Local-First Software: Taking Back Control of Our Data | a mini-doc

I believe that local first is the future, but I also believe that it has to be the future. It's going to change [music] almost every aspect of how we interact with technologies. [music] We want to give users back the control over their data. that I truly believe [music] in here for the next 10 years giving us just better software. [music] It's your data now and forever. And I was sitting there on the train and I was listening to music from station to station. It would go into tunnels where there was no cell service. And I remember looking at my phone and the music I was listening to stopped and the playlist disappeared and I thought, where did it go? I I just had that. This is so stupid. Why have we done this to ourselves? Number one, no spinners. Your software should be instantaneously responsive to you all the time. I'm currently an associate professor at the University of Cambridge. I'm the lab director of Incan Switch. Madam Wiggins, I'm probably best known as founder of Heroku as well as Incan Switch and Muse. The background story of local first starts in about 2014. We went out and spoke with people in the field who are building things, who are studying things, who are writing things. With cloud software, the primary data copy is on the cloud server somewhere. And if that cloud server locks you out, you lose access to all of the data you've ever created. So we saw this sort of like mismatch between the duration of people's work and the lifespan of the software that people were writing. The spark of the idea for local first for me really started with a syncing problem. Seemed very simple. I've got some data on one phone. I want to get it to another person's phone. It's very private data. I don't necessarily want to send it through a cloud service. And I just looked for a technology to do that because it seemed like it would be possible and it didn't really exist. And that got me interested in following this path. And then I came across a research paper on this technology called conflict free replicated data types. This this was just an algorithms paper, you know, but I read it and I thought ah this seems like a good fit for this goal that I have of trying to give the power of back to users again. Yeah. [laughter] Yeah. that that's the concise is going to be the hard part. Let's see. So rolling and we began the research project and it was there that I met Martin Kleman and that was through Peter Van Hardenburg and others who made me aware that there had actually been a realm of computer science still very new at the time a few years old but people who were trying to mathematically prove that it was possible to do this. Not even to build it but just prove that you could. And so we were able to take a lot of that initial foundational research and turn that into a first version of a library we called automerge to try to prove what it would be like in practice to have this very different model from the cloud server ccentric model. And first of all, could we do it? And secondly, what would it feel like to use? And at some point we got far enough to realize, hey, we should give this thing a name. Uh we think that there's there's something really here and something that goes beyond the technology. It was really a set of principles, a set of values even about how computing and storage could work. We went back and forth on all these things. We were calling it serverless software internally. We wrote an essay around it to try and explain what we meant with that. And we came up with the name local first and published that in 2019. Your work should be at your fingertips. It [music] should be really fast. If the Wi-Fi is down, you can keep going. It should be possible [music] to work on on your laptop and later get access to it on your phone. The software should outlast the business, the company, and the [music] computers that it's originally hosted on. Things like Google Docs have given us this world of multiplayer and collaborative editing. Local First [music] thing as well. Security isn't something you layer on after the fact. It's just a property of how the system works. And I think ultimately that is for me what local first is about is [music] data ownership and agency and empowerment for users. My name is Ray McKelie. I'm a product manager at Ditto and I think about developer experience with local first software. We offer a sync engine makes it very easy for people to build applications that work with or without the internet. With local first software the idea is that you have this local copy on your device where nobody can take it away from you. Then that has also some other benefits like for example it's just faster because you don't need to talk over the network every time you want to look at the document or update something. Has the advantage that it works offline. You can edit the data on your own computer and then recynchronize with the cloud when you come back online again later. Things are snappy. That's amazing. But more than that, I think what I'm excited about is the privacy and the control that comes with having things locally. So one of the principles behind local first is that you own your data in spite of the cloud. You can migrate it, you can share it, you can port it away from one provider to the next one. So local first really is the answer to combine both the sort of control uh sovereignty uh ideas around privacy and the real-time availability and collaboration mechanisms. I'm really here for sort of the technical sides uh of local first which is synchronization. Glennar is using some aspects of local first. There's no page loads at all. Immediately when you click a link, you get transported to the next page. In the last 20 years or so, in the stuff I worked on, the user experience has always been like the key driving force behind most decisions. Everything from what the UI looks like to how the infrastructure is built and how the actual product is put together. And we're making something local first not only like provided a better user experience but it was also really important to be competitive in terms of like allowing the early users who decided to switch over to Figma from alternatives to keep working while they were on a train going through a tunnel or something like that. Figma and linear both of those products have gone on to totally dominate their software categories precisely because of the quality of user experience that they were able to build because they were building on this architecture. They had to invest literally millions and millions of dollars to build the core sync engine and local first technology to do it. We have taken so many complex problems out of the equation like error management or error handling by having the synchronization stack you know do everything for you. So that general developers now have the tooling available to be able to build on a local first architecture without having to build all the core technology themselves. We're building Jazz, which is a local first framework where even if you don't care about local first at all, you just want to build an app, it's very easy to do so. But accidentally by using it, you build apps that have all of the local first principles, if your app goes away, or even if our infrastructure company goes away, people will still be able to modify that app a little bit and still use it and get their data out. There's still lots of challenges that need to be solved. Another area that we've been working on on the research side is end-to-end encryption. [music] So this is again something that's enabled by having less of the app run on a server and more of the app run [music] on your local device. It means we can actually make the data encrypted in such a way that the server can't even see the content of your files. [music] The server then is only involved in synchronizing some encrypted blobs uh from one device to another. There's still open challenges with uh that but that's something that is also actively being worked on. Building software in a local first way certainly doesn't come without challenges. It's mostly a challenge though of like rewiring your brain to think different about how you build applications. A lot of the things that we still need to figure out are kind of reaching feature parity with more traditional backend as a service offerings because I believe that if we want to replace the traditional architecture, we need to do everything that the existing architectures do as well. And that's what we're really thinking a lot about now. How can we do all of these things? And so actually local first software which in a way evolved as a manifesto for data ownership and the technology to enable it is now going to underpin the next generation of AI and agentic software. I want to be part of this community because I wholeheartly believe that sharing experiences um no matter where they come from are beneficial to everybody. This is more about ideas that speak to people [music] and people share those ideas instead of one particular technology that is being adopted. The local first paradigm really is one that [music] is very productive for anybody who's thinking about those bigger ways of changing how the [music] internet works and how technology should work. We've been building client server apps in one [music] particular way for many decades. I think this is a new paradigm shift. We're going to be building apps differently for the next few decades [music] and this is really the beginning of it. The fact that so many people are excited about this and share our values is is just wonderful. this is something [music] that we believe in and that we've been finding all the other people who share these frustrations and discomforts [music] and by giving this thing a name, we've started to kind of like band together. I think the way how local first will get widely adopted is by lots of people joining the community [music] using the tools that already exist or building new tools and experimenting what the best way is of making [music] software that fits these values. So, I'd certainly like to think that local first is the future of the [music] web and software in general, but the reality is when you work on a speculative research area like this, you don't know. You're trying to make [music] that future come true. And we'll see if we're successful. The aspiration for this ecosystem [music] is not to supplant or replace the existing software environment. It's to show people that when you don't need this massive complicated architecture, there's a simpler, more straightforward way that you can solve problems for [music] yourself and your community. You know, local first just kind of boils all this down to like one little thing that runs on your computer. [music]