The worst fate, I'm gonna say, is this poor guy who gets torn in half, just straight up ripped in half. [screaming man] [wet ripping] Top separated from bottom. Hi, I'm Lucas Pope, creator of "Return of the Obra Dinn," and this is how localizing the log book put the game in dangerous waters. I worked in AAA for a long time, so I worked on "Uncharted 1," "Uncharted 2." Before that, I worked on a lot of independent games as a small studio. And my biggest title as an independent developer, my current state, is probably "Papers, Please." The core mechanic for "Obra Dinn" is, people would say it's a detective game, and the way I've sort of implemented that is that you're kind of faced with these unnamed people who die in various way and your goal is to name them. So figure out what their actual name is, who they are, their identity, and to say how they died. And it's 60 people on the boat, so you've gotta do it 60 times. "Obra Dinn" was a bit of a weird one. For "Papers, Please," I had this idea that checking passports could be fun and you could make a game out of that. With "Obra Dinn," I didn't really have that. What I had was I had the idea that I wanted to make a 3D game, and then sort of remembering back to my time as a kid playing Macintosh Plus games which were all black and white, I wanted to make a one bit 3D game. So just two colors, black and then white. There's a core element of the game where you're trying to figure out how people died, and actually it's pretty obvious how people died. You just look at them, and they're being stabbed, or something like that. What the game is really asking is testing you to see if you know how they died, so why don't I just develop a system where you just enter textually what they did. So John Williams fell off the mast, or something like that. So that was what I ended up with at the beginning. And I worked that all the way through the development, and that was how it was working. And I kind of noticed along the way that it's actually kind of hard to make a simple grammatical sentence defining how people died, in 48 different ways how they die. So it was difficult to say like, easily correlate how this guy dies with a simple sentence describing it. It wasn't too hard, I was able to work out a kind of very basic system. So like John William fell from the mast, so it would be like two parts to the sentence, or something like that. So building that grammatical sentence was a little bit tricky, and then, generally solved kind of poorly, but okay, good enough, good, ship the game, until I needed to localize it. The moment when I needed to localize it, it became a lot more difficult basically to build a grammatical sentence using the simple interface. Localization is just taking your game which you, say, wrote in English and translating it to Japanese or Spanish or French. It's not just translation, translation is taking the words and making them in the other language. It's actually localizing, so like, cultural concepts you have in English, those need to be converted into similar cultural concepts that they would have in France or Spain, so it's a difficult thing. And "Obra Dinn" is a historical game set in a certain period where different cultures talked about these things differently. They had different terms, they would have different references, different names, all kinds of things that would be different. So when you localize something like that, it's a particular challenge to make sure that the original intent that you have in English is carried over, but also that people can understand it in the same way that they would understand it if they were native English speakers playing it in English. So for example, in kind of a simpler system, you would say, "John killed George." And that kind of thing is very easy to localize. You've got your subject, your object, and your verb. And those get shuffled around for different languages, but generally those are the components that you would use to make that in any language. The problem for "Obra Dinn" comes when it matters how you killed them. So to sort of explain it in an English relative way, in English we have the verb knifed. You can say, "I knifed somebody," and that's both general and specific. It's specific that you use a knife, but it's general that I could have cut your throat, I could have stabbed you, I could've done all kinds of things to kill you with a knife. Most other languages don't have that verb. And to understand that, if you think of the verb to be killed by a sword for example in English. We don't have sworded in English, and I really wish we did 'cause it would have made things a lot easier. So other languages also don't have knifed. When you're designing a grammatical system where you want to work across all these different languages, you need to remember as an English only speaker basically that you can't say John knifed George. You could say John killed George with a knife, but then you're talking about a different sort of grammatical structure for the sentence. And you can imagine that problem expanded across several different languages who all treat sentences differently and also treat gender differently. In English, we don't have gendering. We don't change the verb based on the gender of the subject or the object, but a lot of other languages do, most of the Latin ones do. So you need to think about how this sentence will change if it's a woman killing a man, a man killing a woman, a woman killing a woman, et cetera. The things that you do when you're developing a game in English, generally you project out what they're gonna be when you need to localize it for another language. But this one was particularly difficult to sort of predict and also to deal with because building that sentence is kind of, it's not critical to the gameplay, but it's kind of fun. It was one of the things that I implemented early, and it was fun to just click on the sentence. Part of the allure of the game is that at the very beginning, you can click on that verb, and you can see this crazy list of verbs of like crushed, strangled, eaten, knifed, killed with a sword. And even if those deaths aren't actually in the game, it's enough to trigger the player's imagination about what crazy kind of shit happened on this boat. That element of the game, of that kind of mystery or allure, was important. But when you run into this problem with localization where, how do you express that in all languages? It's enough to make you think twice about the whole thing basically. There are spears in the game that people can be killed by, and they're also spikes in the game people can kill by. And in English, either way, you could be spiked by spear or you could be speared by spear, it's kind of a tossup there in a lot of cases. And based on the scene, maybe you can't tell if it's a spike or a spear. In other languages, those are very distinct verbs in a lot of cases, and they're really dependent on exactly what it is they're being killed by. Suddenly the localizer asking you, "Is this a harpoon, is this a spike, is this a spear, what kind of spike is it?" because every one of those has a different verb for how you describe being killed by it. So the solution was basically just to keep pounding on it. So one of the ways to solve that would have been to make the sentence a little bit more awkward. You would first choose the general way in which they die, so they were killed, they suicided, what else, died like of old age where they were alive, so it was a very short list of things that they could've done. So once you choose killed, "John was killed." Then there's somebody who did it, so you choose George. And then there was like a fourth box which was kind of the implement of their death, so "was killed by George with a gun." The problem with that is that is it's kind of a passive sentence. I much more wanted that initial list of verbs to be the active, so I wanted John shot George. It was more or less just a lot of back and forth. In some cases, it was changing how they died in the scene. So in the scene, you can't see what they're killing them with, so it makes it much more difficult. Okay, I make it more obvious or change the way they're getting killed a little bit so that it is more easily localizable. One of the things about making games by yourself, or at least making the kind of games I make, is I'm not trying to make the perfect game. And actually, this is one of the cases where, especially for this particular instance, it's not perfect. So there are cases where it's confusing how he's dying, and maybe it's okay in English, but it's harder in German. Or maybe they use the wrong, or not quite perfect verb, for Polish, or something like that. I've already kind of given up the idea that I'm gonna make a perfect game, so I'm okay with those things in some cases. I think the biggest lesson from this one is to involve Josue much sooner. As a solo developer, it's hard to know when you gotta involve other people in the stuff you're doing. So I have worked on this game for a real long time before even showing it to anybody and then before talking to localizers to see what kind of problems there would be. The direct lesson that I'm gonna take away is okay, I'll talk to Josue way sooner, basically. The more general lesson that I usually take from these things is don't do what I did last time, which was have a grammatically correct sentence building system in the game. That worked out okay, I had properly sizing text and things like that. So somewhat lesson learned there because when I was laying out those documents, I remembered back to my time on "Papers, Please" where, how hard it was to not consider that from the very beginning. But the sentence grammatical thing came out of nowhere because I hadn't considered really the complications with gendering and verbs and just grammatical structure in different languages. So next time, the specific lesson is if I want to build sentences in the different languages, heads up.