Revisiting Web Intents and the Ongoing Challenges of Decentralized Web Services

Matthias Pfefferle:
Welcome to a new episode of the Open Web Conversation podcast powered by Open Channels fm. My today’s guest is, at least from my perspective, the inventor of Web Intents that almost solved the. The biggest issue in the decentralized for decentralized services. And that was almost 15 years ago. And we want to talk about the idea behind Web Intense, the almost and how we can maybe learn from his experience. Welcome to the podcast. Paul Kinlan.

Paul Kinlan:
Yeah, thank you very much. Very excited to be here. I know we’ve, we’ve chatted a bit in the past. Yeah, quite.

Matthias Pfefferle:
We go quite way back, almost 15 years ago.

Paul Kinlan:
Yeah, yeah, we, we say we’ve never actually seen each other in person. Well, we’re not in person technically, but it’s. Yeah, it’s great to be here. You know, Web Intents is very kind of still close to my heart and I’ve still got some thoughts around some of the, maybe the technologies that are starting to come through again in the browser. Maybe so.

Matthias Pfefferle:
But I do not want to limit you or to how you say that. Yeah. To limit your experience only to Web Intents. So maybe you want to say some more words about who you are and what you are currently doing.

Paul Kinlan:
Okay, yeah, yeah, sure. So my name is Paul. I, I live in North Wales. It’s a very cold and wet day today, which is like most days in North Wales. I’ve been, I’ve been on the Chrome developer relations team since pretty much the start, like its inception. And so I’ve been kind of like building around, I want to say, kind of like the. Well, I need to be careful. It’s not the forefront of the web. Chrome has its own issues where it pushes on the web maybe a little bit too much sometimes. But we, we, we’ve been in this space of just like, how do we think about what is the future of the web? Because Google’s obviously very invested in kind of the success of the web. It has its own opinion as well about kind of maybe how it should evolve or it can evolve. And then obviously there’s tension in the ecosystem which naturally pulls us back a little bit and all those type of things, which is good, by the way. It’s not like, oh my God, this is terrible. So where am I going anyway? Yeah, so I’ve been around the Chrome developer relations and Chrome for I think it’s 16 years actually this year. And prior to that I’ve, you know, I’ve been a Web developer since 96, maybe 97, you know, and I focused. The thing about me is I’ve. I like APIs, I like exploring. I used to spend so much time on MSDN, you know, like, you go into Visual Studio, press F1, get the help menu out and like, you know, get your CDs and everything in. And I’ve always enjoyed kind of exploring the forefront. And so like the developer relations team for me has always been kind of a nice, happy home where it’s like I get to explore the forefront. I get to think about some of the things that I want to see on the web and very, very lucky position to be in that space. Yeah. And then Web Intents was actually one of those first projects that, you know, I had to, you know, I wanted to try and solve what I thought was a really big issue. But I thought we had a technology. Yeah, yeah, yeah, yeah. It’s huge issue. Right. And it didn’t work in the end. And it was also my first big project failure as well. And so I’m also happy to talk about how I felt around that as well. Like, it’s like it was a big letdown actually. But yeah, yeah, it’s, it’s.

Matthias Pfefferle:
Maybe you can explain to us what Web Intense or what the idea behind Web Intense is so that we can dig a bit deeper into.

Paul Kinlan:
Yeah, yeah. I’ve always been, I’ve always been trying to think about how do we maybe rephrase it? The hyperlink is an amazing invention. Right. And the way that we express it inside a web page I think is incredibly powerful. And I think things like search and, you know, other tech, not like, you know, you know, the web is as it is. It’s been great at linking technology. Like linking pages together. Sorry, not technology like linking pages together. And, you know, I’d also started using Android maybe, I want to say like 2008. I don’t know the exact time, but I was playing around with and building some Android apps on the side. And I really like this Android Intent system where it’s like, you know, you click on a link and you can either say that you own the URL and you want to, or, you know, you can handle the URL and do something for it, or, you know, there’s some functionality that I need. Can you go out and help me find that functionality? And then on the, like, the, like the demand side and then on the supply side, like the app saying like I can do all these things. And then there was this like magical kind of like resolution in between of just like matching intent to like the like the supply. I was like, that’s incredible. We just don’t have that on the web. It’s like. And so I started exploring, you know, different ways that we might do it. And actually there was a. I had a website, I think it’s still around, but it’s like web intense.com because I just bought the name straight away. I was like, oh, I want Android Intense. But for the web, Web Intense is the thing. And so I built this. It was using, I think how to do it now. Shared workers. Because shared workers have come out and like this gets back to me like I just like to explore. Shared workers are relatively new. One instance across many tabs and you can communicate. And it was like all of a sudden I could build a. And this is when we didn’t have the first party, third party segregation of kind of data storage. And so for any website I could open up like a small tab or embed an iframe and then they could talk to anything, anyone else that embeds an iframe. And so it’s post message and all those types of things. And I was like, yeah, yeah, this is, this is, this is like Android Intense for the. It was nothing like Android intent for the web, but it was one of those bits where it’s like it caught some attention internally. And then we were like, actually maybe we should try and formalize this because it seems like a missing problem or like there’s a big problem and there’s no solution right now. And yeah, that’s how it all picked off. It was just like there’s a lot of interest in all of these companies like you know, building browsers and everyone. This is a really big issue, you know, federation, I suppose to want of a better word. It’s just like it gives you the user control. It gives like the ecosystem control. And you know, there’s no middle, like. Well, it can be a middle person I suppose in between to kind of help resolve and root. But you know, ultimately the, the user can actually be in control of the services that they want to use instead of being forced to use certain services by the website. Anyway, there’s a lot of interest to do it and it always comes up every couple of years.

Matthias Pfefferle:
Right.

Paul Kinlan:
As well. And so yeah, sorry, I actually wandered very far off. I think maybe the initial question that you had.

Matthias Pfefferle:
Yeah, but that’s, that’s totally fine. Because that is. It’s funny because that was really. I think it was around 2010 when you started at least the formalizing process.

Paul Kinlan:
Yeah, I think it was like 2011 maybe.

Matthias Pfefferle:
Yeah, yeah. Somewhere around that. And it’s funny because it’s still a problem. So you tried to solve that almost 15 years ago and up to now it’s still a problem. And it feels that nobody, maybe cad is the wrong word, but nobody tried to solve that in between this time at least on such a fundamental level that you started with web intents. Because currently I would say really one of the biggest problems in all of the decentralized Fediverse stuff is how can you connect different servers together. So it’s simply I visit a site, click the like button and then a like was created on my site, not on the side of the person I visited. And that is with all the other intents too. Sharing, following. When I click the follow button, how does the site I visit know where I want to follow this person? And yeah, you can go on with that on and on. And yeah, it’s. It’s simply. There’s a huge problem when you want to kind of build or have the same experience with a closed social network. Yeah. That we have to solve and. Yeah. What’s the.

Paul Kinlan:
So I have a question for you then basically. So you know I’ve. I’ve actually been tracking some of these kind of systems. I have a search for web intents so it always comes up and you know, I think I wrote a blog post about maybe how we could, you know, abuse register protocol handler to do like the follow intent.

Matthias Pfefferle:
Right.

Paul Kinlan:
For master. Well, it was for the Mastodon client maybe, I suppose. And so like you say, it’s not been done. I’ve got some proposals. It’s not been done. Why do you think it’s not been done in the kind of the federated know Fediverse social system at the moment. Is it. No, I don’t want to say. Politics.

Matthias Pfefferle:
No, I wouldn’t say it’s politics. It’s more that it’s. It’s a hand act problem. So I think there’s a need. But to solve that properly it is. We. We need the. The middleman you called it. We need something and I would say the browser is not the worst choice. We need something that delegates between. There’s a like button and there’s a service that should handle the like request. How can we get to know these two services and because there is no real. It’s hard for me as a web developer to force a browser vendor to implement something like that. So I would say it’s kind of that problem. And on the other side, I think. Because you mentioned the custom protocol handlers.

Paul Kinlan:
Yep.

Matthias Pfefferle:
The. Because that is supported by at least some browser vendors.

Paul Kinlan:
Yeah. So mobile is the one that’s the holdout, right?

Matthias Pfefferle:
Yeah, but it’s. So there was. I think it was implemented into Mastodon some five, six years ago, and they complained. The user experience.

Paul Kinlan:
Yeah, yeah, yeah.

Matthias Pfefferle:
And from my perspective, that also ends up in the hen egg problem. You complain the user experience. So you do not use it. So if you do not use it, the browser or the standards compliance. Yeah, the W3C.

Paul Kinlan:
I mean, that was my perception. That’s kind of my perception as well. Right. Because I do track some of the conversations that are still going on around this. And it was a very similar issue. Right. Like, we ended up having. There was a. Yeah, the chicken and egg problem, which is, you know, I click on edit an image or I want to edit an image. Right. Like, well, what happens if we don’t? There’s no app on the other side. Okay. Well, yeah, we need to think something. So, you know, we ended up having the Chrome Web Store be the kind of the resolution service, like the broker, essentially. You know, we got into some issues like, you know, we, We. We tried to bootstrap it with the view RSS intent. Basically, I want to view an RSS feed and, like, let any application handle that. And then I think Dave Weiner was probably the first and biggest critic, and it just brought up a lot of attention. I don’t think he was wrong, by the way, necessarily. Like, he just wanted, I don’t want an app. I just want the XML feed because there was XSLT and you could put style sheets on and all that type of stuff. And we didn’t think about the implications of how people use existing XML or RSS feeds in the browser. But it drove a lot of traffic because there is a demand problem. Like, people click on an RSS feed and then get nothing. They get like, xml and it’s like, yeah. And so we were driving heaps of traffic to all these RSS feeds. RSS readers, sorry. And it was. It was great. But yeah, there’s a. There’s an issue of, like, what happens if there’s no app? What happens if, you know, like, Firefox doesn’t support this. And Firefox was. And I pals with Mania, who was kind of working around it, but he had a proposal called Web Activities that kind of Came from Firefox OS as well. And so like, you know, we had a whole like tensions with the browser ecosystem and we never got kind of standardization or commitment from other browsers really to ship this. Um, and so yet you end up with a. Well, it’s not really implemented everywhere. I don’t, I, I actually, I want the integrations to work and like you know, we had fallback story anyway, it sounds like a very similar set of things and my argument’s always been like the web is just progressive by default, right? Like, or it can be right if we choose to be. And so you know, I was thinking about this thing of you know, a client and a server, you know, a Mastodon client. Jump back to your own server. I suppose, you know, if the browser doesn’t support the technology, you’ve still got the paste, paste your, you know, your server or your handle into the thing and it’s like an additive, everything’s additive on top. But I do feel like as an industry we always want the complete solution. And I just feel like with this, the way that kind of like intent based systems work is you’re never going to get the complete solution and the web is never going to get the. Like I had this, this blog post on my, my personal blog called the, like the Lumpy Web. There are certain technologies which you know, you can avoid, right? Like blue web on web, Bluetooth. Right. If you don’t care about Bluetooth that you never need it. But like if you want to build a very Bluetooth specific like centric service, you, you can only ever target Chrome. But you know that, you know that going in when you’ve got all these kind of like things that look like core pillars and infrastructure in the ecosystem and it’s not implemented in Safari or it’s not implemented in Chrome and it’s just like. Oh right. It’s like it looks like the solution and then you can’t actually ever use it, but it looks like it’s a core piece of infrastructure and it’s like it probably is a core piece of infrastructure but it’s like it’s not a load bearing piece of infrastructure anymore because like just no one like it’s, it’s not ubiquitous and it’s like anyway, yeah, we get into this space where yeah I think sometimes as engineers and you know, creators of platforms and stuff, like, and this is the ecosystem, I mean not Chrome, like we want the ideal best working solution and you know, what we ended up with with web intents is you know, the main thing that People wanted to do was share links, right? Let’s get Navigator Share, right? Let’s just do that one first, right? And then all of a sudden like you start to bite off like small parts of the problem. And you know, Web Intents, I suppose, as a concept lives on in Navigator Share. It’s also. You can do the Share Receiver, is it Target receiver. I can’t. I can’t remember what it’s called now. And so your applications can register to receive the share and like it can register to receive files and it’s like, oh, okay. Actually, we’ve solved some of these problems and some people have been very successful with it. And even though it’s not yet ubiquitous across the, you know, all the browser ecosystem, in a lot of cases it’s not the core feature, right? Like the. The or the core need of your. Your application. And so I think the. Was it, you know, perfect is the enemy. The. I can’t remember. I. I don’t know what the phrase is, but like, it’s that kind of. I always worry about all any of these kind of like massive philosophical change for how we think about something on the web. It would be a great solution. I think Web Intents was actually. I don’t think Web Intents was the right solution. I’m happy to get into that later. We made a lot of mistakes. Designing that API especially would have bit us like coming five years later, 10 years later, it would have really hit us and bit us. What are the smallest things that you can solve? First is something I learned from this whole experience. And for me, with say the I, obviously I got my own Mastodon service Status Kinlan Me and Paul is my handle if anyone wants to follow me. But you know, actually that’s the thing is I actually feel quite alone on my own server, right? It’s just my own server with me on and I don’t go to another site and follow. Like if I’m not in my own Mastodon client on Status Kinlan me, if I find someone’s profile, I’m really not encouraged to follow them, right? Because I. It’s like press follow and I can just type. I can just type a URL in, but it doesn’t autocomplete properly and a whole bunch of other things in that space. I’m just like, oh, you know what? I don’t even know whether this. I feel really bad for saying this. I don’t even know whether this person is worth the effort of me to. I don’t mean it that way, but like it’s.

Matthias Pfefferle:
Yeah, but I totally agree because in the end that is the big hurdle for all geeks like us that host their own little island in a decentralized service. Because that all will be solved on the bigger instances with proper autocomplete. With most of the big interesting people are already known on that instance, so you can simply search for them. But if we are talking about single user mastodon instances or WordPress installations, it gets hard because the instance itself is not known in the bigger ecosphere and it’s hard to find and follow others. And if it’s hard for us to copy and paste URLs or email like identifiers, it’s even harder for, let’s say, normal persons that try to use that.

Paul Kinlan:
Yeah, yeah, yeah. It got me thinking. So just like when we go back to this kind of like, bit we were talking about before, which was, you know, discovery of services, you know, it did make me think where. It’s like, you know, when we said that we made the. Obviously the Chrome web store was the broker. It does make me think around the client could actually be the broker. Like your Mastodon instance and my Mastodon, we’re all sharing the same. I know, I mean, obviously I know you’re on the WordPress side as well. But like, like if you’ve got Mastodon instance, I’ve got Mastodon instance.

Matthias Pfefferle:
Let’s talk about Mastodon.

Paul Kinlan:
That’s. Yeah, yeah, it’s like, yeah, yeah, like Mastodon. We can all point back to Mastodon, right? And there’ll be some way of like sharing state between them that can be the broker. But like, it’s not like, I get it because like people say, well, it’s still brittle because it’s not completely federated. Like the browser could be the complete federation piece. Like where it’s like it’s the user kind of maintains all this stuff. I just don’t think we’re ever going to necessarily get there. Right. And influencing browser standards and specifications is a really hard job, you know, unless the ecosystem is also willing to start landing code in browsers as well and also get the commitment from the browser to accept that code, I think it’s going to be an uphill struggle. But, you know, a common commonality is the, you know, I don’t know, you may be using Peer Tube or whatever it is, any of these other ones. It’s like there’s a creator of the software and they can also potentially maintain a way of being the broker for the actions and the intents, because that’s the main discovery service. And so it’s like, okay, and then maybe it’s those islands, those big islands or the software itself that actually talks to, I don’t know, Peertube talks to Mastodon infrastructure or whatever it is. But yeah, I just don’t know. I’m just. I always get back to this bit of the philosophy behind this, I think is great. The web wasn’t built with that philosophy in the first place necessarily. Android was. Android was just like, by default, this intent system is just like built in. But even then, I think if you look at the Android ecosystem, Intents are used obviously for the same app, opening up different kind of views and a whole bunch of other activities. Sorry, but you know, the actual ecosystem of open, like, Intents and knowing, like, hey, I’m going to call the. I don’t know, I don’t even know what app I’m going to use. But like, you know, actually call in another application. No, no developer tells you call this intent. Like, I’m not going to. I’m going to give you an API. Like, you know, just call this intent and open this. Like, very few applications do that. And like there was a project at the time of when Web Intents started, when I was working on that, it was called. I think it was called Open Intents. And the idea behind it was to kind of aggregate and index, like all of the known Android apps and this API surface areas that they did it. I think it died of death. Right. Like, it was just one of those bits where it’s like people weren’t encouraged to kind of go and like do communication with other applications. And so, yeah, there was a bit there. It’s like Android was built with intensive at its, at its heart for this kind of like, new way of linking to functionality.

Matthias Pfefferle:
So you are saying that even with the system baked into the os, it won’t be used by app developers?

Paul Kinlan:
Yeah, maybe. Maybe some will. Um, but like, you know, at the same time. So again, so this is the other bit where it’s like, I didn’t see a resolution service. Like, obviously Android’s got the Play Store. I didn’t see a service that went and say, hey, I need this intent, right? Or I’ve got this intent for this data type. Can you go and find me a service that does this? I don’t think they did that. It was just a mistake. And so maybe, maybe that’s the issue for us. I’m really wandering around now, but. And I’m just kind of philosophizing what happened but so obviously with this view RSS kind of thing that we did, it brought a lot of attention in kind of like senior leadership inside Chrome at the time. And it was just like, what are we doing? Like how do I trust any of these things inside the list of apps, right? Like we’ve got, I think maybe Feedly was on it and a couple of others were kind of sharing like saying that they publish intents and like the VP at the time was just like I don’t trust any one of these applications. I’m going to give some data to. I just don’t know at the moment. And that was the start of like the, like we got way more attention than we probably should have done. And then it was the start of the like oh my God, actually is this project the right thing for us to be doing? Um, and so yeah, there was a bit there of yeah, Android never solved that problem. You know, trust is a really important thing around this. Especially when it’s, you know, you’re sharing out like data out like you know, you’ve got maybe a personal image and then you just pick, you don’t know any of the like terms and service of what the image host might do and so you end up with this kind of like, you know, trust issue. I don’t know whether it’s the same for this though like, because by the way, when I, when I look at a bunch of this kind of like you know, intent based stuff with the, the Fediverse, it’s to always come back to your site in most cases, right? Like it’s the, I’m on another site and I actually want to send the data back to my instance and make sure that it’s just like seamless and beautiful and smooth which is I think is the little bit of the inverse. Like we’re going from like my site or my thing out to any one of the service that we don’t know that exists anywhere else actually, maybe, maybe that’s actually a benefit to the, you know, the kind of the Fediverse side of things. It’s like it’s actually really coming back to you versus you going out to other services. Maybe I don’t know at the moment.

Matthias Pfefferle:
And so yes, but as you mentioned, I, I kind of understand why the current implementation go like you described, kind of mastodon having a mastodon index and they use this index to somehow autocomplete so that it’s easier for you to go back to your site to start the follow process or the share or you name it. But on the other hand, it’s very limiting because Macedon is building its kind of ecosystem thingy. Pixel Phat is starting with something similar. So it’s always there’s an index, we use that index for auto completion and every service is a bit biased because they focus on their service. So it again does not really feel decentralized and it is still hard for people because the usernames are twice, triple, 10th. It depends on how you, what name you use. So even an autocomplete can be very difficult because you at least have to start also with knowing domains and stuff like that. So kind of makes it a little easier. But from my perspective it doesn’t really solve the problem.

Paul Kinlan:
Yeah, I suppose I don’t quite know enough about like what each platform inside, you know, built on Activity Pub is actually trying to do and whether they’re even trying to solve this thing. It’s like, yeah, I don’t know whether you’ll end up with a lot of different solutions to start off with and then, then formalize it later on. I always get the sense of that. It’s like, yeah, we should do it but then we can’t agree on a plan of how to do it and then, and then you end up like, you know, finding yourself.

Matthias Pfefferle:
You were my last hope. So I think in the end there were a lot of good ideas like formalizing for example endpoints so that there

Paul Kinlan:
is

Matthias Pfefferle:
a known endpoint that describes which endpoint should be called which when it’s about a, like when it’s about a follow. So the, the only big issue these days is how can we make the identify ubiquitous? How can this. Yeah, yeah, yeah, yeah. Simply be known. Everything else I would say is solved. It’s simply how does the other side know who you are?

Paul Kinlan:
Yeah, so that was, Sorry, that was the. Is it the fep? I don’t, I don’t know the, the terminology. Is it FEP or FEP the 3B 861 is it? That is like the. Actually like that’s, that’s the server to server infrastructure. And then you know, you haven’t got the. Yeah, I mean, yeah, it’s. This is so going back to even before web Intents which was the. I need to knock my coffee over then.

Matthias Pfefferle:
Sorry.

Paul Kinlan:
It’s like gesticulating everywhere. Going Back to Web intense.com, which was my like emulation of. Basically it was an IPC that worked quite well in a world where we didn’t have third party segregation between, you know, the double key cache or the triple key Cache for. I mean, I get why we’ve done it for privacy. It makes a lot of sense. But you know, you end up in this situation where you can’t have one shared iframe across a number of running websites, which means that you can’t have this kind of concept of, well, I’m going to store some state somewhere and then go and query it from another site. Right. And so, you know, as an example, one way that could have done this, right, maybe is there is. There is some centralized service or whatever it is. Maybe it’s Mastodon’s kind of back. I know that there’s some issues with just like, well, maybe we don’t want to centralize control too much for Mastodon clients and stuff, but if you can kind of Mastodon. I know I’m making this up. Mastodon.com My instance and your instance have a iframe embedding for like, you know, current user. Right. Right now. And this is the exact. It’s. It’s the right solution, which is you shouldn’t be able to discover the user from one page onto another page. Shouldn’t be able to do it. But you know, we’re. We’re at this situation where, you know, they’re all webintents.com, you know, you would just share state and you’ll be able to communicate between them. They wouldn’t even know that they’re talking to each other directly. So this. So I’m gonna tangent this, by the way. There’s a technology on the web that no one uses. It’s incredibly powerful and it’s been around since, I swear, it’s like 2010. Right. And so we have this way of sending, you know, you post message. So you got message channel right? Between, you know, two, two different ins. Like two different sites. Like iframe window, you can open up a new window and have a message channel between them. Cause you can post message and then receive the messages coming through. On top of that you have this thing called transferables where you can actually transfer the raw value from one page into another domain or origin or whatever. Actually pretty cool. There’s a thing called the message port, right? And so the message port is like essentially the two ends of post message. So it’s like you’ve got the thing that you can push into and the thing that receives the thing, right? And then obviously you can go back to the other way as well. That’s also transferable. So you can create a message port on one site. You can open up. I have A demo for this somewhere. Maybe it’s on Glitch. I’ve probably lost it right now. And so you can open up a new tab, right? A new window. I can create the message channel on my instance, pass the message port to this new tab, which is some resolving service somewhere, right? It’s just like you open it up maybe. Actually, this is how we might do it actually. Sorry, I’m just thinking about this on the fly now. Because there is no double key cache between the two. You just opening up a new window, it’s like you get some information. That’s great. But then you can open up the third page, right? So you can have a picker. So this is the idea was that you have a picker of different sites and services. Sorry, I think I’m right. I might actually try and build this properly. But. So this new tab, the first tab that you’ve opened up is. Is your kind of like open id. You know, like the kind of wall. Like the wall of things or, you know, the list of services that you’ve got. You open that up. That’s great. What. What can I do with that? Well, obviously you can send the data back, which is actually pretty cool. But if you want to actually pass it through to the next instance, like, how do you do that? And obviously you can, you know, query string it and stuff, which is in. Interesting. That’s a normal way of doing it. But because it’s a transferable message port, you can just pass the message port to the other Mastodon client or whatever it is, close your picker window, and the two Mastodon clients are actually talking to each other, but they don’t know who they’re talking to. They just know that there’s something on the other end of the channel. And, like, it’s one of those bits of. Just like, no one ever talks about message channels and message ports. And like, what you can do with the trans. Like the transports. But like, it’s incredibly powerful what you can do. So you could have this picker that then opens up the actual. Any one of your, like, main servers that you’ve got. Like, you might have, like, you know, I’m on status Kinlan me, I might be on. Actually, I can’t remember any of the names of other servers now. I might have multiple accounts. Great. Well, that picket does the thing, and then I just route it through to the other account. And then I’ve got the two browsers talking to each other. That’s web. That’s a web technology. It doesn’t solve Native apps kind of by talking to each other and stuff. But there’s a bit there of. There are a lot of other technologies that we can explore inside the browser and if we want this to be mediated by the browser and kind of looking at it like, obviously register protocol handler was one of the things I was looking at at one point, but there is the, like, there’s a lot of other things that we can do and you know, there’s a world where the. Again, I’m waffling like crazy at the moment now. But like maybe getting back to the original thing I was going to say was like, the double key cache is critical. Right. Like, we need that whole thing because we just don’t want to leak stuff. Um, but there are other solutions that might be possible for us to kind of actually work out how to actually have two instances communicate and talk to each other delegated with the user in control. And I just want one of those bits of. Just like the web is just so vast and broad in terms of APIs and technology that, you know, the solution might just be sitting in front of our face and we’ve just never considered it before. Right. As well. And so, yeah, I don’t know, it’s one of those bits of like shared workers. I thought was our, like promised, you know, the promised solution then you know, what doesn’t work on Android. Right? Like, it works, you know, it’s man, like, it’s one of those bits of just like shared workers. By the way, we give Safari a lot of stick for not, you know, doing stuff on the web properly. They put shared workers on mobile a good couple of years ago now and like we’re sitting there going, we haven’t got it. Anyway, sorry,

Matthias Pfefferle:
so that was one of your more recent blog post with reinventing web intents.

Paul Kinlan:
Yeah, actually, let me check. What did I say?

Matthias Pfefferle:
I think that was on your site 2017 where you get a bit into the.

Paul Kinlan:
Oh, yeah, maybe that was it. Yeah, yeah, yeah, yeah, yeah, yeah.

Matthias Pfefferle:
You already kind of formalized all of that already and it still didn’t catch up.

Paul Kinlan:
So that was. That was the exact. Yeah, man, I’m just reinventing the reinvention again. Just in my head. Like I talked about this before.

Matthias Pfefferle:
Yeah. So there. We will link your blog in the. In the show notes that it is. It is amazing. All you can follow the whole process of Web intents are awesome. And we working on that. Then. Okay, Web intents failed. Why did they fail? Web intents are still important. What could be the next step then? Rewise some months later. It’s really interesting. Follow that process and follow your thought process.

Paul Kinlan:
Yeah, so I think it’s so funny that like yeah, I got this idea, I’m inventing it right now. No, I talked about it a little while ago. That’s why I had it in my head.

Matthias Pfefferle:
A little while is 10 years now, right?

Paul Kinlan:
It’s gone up to 10 years.

Matthias Pfefferle:
It is almost 10 years ago. A little while. It’s funny because most of the things you mentioned also with the post message thing was also something I stumbled upon ages ago. It was called X auth that kind of tried to solve that problem for OpenID and OAuth and stuff like that to kind of have being that iframey domain that could be used as a message broker. But it was always, oh, that is an awesome idea. But then it felt into the let’s formalize that and you had a working solution and then it was kind of over engineered in a way. So yeah, we have to think about that properly. We can’t simply use that even if it’s working. We have to formalize that stuff. And it never was formalized and it was never used. So it’s a shame that most of the time it is even if you have or if you try to work on a simple solution that it simply didn’t catch up.

Paul Kinlan:
Yeah, yeah. I mean this maybe gets to the point before that you were talking about was just like, well maybe we want something about but like we want something in the browser and it’s like the, we can build all these like shims and things. Like you know, maybe it’s like post message and you know, you know, passing stuff along, you know, transferable message ports and stuff. I do get this sense that everyone’s just like, yeah, but like that’s, that’s JavaScript, right? And like what we want is an API that’s like Navigator dot authenticate and it’s like okay, right now you have to get really deep into the weeds, right. Versus like everyone agreeing on a common kind of maybe message format and like a way of yeah, cool. Like we’ll just do it over iframes and everything. And it’s. I think there’s a bit here of I’m, you know, I’m, I’m un. I’m gonna go, I’m gonna go for a bigger tangent than my normal kind of conversations. But like you know, I look at how we build on the web and I’m gonna pick on an API. People work on these things. So I feel really Bad calling it out. But it’s one of those things, you know, you’ve got like, things like CSS carousel. It’s very neat and you know, it’s not carousel, but like, it’s a core set of primitives that enable carousels. You know, part of the thing is like, well, you know, we’re doing it because like, the developer experience of making a carousel is terrible. Right? You have to install a module and some JavaScript and all that. So. Yeah, but we’ve got billions of carousels out there and they’re working pretty well. Right. Just across the board. And like, people who want to maintain and build a business around carousels are incentivized to maybe do the best, you know, kind of user space version of the thing. And like, there’s a real tension between, like, should we standardize something that kind of works pretty well and it’s developed entirely in the user space and run inside the user space, or should we bring it into like the kernel of the browser? And I actually don’t, I don’t know the answer, but I do feel like there’s a bit here of, you know, I think a lot of people feel like the validity of the thing, or maybe it’s permanent or something is like it has to be rooted inside an API, inside the browser, versus actually we got fine solutions in user space and they all work. And you know what, the industry’s kind of like, you know, rallied around or standardized on certain kind of like, concepts and. Yeah, so I kind of worry sometimes that we go like, you know, the, the perfect solution has to be a native API, right, you know, inside the browser. And it’s like, we could just solve the problem today, right? Maybe like we, we’ve got some solutions and it’s just like, yeah, I don’t know, I don’t know where I sit. And it’s, it’s, it’s just a tension that’s been in my, kind of in my, my head for the past like five or six years of.

Matthias Pfefferle:
It’s, it’s, it’s funny because I understand your view on that, but from my perspective, not going too deep into that special case, but in general, I really like the idea of there is a working solution that solves a need and it’s working and everyone tries to build that because it really is a need. Let’s make it easier and have a standard for that. I think that is kind of the right way to do it. Kind of having a lot of proof of concepts and then try to finalize that into an easier and standardized way. But I can’t understand why this is not a thing in the intense area. So it’s, from my perspective it, it feels like there is either kind of an island like solution or the shouting at W3C that we need a standard because we can’t build otherwise. So yeah, for me there is a gap between these two extremes and there’s no real work in between.

Paul Kinlan:
So yeah, I’ve got an idea. I mean when we talk about intents, we use it in the browser context for like the old web intents, right? Maybe a little bit different to kind of the, you know, the federation, like what we’re talking about in terms of activity pub. I think that it’s a massive mental hurdle to get across an over. Hey, there’s a fundamental way we’re going to change the way that the browser works. Like it’s going to like you want to find application functionality, we’re going to do this thing and it’s just like. Hang on a sec, Paul, right, we’ve got links and then like the web’s been working fine for now and it’s like you want to change the way that like LinkedIn works. Is there a market for this? And it’s like, and like you get into all these questions where it’s like a fundamental it. Like I think linking to functionality, which is what I was always arguing like web intents would be, is just an incredible mind shift, right, for a lot of people. And it’s very hard for people who run, you know, browsers of billion dollar businesses, right, just to, just to build them nowadays, right? It’s very hard for them like when they’re choosing to allocate all their resources and it’s like, well actually we can add this feature which we know a customer wants, which does this, this and this. It’s actually relatively easy, I think for them. Not relatively easy, but it’s easier for them to think about how do they prioritize that against this like paradigm shift, right? Because I think intent based stuff is a paradigm shift for the browser. And it gets me into this bit where, you know, I did this demonstration and I’ll talk about large language models, but I had a very, a very interesting demo. I think I blogged it on one of my things. But I really love the concept of transclusions, right. You know, it’s like being able to kind of pull content and connect content from two different pages. It’s not a new technology as in it’s, it’s new in the sense that it doesn’t exist on the web properly. You’ve got iframes, but it’s not exactly

Matthias Pfefferle:
what I mean,

Paul Kinlan:
was it Ted Nelson was doing, like, transclusions and just trying to describe them 25, 30 years or maybe even way before the. Even the, like the web as we know it even came into existence. I think it was like sometime in the 70s, like, describing this, like, concept of transclusions never quite got to it. And I was like, well, we have a way now of munging text and understanding kind of like the intent of the text, right? And like the meaning of the text with large language models. And I know large language models in a lot of cases are a very divisive topic around kind of the web and everything. But what I was. I had this demo where you click on the link and instead of open up a new page, it’s like, it assumes that the link is like the link in the, like the client page has got the context around it, right? You know, it’s like the. The content, the text in the link is some meaning. The text around the link has got some meaning. The thing you’re linking to might be relevant, right? And so it’s like it takes the page, the context of the page that you’ve linked to, and then merges it into the current page, right? And I was just like, it’s a very interesting way of browsing. Way, like you shift and shift, click. It doesn’t open a new tab and you navigate to. Takes the context of that page and then just like blends it into the current page. And it’s like, it’s a very good demo. It’s very compelling. And actually, when you start to use it, it’s like, it gives you a little bit more context to give you confidence to then go and click on the link because you don’t know what’s on the other side of the link, right? And like, this is the same, by the way, with web intents and app and apps is you don’t know what’s on the other side, right? It’s just a risk, right? The browser’s actually pretty good at mediating some of those risks. But this demo was like. And it was like. I remember I presented it to a whole bunch of, like, VPs and directors, and it was like, whoa, this is like, incredible. Okay, cool, we’re gonna do it. And they were like, yeah, other priorities. And like, you know, it’s a fundamental paradigm shift, right, of like, how you think about. And it. I don’t think this would have worked just I just bluntly I don’t think it would have worked with normal people because the way that links are expected to work has changed. Right. And anyway there’s like you have to deal with all these different hurdles when you are trying to prioritize like yeah, like it. We think that an intent based system is going to be the right solution for the future but if it’s fundamentally different I think it’s, we’ve got to have a way of actually trying to get the buy in right I suppose and the willingness and the commitment and it’s. I, it’s a people problem, it’s not a technology problem. Right. At this point I think as we’ve seen with all the different proposals is there are solutions and how do you build up consensus to actually get these things implemented? I, I haven’t solved that. All right.

Matthias Pfefferle:
woocommerce.com so if you were able to restart all of that, how would you start or what would be the simplest issue you would try to solve?

Paul Kinlan:
I mean you mentioned this before but like you know we’ve got a user space solution, you know a widget or a plugin or whatever it is. Just get that working and like you know, get people using it to actually get to the point where it’s like it’s clear that a more permanent solution’s needed. I don’t feel like we’ve got, I don’t feel like we’ve got any progress anywhere in those spaces. And so for the kind of the Fediverse one I think just like even though there would potentially be a lot of trade offs it might not be completely federated and a whole bunch of other stuff but like you know, solve the user experience, experience and the problem as best you can and Then work out whether, you know, other things are needed. I think that’s one piece. The success we had off Web Intents was we got a lot of other APIs. You know, we got Navigator, Share. Share Target, that was the name of it. Share Target API. You know, we got a couple of those different types of things. We’ve got like, like I think file system handling now, not file system handling, the mime type handling in progressive web apps, you know, we’ve got all these different solutions that have come through. And you know, also at the same time, you know, you’ve seen that there isn’t a huge amount of demand to use those features either. And so it’s one of those bits of, like, navigate to that share. I think that was actually a really good solution and it’s probably a really good next step for us. So I would do those and then I would also look at other opportunities, like maybe if I jump back to Web Intents and not say the federated, you know, activity pub stuff. Again, it’s a large language model, but related with. And again, I know that there’s a lot of issues, but like, there’s a thing called WebMDP, right? And it’s in, it’s in development and the idea is like, you know, a website will describe all the tools that it’s got for an agent, you know, some sort of agent to be able to talk to it. Oh, I need to, you know, I’ve got a thing that can edit images. I always use edit images. Great. Well, you know what, a developer’s actually quite incentivized at the moment to kind of define something in that space because, like, if there’s an assumption that agentic software is going to be, you know, a thing in the future, we need to find functionality and we need to be able to kind of then route the agent to that functionality. And I’m like, sounds like web intense. All right. And it’s like you get to a point where, you know, there’s enough value around other parts of the ecosystem where if you get every single, you know, application saying like, what it can do. Well, that’s what we wanted to do with Web Intents. We wanted people to say that, you know, I can edit images or I can book flights or I could do all these different things. We were trying to have this way of declarity describing what’s possible. And then we would have the web store index as much of that as possible. It was via the manifest file. So it wasn’t on the pure open web. Well, you Know What? Something like WebMTP is also starting to look at this type of thing now where it’s like if every single site tells you what it can do, what it can work on someone, I don’t know whether we’re gonna do it. I suspect we probably wouldn’t because I’m not necessarily that forward thinking sometimes. Like if every single website is starting to describe what it can do, all of a sudden you can build an index of functionality and it doesn’t have to be agents calling it, it’s people calling these things. I want a website that you know, books flights, right? Boom, click on the link and it’s like, okay, cool, I know that you’ve been to these things. These two sites offer this functionality. I will send you to it. And it’s like, you know what, we’ve got the whole thing, we got web intents back again. But like without even knowing it and it’s, yeah, there’s a lot of options, right? And it’s like, yeah, I don’t know where we’ll go with those, with those types of things. And again it comes down to like some other different raw fundamental primitives maybe that will actually reconnect like the. Oh actually this is how we’re going to do it. And so instead of like the pure web intents like as the concept and so yeah, I don’t know. My advice to anyone who’s going into this space is just like at least just try and solve a problem right now and then try and get like consistency in usage and then standardize it a little bit later on. Because if you can show demand, I’m sure people will actually start to think about it a bit more or maybe just go to that reinventing web intents post. Quite happy with that post actually. What do you think? Am I just curious? What do you think would be the best way if you had your two pennies worth, that you could tell the ecosystem what to do and what they should focus on?

Matthias Pfefferle:
That’s hard to say now that we talked about for about an hour because before we talked I really thought the browser is the one who has to solve the problem. It’s kind of, it has to know who I am and pre fill or kind of automatically trigger the process I wanted to start. But you’re right in the direction of this is maybe the wrong way because there is no. Even if we all think there’s a need for that, there is no real. We haven’t tried hard enough so that we can prove, okay, there is a need and we can’t solve it on our own. So there needs to be something. So maybe we really have to go back to see how we can work around the issue. Like for example with the post message thing through an iframe or build browser plugins that simply know who you are and autocomplete forms you could have something

Paul Kinlan:
like the, I said the W3C but like you know, actually I don’t know how the activity pub standardization works but like you know, there could be some kind of agreed form forum that is like, you know what we’re going to do, the resolution service and you know, it’s maybe easy to fork but it’s easy to manage. There’s consensus built about how we think about things. And I’m not saying use the kind of the, you know, the post message thing but like you would get a solution actually pretty quick that should work across at least multiple platforms in the browser. I don’t think it’d necessarily work on Android Apps and iOS apps. I don’t know how that would work.

Matthias Pfefferle:
Yeah, but I think because I agree with you on maybe start simple because in the end you already said that with Android Intense I think we do not have to solve the problem on every device and for every app, but simply start maybe with the web because all other platforms has their own issues or work differently, have different paradigms, I don’t know. So I would be happy to at least solve or start with solving that particular issue and then evolve on that. So yeah, that from my perspective would already help a lot. But yeah.

Paul Kinlan:
Do you think there’s challenges that I’m happy to keep on chatting by the way, but like do you think there’s. What’s the biggest hurdle then to get over? Maybe there’s like a small common solution. Is it a people thing? Right, where it’s like actually there’s multiple kind of key pieces of infrastructure in kind of the, you know, the federated, you know, activity pub space and. Or is it. Can you solve it with one and then kind of see whether it works

Matthias Pfefferle:
and then I wouldn’t say it’s a people thing, it’s more how you try to solve the issue. So I think it is no problem to have at least at the beginning, a central service if as you said it’s a trusted one, maybe kind of hand that over to the W3C or kind of in the web there is the social web foundation, maybe something like that can then own that piece of infrastructure. But the current solutions are more island solutions and Mastodon is working for example on a kind of also decentralized discovery indexing thingy. But that is quite hard to implement that there is OAuth involved, there is client registration and stuff like that. So this is really something you have to implement on a platform level and it’s not something you can unobtrusively put on onto something kind of that fills out forms or something similar easy JavaScript snippets. So it’s. You really have to not all go all in, but you have to really decide okay, this is worth it and I want to invest in that and implement that into my service. And it’s still. They come from a perspective of everything is a platform in the, in the Fediverse because they run Mastodon, they focus on Mastodon. So from their perspective it’s easy to connect this index thing because the admin is connecting that and then every user has access to that. When you see it from the WordPress perspective or the self hosting perspective, it’s different because the end user has to know how to connect a blog with kind of an index thing, why that is important, what problem it solves and then you have to run through the OAuth handshake and it’s maybe kind of scary. So it’s from, from an end user perspective, it’s not really how you would design a service. So yeah, it kind of brings us into a step more into the direction of this could solve a particular problem. But yeah, it’s from my perspective it’s too big. I want to have something simple like the other side has to only know who I am, nothing more. Everything else is kind of solved. My side would know what endpoint to trigger. I, I simply have to know to, to which side I should delegate to.

Paul Kinlan:
Yeah, yeah. I feel like I’m not saying it’s gonna be the same with everyone else but like for me there was an ego thing right where it’s just like I can solve this entire kind of resolution thing. And it’s like. And you know, it wasn’t until we actually solved one just, just like share is the thing that we need to solve, right? And it’s like I had to have a lot of like personal introspection at least anyway. And so I do think that there’s a. The problem space is so big that. And you know, the actual solution looks relatively simple. Oh it’s just like let’s just get something in the browser or let’s just do this. And yeah, it’s one of those things of just like, well just it’s. It’s very enticing to try and solve this whole thing. And I’m just less convinced that we’re ever going to be able to do it, especially in one shot. Like, kind of like, hey, there’s one API that does the whole thing.

Matthias Pfefferle:
I would agree on that.

Paul Kinlan:
Yeah, yeah, yeah, yeah.

Matthias Pfefferle:
But as I said, I don’t think we would have to. In the end. I would be very happy to at least have really the identity thing solved.

Paul Kinlan:
So, yeah, I think I was looking through a bunch of the proposals. I do think it’s like if you can just make it easy to route back to whatever service that you want. Like the Activity Pub system worked really well. Like I built. I think, I think you actually one of my only followers on my other posteris thing I did, I built like an email based Activity Pub system. I think email’s a great format for a lot of stuff, but it’s separate and it’s not. Again, it’s not the browser but like, yeah, it’s Activity Pub itself is. It’s complex and like the security, like, you know, I always found like signing headers and you know, getting the certificates and it was just that hard step. But I could build my own server and like, it will talk to like Mastodon client. It would talk to every other client out there. And it’s like we’ve got something really cool around Activity Pub, right? And it’s like, yeah, maybe it is this. I just need to find me right from my other. This other site and it’s. That’s it. Maybe. I don’t know. Yeah, it’s.

Matthias Pfefferle:
So how can I bring you to get into that again and help solving the problem?

Paul Kinlan:
Oh, geez.

Matthias Pfefferle:
So. Because you, you were one of the guys that started all of that. So I think it’s only fair that you will be part of the.

Paul Kinlan:
Yeah, I’m always happy to help. You know what I have you got another 10 minutes? Are you okay with 10 minutes?

Matthias Pfefferle:
Yeah, sure.

Paul Kinlan:
Okay. All right. Okay, good, good. Um, so Web Intents was my first foray into web standards and web specifications and I, I kind of vowed to never go back into it. It was one of the worst experiences ever. Oh God. Yeah, yeah, yeah.

Matthias Pfefferle:
I’m sorry to make.

Paul Kinlan:
No, no, no. But it’s. You know what, I realized that. So people are going to kill me for this. But like there’s a. I don’t know what the name of the system is. In Google sometimes they do these like training courses and you know, it’s basically, it’s designed to give yourself some introspection Right. And there’s a colors thing or whatever it was, and like, Google simplified it down to like four colors. Like, an orange is like, you know, people who go off and happy to do 80%, but then they’re not the completer. The starter, but not the completer. And then there’s people who like to be very. Anyway, it’s just one of those bits of. Just like, lots of people in on my team hate these things just because you can’t classify people as these kind of like, different groups. And I’m like, I get it, but I’m not. I’m just using it for me to kind of work. Like, I realized a lot about how I work. And it was all about the same time you were going into these places where, you know, a lot of strong opinions, some very, very. You probably see the meeting minutes actually if you were to go back, you know, very influential people who worked around these spaces, people throwing their weight around. I didn’t handle it very well either because I was just like, this is obviously the right solution. The solution that we had, I don’t think was the right solution. On reflection, it was a. Maybe a. I don’t want to say argumentative, but like, you know, it’s a very robust process to kind of like, work out what the standard is and everything. And I’m not necessarily set up for those kind of discussions personally.

Matthias Pfefferle:
Right.

Paul Kinlan:
And so, you know, diving back into that space would actually be very hard for me. And so one, because I’ve got opinions where I always think I’m right. And right there is that thing. I also hold onto thoughts. As we’ve seen from today, an idea from 10 years ago is still on my mind as being the solution. And so I realized at the time that there are people who are way better in that space and they’re way better at collaborating and negotiating and coming up with consensus and all those types of things where I’m not necessarily built for that. You know, I like to tinker, I like to experiment. Chrome Devrel, for me, has been a great place. I like, I’m tinkering as a. Like, tinkering is my job. Right. But, like, you know, we do attach it to, like, actually, how do we actually improve the web? And so I don’t know where I’d fit. And as I can have ideas, I’m very happy to have ideas and I’m very happy to talk to people. And like, you know, one thing I do like actually is kind of like, we do, like, mentoring type of things. And I Like seeing other people be successful, right? And like, you know, get to a solution and all that type of stuff. Could I do the work that these people do to then kind of like actually get it over the line and get the consensus that they need? Oh, no chance. I, like, I’m not, I’m 100% not, not that kind of person. But it, you know what it took me, it took me doing this whole space because I, I’ll get into some philosophy and like a lot of people on the developer relations team over the years, we, we like doing people like getting involved in standards and being editors of specs. You’ll see a lot of my team as people or people who’ve been on my team as well as people have been involved in specs and you know, my rationalization. It’s not their rationalization, it’s my rationalization. This is. There’s a certain level of cachet, right? Like a seal of like actually these people know what they’re talking about because like we as developer relations, we’re understanding what happens in the ecosystem, but we’re also asking people to build things and do stuff, right? And you know, engineer to engineer, you want to know that you’re talking to an engineer, not someone that’s kind of like in a, like a salesy kind of thing. And like, so one way to do that is like, especially if we’re not building in the browser because it’s, you know, it’s a lot of work to change Chrome. Well, you know, like a lot of my teams, like as patch Chrome, a whole bunch of other stuff as well. Like Francois Beaufort as an example, like, he’s like in Chrome all the time just building things for his devrel job. He’s very good at it. But it’s like a, A lot of times like, you know, spec editing and everything and being involved in the W3C and you know, you know, working, you know, what’s the name of the tpac? I go to tpac and Very valuable. But they also kind of like, well, that’s how I’m actually going to show that I’m, I’ve got the skill and the knowledge and I don’t think that’s the case. Like, it’s just one of those bits of, Just like I think everyone’s going to have their own path, right? And you know, they can do that, but I don’t think it’s needed. And like there’s definitely a lot of people on my team who just like the second they, they touch this space, it’s Just like. Nope.

Matthias Pfefferle:
Right.

Paul Kinlan:
But yeah, there’s a, there’s a bit there just like. Yeah, I completely forgot where I was going with this. But like the. Yeah, I’m not necessarily going to get back in. I’m very happy to help people think about these things and you know, they can bounce idea. People can bounce ideas off me and I can, I can throw ideas out and like um, you know, I also like looking at the, the boundaries of the platform. I can. That’s what, that’s what Devrel has been good for me about. Is this like. Yeah. Yeah. Cool. I’m just gonna walk across the DOM and see whether there’s a new API that’s coming to the browser that hasn’t been documented yet. Oh cool. There has. Right. Like let’s play with it.

Matthias Pfefferle:
And I totally understand that I’m, I’m also not the, the standards guy because it’s. I’m. I’m way too pragmatic for that to, to kind of generalize everything. I’m happy if things work simply and I’m just learning that with my work on Activity Pub a bit more and I’m quite sure how this turns. I’m quite interested how this turns out. So maybe we have another discussion in some years. But it’s funny that you already think about the standardization process. Um, because I would love you to kind of build one of your ideas on. On from your blog that simply works that, that would. I, I would be so happy if we were at least at that point and then think about okay, yeah, I

Paul Kinlan:
might actually go back and I realized I was actually looking at my blog post when you about the reinventing web intents I was like oh yeah, I remember this. And then I was like yeah, it’s on glitch. It doesn’t work anymore but I can, I can probably build it out pretty quick now. And I was like maybe we can get to a point where it’s like I’m very happy to demo and like just try ideas and all those types of things where it’s like, you know, we could, you know, have something where it’s.

Matthias Pfefferle:
So there is. So there is an online Fediverse event thingy called the Fedi Forum and they always search for it. Always. I think it’s a two or three day thing and it always starts with presentations and then it’s a non conference where you can simply talk about ideas. So I think that would be. I can share you the link. I think they tried to have to a year so yeah, that’s quite nice. And that would be a perfect thing to pitch ideas like that. So.

Paul Kinlan:
Yeah, yeah, yeah, yeah.

Matthias Pfefferle:
It was funny because when I prepared for the podcast and had another look at your blog, then I found the Reinventing Web intents.

Paul Kinlan:
I’m glad you did, by the way, because I completely forgot about that and

Matthias Pfefferle:
I thought, yeah, that is totally the solution. And I think somewhere in the blog post you mentioned that you already built something. And I said, okay, I simply have to wait one or two more weeks or month. And then I saw the date when you published it and they’re like, okay, it was 2017. Okay, maybe.

Paul Kinlan:
Yeah, yeah, you know what, it’s one of those bits where it’s like this whole space is. And this is actually one of the things I liked about Activity Pub as well, is that there’s a lot of people who want this. Right. In this whole space of just like federation, the software’s open communication is an important thing between people building social circles. And you know, it’s, you know, it was nice to see all the different types of activity, but it’s not just mastered on Right. And everything as well. It’s a nice community and it’s one of those ones where it’s like, you know, I did a tweet not too long ago. Was it Tweet? Anyway, I posted, but someone’s like, I was trying to push Chrome to like get kind of Activity Pub integrated into the browser. Didn’t get there, didn’t. Wasn’t a thing that we thought we could prioritize. But then I was realized, like, you know, I don’t necessarily know whether how much it makes sense outside of maybe posting and stuff. But it’s nice that you can, you can just. I built my own, like I said, I built my own Activity Publishing kind of host and Instance. And you know, it’s not too complex to do a bunch of these things now as well. It’s pretty cool, right?

Matthias Pfefferle:
And they’re currently really pushing into the client to server API. So that would be.

Paul Kinlan:
Yeah, that’s what I was looking at as well. I didn’t see much progress around it at the time.

Matthias Pfefferle:
And so there’s a working group currently that’s really focusing on that and there are quite some platform providers working on implementations and at least two or three people working on clients. And that would be a perfect browser plugin, I would say.

Paul Kinlan:
Well, I mean this was the bit where I was like, actually at one point I was like, why can’t we just have the share button inside Chrome? It’s like, I’m going to share it. I got my instance registered. I’ll just share it straight to it. And he’s, yeah, I just want to share links, right. And like, not everyone wants to do it, but, you know, like, can you remix web pages? Can you just take a quote out of a web page and share it to your local. Like, there’s all these bits where it’s like, I don’t. I don’t. I never know whether Chrome’s ever going to do this, but, like, maybe another browser could, like, try and experiment with like, what. What does it take to integrate kind of social interaction and sharing? And like, you know, you bookmark something or you have. Instead of a bookmark, you also have a like button for any. Any URL. Great. It’s a post to the. It’s not. It’s a. No, there’s a. There’s a bookmark activity, right? Isn’t there. Is there a bookmark activity?

Matthias Pfefferle:
I’m not sure the indie web has bookmarking, but I’m sure.

Paul Kinlan:
Was it safe? I can’t remember. But like, there’s action and there’s one of those bits where it’s just like, yeah, great. Like, there’s loads of ways that we could think about this. And it’s like, you know, all your interactions within the browser are just like, you know, actions on. Maybe it’s like a private instance that you’ve got of anyway, and then you could. I don’t know. Right. But like, maybe another browser could do it and then show the way forwards.

Matthias Pfefferle:
And so Activity Pub as an intent API, technically, it’s built around intents with the.

Paul Kinlan:
Yeah, so this is the thing. I read the 3B86 and it was going through all these. I’m like, this looks like Activity Pub. Right. Anyway. Right. It’s like, it’s all built off the same stuff and it’s like, oh, okay. And so it does get back to what you mentioned about. Just like, Just like, let me resolve who I am and then. And I know the server and my web finger works and all these other bits and pieces. It’s like, yeah, yeah, yeah, it’s fun. But yeah, in terms of, like, getting back involved and stuff is like, I’d be very happy to, like, you know, catch up with people. If people ever want to know how to maybe think about stuff, they go fight the battles with us in the standards groups and all that. I’m very happy to kind of, I say mentor people, but like, it’s.

Matthias Pfefferle:
I’m always happy I take you by the word. Sorry. I take you by the word.

Paul Kinlan:
Yeah, of course. This is also the bit where Devrel is. I get to sit down and talk to people and work with people and like, you know, we’ve worked with so many people who have just been way more successful than, than me individually. They build businesses off and it’s like, oh, I look at it and I’m like, yeah, I had a hand in that. But like they’re never going to tell anyone. It’s fine. But it’s like, yeah, it’s, it’s a nice, I find it a nice role to be in actually. And so it also aligns with my job. I can put it in my own kind of like my, my performance review.

Matthias Pfefferle:
Perfect. Okay, so then I book you for the next FEDI forum and I will see.

Paul Kinlan:
Yeah, send me the details and see what, Is it an in person thing or is it online?

Matthias Pfefferle:
It’s the same situation as here. You can sit in your chat, let me know, then simply participate. I think they thought about having that in person too, but it started and I think it will go on as an online event.

Paul Kinlan:
Does it use activity pubs to organize everything behind the scenes?

Matthias Pfefferle:
Shame on them. No.

Paul Kinlan:
Yeah, yeah, shame on them.

Matthias Pfefferle:
No, I think there are quite some event platforms out there, but kind of a mixture between events and organizing a conference itself. This is still a problem to solve.

Paul Kinlan:
Yeah.

Matthias Pfefferle:
There is still need for an activity pub enabled in between thingy.

Paul Kinlan:
Yeah, I do was. Sorry, this is a separate conversation up, but I always wondered like, how long does a activity pub instance have to live? Right. I think the expectation is it’s a social network that lives forever versus a snapshot in time of one event that everyone goes to. At the end of the event it closes down, but then you’ve got your entire record of every single thing that’s happened in between. Right. Because it’s all done over activity pub. Right. It’s like, I don’t know, I’m just

Matthias Pfefferle:
like kind of to spin up an instance only for a particular.

Paul Kinlan:
Yeah, yeah, yeah, yeah, yeah, yeah, yeah. And if you can resolve that anyway, it’s one of those bits I don’t, I don’t know yet at the moment. And so I would like to move my presence.

Matthias Pfefferle:
You have to participate a bit more. You don’t have to build that, but you’re full of ideas. That’s amazing.

Paul Kinlan:
Yeah, I’m building so much at the moment. I know again, people don’t like large language models necessarily, but it’s enabled me as a manager to actually get back into building stuff.

Matthias Pfefferle:
Yeah.

Paul Kinlan:
Like, it’s crazy at the moment and so. And I like giving back as well. So it’s. Everything I do is always open source.

Matthias Pfefferle:
So.

Paul Kinlan:
Yeah, let me know, let me know. I’m happy to speak to people, you know, give people my two pennies worth, go away over time. It’s fine.

Matthias Pfefferle:
So thanks a lot for taking your time, Paul, and talk about the bad days, the dark days.

Paul Kinlan:
Yeah, yeah. Specs, mate.

Matthias Pfefferle:
Of your standards work.

Paul Kinlan:
Standards work. A very particular type of person who thrives in that space and I think it’s good for them. It just wasn’t me. And some people are very successful and they’ve been very good. And yeah, I wasn’t.

Matthias Pfefferle:
But, you know, at least you’re. You’re my hero and you solved a problem years ago and I was so shocked that I think you published that on the post that it will simply stop. And I was about why that worked. That worked perfectly fine. And now everything is.

Paul Kinlan:
No. So there’s. There’s a really fun story. So I know you were trying to wrap up. Sorry, but like, I’ve got another fun story. Sorry, but. So I was doing Web Intents and I think Remy Sharpen asked me to speak at FFCONF in Brighton. Yeah, great. It’s cool. And then it literally, I was gonna talk about Web Intents and the project shut down like the week before. And so, you know, I. But it wasn’t public at the time. And so I got on stage and I’m like, here to talk about Web Intents, but don’t know what to say. It’s closed. And everyone’s. Everyone in the audience was just like, what? Because, you know, it’s like, you know, I had a 45 minute slot to do a talk. God, I had a different talk already. It was kind of fine.

Matthias Pfefferle:
Okay, you switch. You really switched talks then.

Paul Kinlan:
Yeah, the week prior. Yeah. Like, we had. So we were also building out this Chrome apps platform. I don’t think it was what Remy wanted in terms of a talk, but it was like, you know, we had this platform where we were trying to push on the boundaries of what the possibilities and capabilities were of the platform. And Chrome apps came again. They’ve gone now as well. And so it was just one of those bits I was trying to get people to think about, like, how the platform might evolve more advanced, like in terms of like interacting with the system. Right. And all that type of stuff as well. Because Web Intensive was one part of like the interact communication. But I was very into, like, I want the web to be the operating system of the future.

Matthias Pfefferle:
Yeah.

Paul Kinlan:
And so to be the operating the system of the future, we need more access to more of the hardware in a safe way. Chrome wasn’t at the time willing to or able. We didn’t think we had the ability to land some very advanced APIs, like if you want to do USB and all type of stuff. We, we didn’t think we could ever get that on the web. Right. Just because of, you know, people were concerned. Right. And so if you silo it inside this kind of like native, like wrapper. Not native wrapper, but like something that’s Chrome only. People know what they’re getting, developers know what they’re getting. Right. And it’s like, maybe in the future we’ll be able to make this a web standard, but show demand first. I don’t think we did show demand, by the way, but, you know, there’s a. Yeah, so I ended up kind of switching it to that one. But yeah, there was a bit where it was just like, really I was booked on to do this talk about the future of web, like the future of the web via Web Intense. And it’. Yeah, it didn’t. It was.

Matthias Pfefferle:
So you are also part of the I can flash my ESP through Chrome thingy.

Paul Kinlan:
Yeah, yeah, yeah, yeah. So it was, I think you may. People might know like the Fugu project and stuff. And we’ve been all over. Actually, it’s, it’s been pretty good. It’s like there was another internal name. I can’t remember what it was called. Not supposed to leak internal names. So there was another name before it was like Fugu was the public name. But yeah, there’s always been this push to kind of like really kind of. Actually this is the tension between multiple browsers and I totally understand it. One, Chrome is very well capitalized to kind of push and it’s got a lot of engineers pushing on the platform. But you know, there is a belief at least like for, for Chrome, you know, through Google, is, is the web, is the, is the platform and we want the web to keep growing in terms of capability and richness and expressiveness and you know, every single, you know, anyway, there’s, there’s a lot in that space. So like, you know, where we choose to invest was making it so that you could do more with the browser. You know, in some cases you get WebRTC, which enables this conversation. In other cases you get, you know, web Bluetooth, which is a web usb, which is very useful, but it’s just such. So niche, right.

Matthias Pfefferle:
I think the whole home automation industry is at least the open source part of that is relying on the USB part. It’s really, really. I can flash software on my ESP stuff using Chrome. I do not have to install CLI stuff and flash that into in the terminal. I simply click a button in my browser. That’s it. That’s amazing.

Paul Kinlan:
You know what, actually I’m going to take that back to the team because like, you know what, what tends to happen is I don’t want a woe as me story but whenever we do these we will do work and like you know, obviously we’ll, we get a lot of feedback from people who are very against versus the very poor and pro. And so what happens is we tend to, we do listen to it like we do listen to all the feedback and we take it to heart and like we want to make sure everything’s kind of the best it possibly can. But yeah, then we don’t see like the success that other people have necessarily as much. And so you know, yeah, like maybe the homebrew kind of or maker ecosystem is doing really well in this space. You know, we’ll see some partners, you know, I think Verizon or someone used a Flash station for some phones and stuff in the past as well. And it’s like yeah, yeah, we never, we tend not to see, we beat ourselves up a lot which I think is fine. Yeah, we never see some of those kind of more like oh yeah, cool. But like, but we do get to this situation where it doesn’t work everywhere. And this gets back to a very early conversation where it’s like, you know, a lot of people when they’re building software rightly have to make these things work everywhere. And you know, that’s also kind of maybe the, the thing that holds some of these proposals back around, you know, things like web intents or you know, federated stuff is just like, you know, what doesn’t work on mobile because register protocol handle is not there. And it’s like, you know, if you can get around it and say well actually we’ve got a progressive experience or you know, parts of the experience just don’t work, it’s fine. We’ve got a fallback that might actually be a, mentally as an ecosystem that’s maybe what we support. And like, you know, I know people like Jeremy Keith, right. Very pro progressive enhancement and like, you know, but, but the industry’s not always like that. It’s like has to be this minimum bar, right. And that’s also why baseline Exists. Right. Like one of the things. So I’m going way over now, mate. Sorry about this, but, like, baseline exists as a, as a concept in terms of how we want to communicate the platform. Because we found that developers didn’t realize what was available. Like, oh my God, Safari is really slow, slowing down the web. And like, if you were to believe that whole there are differences, right? You can see the differences in some charts and stuff, but if you believe that story completely, you end up going back. Well, you know, God, like, I know what’s available. You look Back to like 2017, when that other blog post was made, or that’s what I’m trying to target for. And if you actually look at the progress that we’ve had as a platform, there’s been a massive shift in terms of capability where the majority of users or your people who use your sites actually are way further advanced. Right. And way further on in terms of what is possible. And developers just didn’t realize that gap in between. Right. And so, you know, there’s a bit here of like, you’re thinking about things like progressive enhancement, but like, when you’re thinking about that, you’re also like, way far, like back in the past in terms of your understanding of the platform. And it slows everything down and it slows the quality, reduces the quality of experiences that we can offer people. And it’s. I mean, there’s still challenges, right? I’m not going to say like, baseline is the solution to everything, but, you know, things like progressive enhancement are a way that we know that we can solve a lot of the user experience issues in terms of capabilities and everything in that space. And, you know, our concept of the platform actually is way further behind than it should be in terms of, you know, what is supported everywhere and what you can reliably use. Yeah, I think there’s loads of challenges around these things. And so it’s. This is what I think. It’s always a people thing. Right. It’s meant. It’s like, what do we understand versus, like, the technology is always. I don’t know, there’s a lot of solutions for a lot of things. And it’s, you know, getting over people’s, like, hurdles of whether they choose to do something or not is probably the biggest thing I see. Sorry, I went completely off.

Matthias Pfefferle:
That’s totally fine. It’s. It. The, the podcast lasts for as long as we have something to talk about, so there is no limit.

Paul Kinlan:
I’ve got a lot to talk about. I got a lot of time as well. It’s Friday, Friday morning. The UK America doesn’t wake up for another couple of hours.

Matthias Pfefferle:
So the, the, the only, the only people we could upset are the listeners. And we do not do that for the listeners. It’s simply we having a chance, really. Thanks a lot for taking your time. But as I understood it, right, this is work for you, so even better.

Paul Kinlan:
Yeah, yeah. Talking to people, I love it. Yeah, it’s great. It’s my job.

Matthias Pfefferle:
Getting paid for podcasts.

Paul Kinlan:
Paid for podcasts is great.

Matthias Pfefferle:
And you teased that you want to have, want to start your own podcast about the LLM stuff.

Paul Kinlan:
So, yeah, yeah, we’ll see how it goes. I’ve got my other blog, which is AI focus, which is aifoc us. You cannot find it in search engines. For whatever reason, Google has decided not to index it properly. And also I’m not allowed to ask them to fix it, like, because we can’t. Like, honest results is a really important concept. Okay. So I have no clue why it’s not indexed properly, have we?

Matthias Pfefferle:
What a shame. You’ve worked for Google and you’re not indexed.

Paul Kinlan:
Yeah, I mean, if you use Bingo or Brave Search, it’s there. It’s great. But, you know, my Google, who needs Google? But yeah, so I’m, I’m. One of the things for me right now is to explore this. Like, there is a new technology that’s come out. It’s changing the way that either we think about building software, even creating content, discovering content. You know, is there a world where more people put their content behind a paywall because they don’t want the content to be consumed by, you know, these tools and a whole bunch of other stuff. And so I’m trying to explore all the different challenges in that space because I think it’s important and I think there’s not enough people looking at this intersection of where the web goes. It’s either completely built into, like, I’m going to go LLMs all the way and I build my own app and then like that becomes the platform or no, we’re not going to do it completely. And it’s like, I think the web has got an opportunity to thrive and be the place where it’s like we can create more content, more experiences, more apps, more games, more everything. And it’s also the view on all of that, right? We have all the text turns out large language models like text. They like HTML, JavaScript and CSS and they can output it really, really well now. And it’s like I’m trying to explore all those, all these different tensions and stuff, and it’s not going to be an easy time. I don’t think it’s going to be a pretty hard time and a rough time in terms of some of the change that’s coming through and kind of getting on top of it and so see, we’ll see. But I’m also happy for, like, maybe there’s a, you know, what was it like 25 million web developers. Like, there’s a lot of web developers. Might be 25 million more in the next couple of years. People building web experiences. They might not call themselves web developers, but I think the web’s looking lots of challenges, a lot of opportunity for the web, though. So I’m trying to explore that and I’m going to try and do my own podcast. But we’ll see. I need to stay on track.

Matthias Pfefferle:
Right, let us know. So then we can dig a bit deeper into the AI world and maybe Web Intents will.

Paul Kinlan:
Web Intents and Web Intents.

Matthias Pfefferle:
How to replace Web intents with LLMs doing the.

Paul Kinlan:
I’m going to say. Wait, wait, wait, wait, wait, wait. Do you want another 10 minutes conversation? No, we should probably finish. I mean, there’s a lot of, there’s a lot of stuff that could potentially happen, right. If you can build software unbelievably quickly, you know, do you. Do you actually need like, you can get, you can get something built to solve your issue and so do you. Do you even need to resolve another site? I don’t know. Longer question.

Matthias Pfefferle:
Just this morning I stumbled upon a site. I forgot the name it was about. It is kind of an automatic measurement how likely your company will be replaced with AI in the future. It’s something called like, can my SaaS service be a markdown file? And then you put in names and it kind of gives you a percentage. And I tried it with Activity Publish and it said you can’t replace a protocol with AI. And then there was kind of a. You see that with rss, it died so often or it was called so many times. It was really funny. And I’m really not sure about the AI kind of. I would love to see Activity Pub evolve over time. But it’s also an interesting discussion. If you have an agent that can do stuff for you, do you really need open protocols in the end? So there is a baseline like HTTP and stuff like that, but do you really need a unified standard or is it. Do you simply need a API that is.

Paul Kinlan:
Sorry, Sorry about this. I’m going to go on another rant Now I don’t know and I don’t know what the future is for standardization in terms of things like the W3C and it’s a whole new area. But if you get to a world like there’s a chip now which is 17,000 tokens per second or yeah, 17,000 tokens per second, maybe we should end on this. Actually, I don’t know whether we should. But like, um, I don’t necessarily know about pages. So like everyone’s worried about kind of large language models generating content for pages and a whole bunch of. I, I kind of get it. Um, but if I was to say, like, you know, I want to build a system, you know, an app that yeah, updates my, you know, flashes my chip, you know, whatever it is, Arduino or whatever it is, and we say we didn’t invent web USB and web Bluetooth, but we had this technology that generated code unbelievably quickly and it was small enough, you know, to be on a chip, on a computer or whatever it is. So it’s all local. Why don’t you just build the browser around the page? Right? And like we, we know how the nate, like the actual native interaction works with the operating system for Bluetooth and USB and all those types of things. Why do we need a standard to define how, you know, a, a web page would talk to that hardware when the, the systems could just be built. And so I’m, I’m actually preparing this blog post at the moment, trying to work it out because it’s a little bit more science fiction. I don’t think it’s going to happen. But like there is a world where it’s like the page is the source of truth, what the author’s intent is the source, but then all the infrastructure behind it is just built by the browser as it needs it. And then you don’t need to think about a specification for interacting with the hardware. But then the activity part bit though is kind of interesting because it’s the protocol to talk to other machines and other tools and your silo and my silo and everyone else’s kind of connecting. Like I don’t think the software is gonna know how to do that right at the moment. I just can’t. I can’t see those things. It’s like email. It’s a pretty well established kind of communication standard and the tools can use it and like, I don’t know how I’d get like my vibe coded thing onto your machine to then talk. No, no, no. Actually we need the protocols and we need everything else in between.

Matthias Pfefferle:
Yeah. But take it on a different level. So in the end it’s about communication. So you have activity pub to communicate from your instance to another instance. But what’s the difference between this communication? I hope they will not hate me for that. When you could have an agent that simply goes to your site, to the comment field and write a comment there and checks the site if you reply to that comment. So I think there needs to be ways you could use to interact. But the question is, do they have to be standardized or do they have to be simply there understandable for an agent.

Paul Kinlan:
Yeah, I don’t know. I don’t know. Yeah, it’s actually a really good question. I don’t.

Matthias Pfefferle:
We are in a weird state.

Paul Kinlan:
Yeah, yeah, yeah. And it’s divisive as well. Right. But that’s why I’m also thinking about it is because I want to be on the forefront of a shift and a change in, like, how does the web actually thrive and survive and grow? Because I think the web is actually a fundamental platform that it’s changed everything. Right. I wouldn’t be here, obviously, without this. And like, I think it’s still gonna be the central part of how we. How all of our tools talk to each other, how all of our people just talk to normal people like we’re doing now and all this type of stuff. It’s like, I’m very hopeful, but there’s also gonna be a whole heap of challenges. And, you know, people have questions about like, you know, privacy and energy use and, you know, copyright and, you know, these are all things that we’re going to have to try and solve as well. And it’s like. Yeah. So I don’t, I don’t want to seem like, I’m sorry, we probably shouldn’t have ended on LLMs. Right. But like, given it wasn’t about that in the first place. But yeah, I don’t want people to go away like, oh, my God. We’re just talking about. This is the thing. Everyone’s talking about it. Like everyone was talking about crypto or something. But there is something. There’s a shift, something that’s happening and the technology exists and the web exists. And I think the web is the best platform. Right. And it’s.

Matthias Pfefferle:
Yeah. And I think you have to care about the development because if you want to protect the web itself. So you have to know what’s going on to protect the web. We currently have. So it’s not about, at least for me, it’s not about blindly following the hype but it’s. But I think on the other side I can’t ignore it completely. So it’s somewhere for the good and the bad. So it’s.

Paul Kinlan:
Yeah, yeah, yeah, yeah.

Matthias Pfefferle:
You have to somehow adapt to. Yeah be able to follow that. And yeah, it’s hard, it’s funny.

Paul Kinlan:
But this is my tinkerer brain. I like it in that space but we have to deal with challenges and I’m also happy to help with Activity Pub as well. If anyone ever asks.

Matthias Pfefferle:
Coming back to Activity Pub.

Paul Kinlan:
Yeah, I’ll bring, try and try and bring it back.

Matthias Pfefferle:
I would love to have you in one of the next 30 forums talking about your experience and maybe present a possible quick and dirty in between solution that we can already use based on a 2017 blog post.

Paul Kinlan:
Yeah, that completely forgot about that, you know. Brilliant.

Matthias Pfefferle:
It was me. A pleasure, Paul.

Paul Kinlan:
Yeah, thank you very much for having me and thank you everyone for listening.