Justin Wohlstadter describes building Wonder, a learning platform designed as exploratory pathways through web content modeled on human apprenticeship. He recounts failed approaches including a Chrome extension that injected code into pages and an iOS browser, and explains why controlling the environment proved harder than augmenting existing user flows.
Chapters
I had a question. I don't know if we should wait until you finish your presentation and then ask, or should we— I want to see what you have to present. Okay, yeah, yeah. So maybe I'll hop into some of the history stuff and then we can get back to the product stuff. I love the product stuff, so we'll get back there. So as I mentioned before, I'm sure you guys have seen this little drawing a lot, right? Like, it's not a fast and quick and easy and straight path to success. Yes. The last CEO I worked for, Andrew Toy, who now runs Android for Work, he was the founder and CEO of Divide. He had a great quote he used to tell me, which was, you don't raise money to build a product. You raise money to screw up 5 or 6 times and then build a product. And it's so true, right? We raised money, and had someone told me 2 and a half years ago that all you have to do is build a text box where people can enter their question and build a research network, We would have built it in 3 weeks, at least the first version of it in 3 weeks, right? Instead, we basically had to go through 2 and a half years of building the wrong thing and screwing up and making a lot of mistakes to get to the point where we now can build a very simple, beautiful product. So it doesn't just end up being this way. It takes a lot to get there. So what I want to do is I want to walk you through a few of the versions of our product that many people have never seen because we never But before that, I'll start with the vision that kind of guided us all the way through. So as I mentioned, I did my master's in the future of education. The reason I built Wonder is because I spent a lot of time thinking about how we harness all this great information on the web and how we use it to replicate this learning, this human learning model that has existed for thousands of years. Existed for millennia in the real world, which is human to human, right? Like this apprenticeship model that we've had, you know, where someone who's very wise and learned on a topic area pairs with someone who knows nothing and tutors them over many years to get them to the point where they too are a master and an expert. On the web, we have all this information and none of it is— there's no, you know, master or guide To walk you through it. And I spent a lot of time thinking about, you know, what if we could build, you know, this future interface where a kid would walk into a library one day, right, and walk up to a wall and be able to endlessly explore pathways through knowledge by seeing the path that experts, that others had taken in the past. And that was the inspiration for Wonder, right? Could we apply that human learning model that has existed forever to all this new cool technology we were building. In the beginning, I took this way too literally. And we built, like, crazy interfaces, which are even hard to show on here, to do this. Like, things where you actually felt like you were moving through the web. But I'll show you some more basic implementations. So the first version, the idea was the browser seems like a natural place to do this. Right? The browser is where we access information on the web. So what if we built an interface on top of the browser to do this? If you think about browsers today, and this is still the case, browsers are dumb frames. Right? So you go to a browser, and the way you navigate this amazing thing that is the internet is you type in a URL, a location, and it takes you there. Right? It's basically the equivalent of, like, us having teleportation. Right? And like, we want to go to the store? Great, we press a button and we're at the store, right? And we want to go to, you know, Africa? We press a button and we go there, right? It's cool. It's fast. It's efficient, right? But there's something beautiful about wandering down the street and being able to actually find that, discover that cool coffee shop on the corner or that bookstore that you would have never passed, right? We don't have that experience on the web. There's no, like, path that you can take. So the idea was, what if we built into the browser itself? At first we were like, there's no way in hell we're going to build our own browser, because no one's going to download a browser on their desktop, right? So the most natural extension is let's build a Chrome extension. It's easy, it's quick to build. We'll just put that together quickly, and we'll test it, and we'll get it out there, and we'll see how it works. So the first implementation of Wunder, which was— or it was actually called Viber at the time, or Wayfinder. Um, this is here at the bottom. Hopefully you guys can actually see that. Um, so literally we had this like pathway and like branching paths that you could see on any page that you could kind of hide. And as you went through the web, you would create this trail and be able to explore other trails that people had taken. Um, we did a lot of stupid things that we shouldn't have done that were way too complex. We tracked like the weather at the time when you were on pages, the location. Because the idea was Like, what if you could remember where you were when you explored a part of this journey? We took it, again, literally. It was way too complex. Not only is it really difficult to build on top of someone else's interface, which is basically what you do when you build an extension, you can't control the whole environment, but we just built so much unnecessary stuff that we didn't have to build. We went through literally 12 designs here. Here's another later design on the right-hand bar. So again, this whole concept of pathways, and you could save and share your pathway and explore other people's, which is why the explore button. It was crazy complex. We had to deal— we had to basically inject— how many of you have a coding background or something? Yeah, so basically we had to like inject code into every page that someone loaded, which was crazy intensive and would crash people's browsers and it was awful and don't do it. If you take anything away from this talk, don't do it. So that was nuts and we realized we were getting nowhere. One of the big mistakes we made is we actually never We never really released it. We kept iterating and we just weren't happy with it and we thought designing and designing and designing would eventually solve this problem and it didn't. It was just the place that we were building it and the implementation itself was just way too complex. How many users did you have at this time? We never launched it. Okay. Yeah. So we put it in the Chrome Store I think at one point and we had a few dozen friends on it, but we were never happy enough with it. I'm a perfectionist. Sure. And so I was like, this is not going out unless it actually feels good. And it didn't feel right. So then we were like, you know what, maybe the browser was just a mistake. So what if we were able to control our environment? People are much more willing to download an app than they are to download a desktop browser, right? So let's build for iOS, right? So we decided to build an iOS browser. We did a bunch of crazy designs, like, again, like pathways through the web. You could, you know, see forking, branching paths. We could control a whole bunch of things. Environment, which was a lot more stable and efficient. But again, it was— we realized it was very difficult to get people into a new flow, right? Like, people are used to using their browser, right? So for us to educate people that they shouldn't open up Safari or Chrome and instead that they should open up Wayfinder, right, and do all of their browsing here is a very difficult proposition. So another lesson here is like, don't take people out of the flow of like their normal daily lives, right, and what they're used to. You can supplement it, right? You can augment it. But removing people and forcing a new behavior is very dangerous and very difficult to actually do. And I'll get more into that. But the other thing is that when you're designing an interface, I don't know how many of you have design backgrounds, but it's basically, it's very difficult to build an interface that does everything that something someone's used to does, but then does a whole different set of behaviors, right? We're basically trying to merge 2 functions: normal browsing today and this like exploratory pathing into one interface, and it was just a disaster. Again, we never really released it publicly. We were never happy with it. We released it to like 100 people of our friends, and we were just We were never satisfied with it, so we kept working on it. This was literally 2 years in.
