium that includes IPFS support that you can get. So those are also interesting ways of tinkering with IPFS. And also, so as I mentioned, I've been working on this thing called LUCID. It's really an experimental playground, so caveat emptor, don't go digging into that unless you are willing to play with things that are probably broken. But really what I'm looking for is increased hack ability. I think a lot of these DWEB systems lack a degree of hack ability. And when you're trying to build something new, it's important to make it more hackable so that people can try really stupid things really quickly. And some of those stupid things might turn out to not be so stupid. And so with that in mind, one of the things reasons that I've been looking at Nostr, as I mentioned earlier, is not because I'm enjoying the experience of the social media itself, it's a bit too Bitcoin-y for my tastes. But because it's an extremely hackable protocol, and therefore I'm trying to figure out interesting ways of injecting pieces of web into social media. And so one thing I'm working on is, hey, if you have a small bundle of web content and you manage to serve it in a way that has very strong security so that it can't phone back home so that it can't report on its usage so that it can't access arbitrary APIs on your behalf, really something that's really contained, you could create little islands of interactivity that use almost the full power of the web, or at least the full power of the client-side web. And so by serving those in a content-addressed way, you can serve them in a more private way. You can serve them in a more reliable way. And yes, these are things that I call tiles that I've also been developing as part of this LUCID project. And I'm currently trying to build a version of Nostr that supports tiles so that you can basically send people tiles that are full-fledged, HTML, the CSS, you name it, interactive systems, but that can be embedded safely in anything and don't have privacy downsides or just don't have the traditional privacy downsides of embeds and are very secure. So this might seem very abstract if we were in front of a browser and a code editor, I would show you. I'm of course happy to discuss this further with people on the internet.
Brian Kardell: Yeah. Well, speaking of that, where can people find you on the internet and they want to chat with you or maybe yell at you or maybe send you money?
Robin Berjon: All of the above as a welcome. So my blog is @berjon.com. My email is robin@burgeon.com. My Bluesky account is robin.burjon.com and Fedi is robin@mastodon.social. You'll notice a theme there. Oh, and my GitHub is darobin. So all of these are welcome. Yeah. Please don't hesitate to reach out. Oh, also robin.77 on Signal if that's your thing.
Brian Kardell: Can you maybe spell Berjon in case people-
Robin Berjon: Yeah. B-E-R-J-O-N. Hon, maybe.
Brian Kardell: Awesome. Awesome. Is there anything else that you wanted to say, Eric?
Eric Meyer: I wanted to say thank you very much, Robin. That was a great conversation. Really enjoyed it.
Brian Kardell: Yeah, me too. Me too.
Robin Berjon: Thanks to you too for having me. It's been a great pleasure.