--- date: 2019-03-10 modified_at: 2019-05-20 tags: [entrepreneurship, productivity, programming] description: How I built Dunbar, a relationship management app, in just 26 hours by leveraging high motivation, Expo, minimal boilerplate, and open source libraries. --- # I just made an app in 26 hours I got this Aha-eureka moment when I started thinking about relationships, at about 22:30 on a lonely friday night. How do I maintain my relationships? Well, it's hard. All current platforms don't really help you with it. If they do anything, they make it harder. You see, on Facebook and other social media I have ±600 friends. This is not a maintainable amount of relationships. When I go to these platforms, I don't connect with my close friends, I just sparsify my attention to many loose connections, and this only makes maintaining the closer relationships harder. Thanks to Robin Dunbar https://en.wikipedia.org/wiki/Robin_Dunbar, it's known that 150 is a magic number of the limit of relationships an average person can maintain. I really think there should exist a tool to easily see and maintain the relationships that you care about, and be able to reach this magic number. To create a PoC I just need a very simple offline app that can get some contacts from my contact list, put them on a screen, and have a cool UI that has leads to real human interaction. Let's create an app, and call it Dunbar Code I started by creating a new folder, github repo, and launch expo init in the terminal. I made a few sketches of what the app should look like on the home screen. The rest of the app was a rather straightforward copy/paste from the iPhone phone app I already use now. So after an hour, I had a list of functionalities and screens the app needed to have (See this GitHub Issue https://github.com/EAT-CODE-KITE-REPEAT/Dunbar/issues/1). Then I started coding, and I kept adding more specifics about the implementation in the issue on the fly. In total it took me 6 morning- or evening-sessions of, in total, 26 hours. Of these 26 hours, 20 hours was coding. 4 hours was planning out features (create a huge backlog), brainstorming, pre-validating with others, and testing, and 2 hours was bringing it to the stores of Android and iOS. The first iteration (Iteration 0 - the Proof of Concept) code will remain open source in this repo https://github.com/EAT-CODE-KITE-REPEAT/Dunbar/. The app is already usable, but further iterations will improve it greatly. Validate This week, I'm going to validate the PoC and get feedback from others. Then I'm going to start with Iteration 1 and, hopefully this month, set it live. I can't wait to see what others think of it! I can't wait to see if it gets any traction. This could very well be a hit app! How can it be so fast? There are many reasons for it, but I think these are the biggest ones: 1. My motivation to create it was super high, so I was about 50% more productive. A few times I was in a busy cafe, and I was completely zoned out and focused on writing code, while others were talking around me. Also, turning off my phone wasn't needed because I didn't care to check it at all! 2. Programming is about reading old code and other people's code more than writing new. Normally, reading old code takes 75% of the time when writing code. Now I didn't have to read ANY code, making it 4x as fast. Let's say that, once I'm in a further iteration, I can't always be just writing and never reading code. Code just has to change. But the big advantage shouldn't be overseen that I am the one that wrote all the code. And if that's recent, this will mean that I'm not wasting as much time on reading code. Let's say that in further iterations, the time reading code will be about 50%, and writing code also 50%. That means that I'm 2x as fast at that on average! 3. At most projects I've worked on there was complicated linkage with native code and no building automation from Expo https://expo.io. This easily takes about 30% of all coding time. Using Expo and staying in the realm of their possibilities they offer, removes all native Android and iOS stuff, making it 1.3x as fast. 4. On most projects I've worked, discussions, prioritizing, planning and reviewing takes about 3 hours every 8-hour day (37.5%). Because I did this alone, I just needed 4 hours out of 26, which is 15%. This is another improvement of time of 19%. 5. I used some smart ways to reduce boilerplate a lot (one of which is Redux light, I call it). This left me with an app of just 1500 lines of code. On most projects I've worked on, the boilerplate was already set up in a very extensive way, and it's not easy to change this. Therefore, I think I needed about 20% less code, if not more. 6. I was able to rely on some libraries I wrote in the past and open sourced, one of which is React Native Data Forms https://github.com/EAT-CODE-KITE-REPEAT/react-native-data-forms. This library creates a form in ±100 lines instead of a few thousand lines. Because of this, I probably needed to use 50% less code. (improvement with factor 1.5) 7. I kept components and my UI/UX simple and close to native UI. This meant that I didn't have to deal with any complicated animations or modals, or anything. These things can slow down your feature development a lot, maybe up to 100%, but let's say 30%. I've seen this a lot on my other projects. These 7 reasons meant I was able to make an app 1.5 x 2 x 1.3 x 1.19 x 1.2 x 1.5 x 1.3 = almost 11 times as fast as when I'm working on a slow legacy perfectionized product in a big company. Can this really be true? Yes, I think it can. I made it in 26 hours. The next iteration will probably take ±62 hours, which brings me to a total of 88 hours. This would take 3 developers ±2 months to build something similar if they didn't have these 7 advantages. This would cost a company at least €28.800,-. Not bad if I can produce the same result in 2 weeks on my own :)