RogueForge

Talks

Ken Arnold: Rogue and the Seduction of Infinite Simplicity

Talk
October 2012
"Rogue and the Seduction of Infinite Simplicity"
Northeastern University Game Development Club, Boston · video, 53:47

Ken Arnold, who wrote the curses library and worked on Rogue with Michael Toy at Berkeley, talks to the Northeastern University Game Development Club in Boston: how Toy came looking for him with curses questions, Rogue's spread on the Berkeley Unix tapes and the ARPANET, the rules behind its random dungeons, starvation and fixed starting stats as answers to players who gamed the system, securing the score and save files (and how Ken Thompson got round it), why people played it for days on end, and the roguelikes that followed. In the questions he talks about Rog-O-Matic, the one game he won, and how the source of Rogue 5.3, which he demonstrates, survived only because someone stole a copy of a backup tape.

See also the 2016 panel with Glenn Wichman and Michael Toy.

About this transcript. Made from YouTube's automatic captions and lightly edited for readability: punctuation added, false starts, repetitions and filler words removed, and obvious mishearings corrected (names and titles, Wichman, Jon Lane, ARPANET, USENIX, inode, Rog-O-Matic, Ken Thompson, FTL, Dredmor, Geometry Wars). The wording is otherwise his, slips included; words in [brackets] were supplied, and [unclear] marks words the captions lost. The captions do not say who is speaking: the host introduces Arnold and closes the session, and in the questions members of the audience ask and Arnold answers, with a new paragraph, or a dash, where the speaker changes. The talk has no chapters on YouTube; the section headings were added for this page. Each timestamp links to that point in the video.

Introduction

0:08 Please, a round of applause for Ken. — Thank you for inviting me to your [unclear] pizza.

0:16 When Michael asked me to come here, it wasn't the most obvious thing in the world to me, because Rogue is a game that had its heyday in the 80s, before, I think, any of you were born. [From the audience: "I was born."] Well done, sir. But it actually continues to have an impact today, and I'm going to talk about that as we get along. But originally — well.

What Rogue is

0:49 Let me just start: what is Rogue? Has anybody here ever played Rogue? That's not bad. And roguelike games — NetHack, Moria? Okay, so there we are. I'm going to go through this intro, what it is, and I'm going to show it, and then we'll talk some. So Rogue lives in a D&D-style universe. You start at level one, you're weak, low level, and you build levels as you go.

1:14 Every level is a 3×3 grid of rooms; sometimes a room is missing, just to keep you on your toes. The goal is to fetch the Amulet of Yendor on level 26, and the odds of winning — that was an estimate — [were] few. In fact, Michael Toy, who was the original developer, primary developer — what he said was (I'm going to have to translate that into modern speech, because the scales aren't the same anymore) it would be like if every time somebody won a game, their name went up on the front page of Google. He said that shouldn't bother anybody, because the information flow should be so low. So he was figuring, in that kind of stretch, maybe one every day or two in the entire thing.

From Santa Cruz to Berkeley

2:02 So originally Michael and Glenn Wichman wrote this when they were at UC Santa Cruz, hanging out in the late — very late — 70s. It was very quickly very popular on the Unix systems of the time. Now, what was going on there at the time was the ARPANET, which is the precursor to the internet. The ARPANET was being developed, and Berkeley was being paid to develop BSD Unix as the primary new operating system for the internet. There were some other operating systems that already existed that were being plugged in, but the idea was that Unix could be made into a very good network operating system, or operating system for the network.

2:51 So all the Unix distributions that weren't coming from Bell Labs were coming from Berkeley. They were sending out computer mag tapes, and there were probably about 60 or 70 institutions that directly got those tapes, and then they made copies and sent out tapes. So when we say that this was popular on Unix systems, there were probably on the order of 1,500 to 2,000 computers at the time, and each computer was shared among multiple people, so you might say a computer had an average of between 10 and 30 users.

3:33 So Mike transferred to Berkeley in 1980, and he went looking for me to ask curses questions. (I'll show you Rogue in a minute, but it's built on top of a library I wrote called curses, which was designed for character-matrix terminals. It'll be easier to describe it when I point at it, and I'll do it fast; anyway, you probably know.) So he was using this library that I wrote, so he came looking for me; and I had been playing this for a while, and I had my frustrations with the game, so I went looking for him. So we sort of ran smack into each other, and then teamed up, and we did the primary development from Rogue 2.0 to 5.3, the last we put out. And then Jon Lane ported it to the IBM PC. This was well before Windows.

A game its author could enjoy

4:25 The main thing that Michael was trying to do when he started out was to write a game that he personally could enjoy playing after having written it. The reason this was particularly interesting to him is that the primary games at the time were Adventure and Zork and games of that ilk, which were essentially mazes. You run into the wizard, and there's a magical set of things you have to do to get past the wizard, and then you would come to the gulch where the dragon would attack. But they were the same, right? If you knew what to do, you could write a program that would just send the right text to the terminal at the right time, and you'd be done. And of course, once you've written one of those things, there was no point in playing it; you knew all the answers. And his idea, which shockingly was new at the time, was he wanted to write a game where he could have adventures but be able to play it himself, even though he'd written it. We used to call these rat-in-a-maze games, and frankly they bored the crap out of me too, so we were well matched.

5:27 So the idea is you have random generation with the basic rules, and you regenerate the universe each time, but the metaverse is always the same. There's a fundamental set of rules that govern play, govern the kinds of items that can be found, how they interact with each other. So the question is not "What are the rules of this universe?" — after you get playing, you learn the rules that apply to each universe — but "How do I apply the knowledge I have gained previously to this new world that I have in front of me, that I haven't seen before?"

6:00 It was really important not to reward mindless play. If people played Rogue, they would literally sit at the terminal and play for hours — I know this doesn't shock you at all — and this was also pretty new at the time. So what we didn't want to reward is, for example — there was a strategy where you could drop a scroll of scare monster and stand on it, so the monsters wouldn't hit you, and they'd gather around you, and you put your book on the space bar, and then you go get lunch. And then you come back, and by then you have a million monsters gathered around you, and you've fainted, but then you eat some food, so now you can fight again, and then you have a million monsters to kill off, and lo and behold, a lot of experience. That's when we introduced dying of hunger. Because we don't want to reward that kind of — we want people to be playing the game, not outsmarting it with clever little tricks they can play to get around the game, and people [were] doing that all the time. And as I said before, it should be hard to win. In fact, when you look at roguelike games, this is one of the really defining characteristics, I think: it's going to be hard to win, and we are going to keep screwing with you forever to make it happen.

7:17 So it's a turn-based game: all combat, everything, is turn-based. It's done on a terminal, and since a lot of you have played it, you know what it looks like. I'll show it in a second, for those of you who don't. You look down on the grid of rooms. The graphics are limited to printable characters — at that point, that was all you could rely upon. The window size: 24 lines, 80 characters, dead-ass standard. You are the @, monsters are A through Z, scrolls are question marks, rings are equals signs, armor is a bracket, and so on; there's a set of other symbolic characters that represent things you pick up in the dungeon. And it's a permadeath game.

7:58 And this is where — when you look at the roguelike games, there's controversy still over whether or not you are a roguelike game if you don't have permadeath. But it was important in the way we played it: you play and you play and you play, and then you screw up, and you just have to start over.

A demonstration

8:22 So let me just switch over and show. Now, I have to say I just spent a couple of hours today trying to get the actual original source of version 5.3 running, and it sort of works, but if it crashes, don't be shocked. It's a little hard to see, because my cursor is yellow, but there's an @ sign under here, and that's you, and the B, that's the bat, and you go up and attack the — that was exciting. Pick up gold, run through the corridor, get over here; that's some food, more gold. Let's see, slower here, so you can see movement. So now you see we have gone another — this is the middle room of the 3×3 grid. [unclear] Yay. And so on. That there is the down staircase, and just for amusement I'm going to see if it works, because I actually didn't try that. Yeah, there we go. So you can see the graphics are pretty limited. There were people who went to some trouble to use some of the fancier graphics characters, but the basic matrix thing still exists.

9:54 You can get Rogue for your phone; I believe it's available on both Android and iPhone. And it has two different aspects. One is — I know you can't all see this very easily — that looks very much like the one that's on the board; it just has colored dots for the floor instead of white dots. And then if you turn it this way, it actually puts up some tiles, so you're not an @ sign anymore; you're a picture of a person going through the dungeon, and walls of bricks and things like that. So you can play on your phone if you are masochistic, because it really is a keyboard-based game, and you try and play on —

The rules of the universe

10:44 So this is the metaverse, the rules of the underlying universe generation system. At the highest scale, each level has rooms, doors (which can be hidden, and you have to search for them), hallways and a staircase down. What's random is where exactly the rooms are placed within the grid, which rooms connect to which, which rooms are lit and which rooms are dark. On level three there's a certain set of monsters that you can potentially encounter, but their placement and their generation — wandering monsters that show up during play — is random.

11:23 There's a fixed set of magical items: there's a potion of speed, and there's a scroll of teleportation, and so on. They're labeled randomly each time you start the game, so when you pick something up the first time, you don't know what it is. You can discover it many ways: there's a scroll of identify; you can use it and see what it does — that's sometimes smart and sometimes very stupid, but in desperation you'll do anything. There are all sorts of ways to try and figure it out. Once you do that, the game remembers what label is on the scroll and will tell you "That's a scroll of identify" the next time you pick one up. But when you first get into the dungeon, you're a neophyte; every time you start, you're a neophyte, and you don't know what a scroll of identify actually looks like.

12:04 Sort of fun playing around with colors, because it's all words. For example, I used plaid as a color for the potion: a plaid potion. And again, armor and weapons are of known types, and what you find is random. And your starting stats are always the same. We initially did randomize people's starting stats, and — this is the "do not reward people who want to play in a boring way" — they just started rerolling over and over and over until they got good stats, and then proceeded. So we said no, can't do that.

Popularity and cheating

12:44 So when they were sending out BSD Unix, it was on the tape, and it went to everybody who had an ARPANET node. When we spoke at USENIX in '82, which was in Boston, Mike and I gave a talk on Rogue, and it was the most highly attended talk. USENIX is where the Unix people gathered and talked, and Rogue filled the seats. It's hard to talk about stats and how popular things were, but —

13:14 This kind of behavior was pretty new. This kind of behavior, of course, we all do, but this was the first time anybody ever really did this with a computer game, so we were stumped — absolutely jaw-dropped. People would talk to us about not doing work for three days on end while they played Rogue, and missing dates, those people who had girlfriends. And people would get together and do parallel play: they'd all sit in the same room and play Rogue. Since it's a single-player game, that means they're all just ignoring each other in parallel. And that was a new experience; [I'd] never seen anything else do that.

14:00 And of course, popularity led to cheating. There's a shared score file on each computer. Since the computers were shared, that meant — not everybody in your network, because the network software was pretty new then, but everybody who shared that machine with you would share a score file with you. And that meant, to us — Mike and I felt that we had some obligation to at least discourage people from screwing around with the score file and getting scores in ways that were not playing straight. Somebody who doesn't know how to hack a computer ought to be able to get the high score, right? That seems totally different. So we had to secure both the score file and the save file. We did a lot of checks on the save file: when we saved the file to disk, we would know the inode number (its identity on the disk), and the save timestamp, and all sorts of things, to try and make sure that nobody had played with it.

14:55 At one point we got to notice that Ken Thompson, who was one of the inventors of Unix, used to play with a disk: he would mount it, and then he would raw-copy the disk so that he could start it over and over and over again. Okay, I can't stop that, but most people don't have that access.

Why people played

15:10 So to us this was a really shocking thing. Why is this — what about this is popular? Why are people playing this instead of, I don't know — there was a solitaire game, there was Hunt the Wumpus, there was a 20 Questions game, there was Adventure, which people would play a lot until they solved it. So it was popular partly because these kinds of cursor-addressable terminals, where you could draw a picture on the window like that and then go back to a point and change it — that was brand spanking new. That's how long ago this was.

15:51 So partly it was using the latest graphics system. There were much better graphics systems, but they cost way too much money for most people to put their hands on. Starting play was very simple, so it was easy to get up and start down this path, but it was impossible to win, and that was, I think, important to its popularity. You get to the point where you absolutely frustrate the ever-loving crap out of people who are trying to play the game, and they are goddamn going to beat it next time; they keep coming back, going to beat it. So this was surprising to me, but the fact that it was so difficult to win was partly what kept people coming back. It was as immersive as was possible, again, for the time, because that's what graphics systems were. But it was direct, interactive: it didn't print out a bunch of text that you had to read and then, in your mind, create the picture. The picture was right in front of you.

16:39 Which led us to this brilliant understanding — and you all probably have that anyway — but immersivity is in your head. What constitutes being inside a space happens between the ears, not on the screen. It's a matter of expectations. [I remember] seeing a psychological experiment where they put a screen in front of your eyes, and then they touch a dummy while they're touching you, and your mind maps the two of them together. And so then when they hit the dummy in the stomach, you flinch, because you think you're about to be hit. It's all up here. It's a matter of what people expect to see; the level of graphics is all a matter of expectations.

17:18 To us, in some sense, this felt like how slot machines are addictive. That is, it's easy to put the first quarter in, and we give you a lot of little wins, because you beat this monster, and you beat that monster, and you figured out how to get through this doorway, and figured out the maze of this particular level. But you never get to the big win, so [you] just keep going; the big win is just over the hill.

17:42 And also there's constant change and motion. I don't know about you, but it's a fairly standard human experience: you may sit on the beach and watch the waves roll in, or look at a fire, and find that it's not very exciting, but nonetheless it's rather calming and attractive; most people [can] sit and do that for a long time. There's a certain rate of change that's seductive. And on the screen, as you saw, you move around, the room pops up, you go down a level, there's a whole new level, and yet it persists long enough, because you have to look around the room. The same is true, again, of slot machines: you pull the lever and it rolls, and it goes dink, dink, and there's a pause, and dink, dink, when you roll it over again. So there's this rate of change that isn't too fast but isn't too slow that I think pulls people along.

The games that followed

18:43 So it spawned — much to our original — you have no idea how weird this was to us. Nobody had played a game in this kind of way before, and people were asking us how it worked, why it was popular, and we didn't know the answers to this. We were actually programmers; we weren't even game designers — there was no such field. We weren't in the psych department. We had no idea why this was happening, and people were piling in and trying to pick this thing apart. And one of the things that surprised us a great deal was that people started writing their own games that were in the same style, in the category, as you all know, [that] still exist.

19:34 NetHack is the closest one to the original Rogue design. It still has dots for floors and dashes and vertical bars for walls, so it's a very character-matrix-oriented thing. It has some graphic tiles, for people who like graphic tiles, but primarily it's again a character-matrix-based game. And that's true, of course, of many of the early ones, because those are the displays people had: Larn, [unclear], Omega, which followed Moria.

19:58 Diablo is one of the few modern games that still fits in this. I haven't played Diablo III, which is the current release, and I'm told it's lost a good bit of this, but I haven't played it myself, so I make no claim. But at least in the earlier versions of the game, each time you went through, at least large portions of the universe you were interacting with were generated on the fly, so if you went back and started up again, you would have to figure it out again.

20:21 Does anybody know the Mystery Dungeon series in Japan? I would dearly love to talk to some of you guys about that, because actually, until today, when I was researching it, I had no idea how popular this was, but these things apparently sell millions of copies. So I would love to hear what that's about. And then FTL, which just came out in September, and I haven't had a chance to get very far into.

20:53 But I want you to think about the gap between these, which are again essentially character-matrix display games, and there was also a very long period of time — the last NetHack release was 2003, for example. So most of these over time faded away, and then there were just a few carrying on, a few things that picked up later. But FTL: Faster Than Light came out, and then there's the one I'm going to have to ask you to remind me of — [From the audience: "Binding of Isaac?"] Say again? ["Binding of Isaac."] Oh, yeah — which is very much like Rogue. I played that; the same kind of style of game.

21:37 So I don't know: is there anything major that's current that I left off this list that anybody knows about? [From the audience: "There was ADOM … he had a Kickstarter, and he's coming out with the second one."] Oh, okay, I'll look at that. Dredmor is relatively recent, but I don't know how popular that really is. [From the audience: "Dwarf Fortress."] What? ["Dwarf Fortress."] Well, there is a character-based — yeah. I'm afraid over time I've gotten real jobs and don't have time to put my hands on all this stuff, so I'm definitely behind, but I try and pick up a few.

What makes a roguelike

22:45 So the real — I thought I had a slide in here, but I haven't come to it, so I think I may have lost it — is the section on what's the core constituency of what a roguelike game is. A lot of it is this: the random generation is critical, but a lot of it is difficulty of winning, and how much grief you get for losing. So the permadeath question is something people argue about forever: is that a good feature or a bad thing? Me, I think it's critical. I think it really makes a difference if you feel like everything's on the line when you make an attack. But other people, of course, like to be able to — they don't like to go backwards.

23:26 So this struck me many years ago, which is that most of the first-person shooters and the RPGs that [people] play are all rat-in-a-maze things: once you've done it, you're done. And considering that Rogue came in early and was very popular, I was surprised that there weren't more games that took the lesson about randomization to heart. So once you play the game, you're done. You might replay it on hard or killer, or try and see if you can beat your old time, but at some point you're just [treading] the same ground. One of the reasons I was thinking this might be true is that, of course, once you're done with it, you have to buy another game, so this may be why people don't think it's a good thing. But the Mystery Dungeon games are very popular, and they sell, so I don't know whether that's it or not.

24:22 And then the sandbox games, like Minecraft, famous at the moment: they have randomization, but they don't have any of these other attributes of play. But at least you walk into a world, and it's different each time, even though it's based on the same set of rules. So it was surprising to me at the time that Rogue became popular, and it's surprising to me, considering it was popular, how little it was followed. So both ways, whatever.

Lessons

24:58 And so I know the lessons we pulled out of this — I pulled out of this, and this is not my field, so I want to hear from you people who have been exposed to this. The ones that are easy for me to pull out — [unclear] — excuse me, I seem to have lost the last slide there. So, first of all, graphics obviously can be very simple. As long as people can recognize themselves and place themselves in the game, they're okay. Another is that people are driven to small, consecutive building goals, and [you] really can turn their frustration into addiction. That's really what it boils down to, at least as we saw it: once you have become sufficiently frustrated, you are going to get it next time. You say to yourself, "I am going to beat it next time. I was close, and I'll do better," and people will iterate over and over and over again.

26:01 The other main thing — it's been interesting to me to see. FTL is new, and even though Diablo seems to have pulled back from it, there [are] other games. There seems to be at the moment — again, from my unprofessional and unimmersive opinion — some bubbling up of this style of game again; people are trying to make it. You look at, say, FTL: it came out of Kickstarter, and what that means is they don't have to please anybody else except the people who want it right then. They don't have this planned-obsolescence intrinsic benefit that the other companies get: if they do a planned-obsolescence system, the game is boring, you buy a new one. But all FTL has to do is make the people who want to buy it right now happy; that's their entire goal. And as you know, Kickstarter and other equivalent things make it easier and easier to fund things in that kind of way, that go direct to the consumer and please the consumer, rather than please some general business model that you can turn into a business plan that you can sell on the stock market. I think more of these are bubbling up. So I don't have any major, huge things out of this, but in the sense of — again, this wasn't my profession, to analyze this — there are some lessons I've brought from it. I'm interested in hearing from you people who have played them and liked them, any differences you might have with what I said about it. And I continue to be surprised that people who were born long after these games were popular are still interested in this and still pushing, so that makes me happy to be here.

27:39 Anyway, that is the end of my prepared remarks.

Questions: commands and auto-explore

27:48 Any questions, comments? — Yes. Can you speak at all about whether one of the [reasons] that you had 20 people congregating for your next session was the complexity inherent in the game, compared to many current games — like the keyboard interface being completely new at the time, and still not being used since — or anything else other than —

28:05 Well, I mentioned curses earlier. Curses is a generalized package that Rogue is built on top of: you say, "Put these characters on the screen; this is what I want the screen to look like," and it figures out how to put the right characters in the right place, and then, next time, which characters have to be changed and which ones are the same, so it does a minimal update. And Rogue is the first major curses-based game, I guess I'd say. So is that the —

28:38 I'm not sure — [I mean] the keyboard control. So much of it is based on memorization of what does what. Did that actually draw people together, or offer additional difficulty?

28:47 That's interesting, because of course, in the day in which we were writing this, we didn't have any choice. Nobody had menus; that was a completely different thing. But most games didn't have as many commands, partly because they didn't have as many things to do. So there was a complexity of the universe as a whole. People could say, "Oh, I ran into a troll, and I did this and this and this," and people would certainly hang out and have those conversations; that was part of the stuff that brought people together. And then there's just general commiseration, right? Parties where people are doing this together: you go "Oh!", and people go "Oh, what happened?", and then you tell your story, and that's all good stuff; everybody can see what's happening together there. The keyboard complexity, as I say — I think for the people of the time, it wouldn't have occurred to them that it would be any other way. So that may have been a factor, but it's one that I'm totally blind to, because the options which we —

30:00 Yeah. One of the roguelikes that I played, which is called Dungeon Crawl, has an O, which will automatically have your character explore the level until you find something interesting, like a chest or a monster, as opposed to [unclear].

30:18 That is very interesting.

30:26 Try to imagine: so I come on a level, I hit O, and then I go get a cup of coffee — well, actually, that doesn't take that long; [I] turn away. I don't have to go look around. It seems to me at that point I'm not the one doing the exploring, and personally I tend to like — I like walking through the room and seeing what's there; that's part of the interest to me. When I play Minecraft, I like to mine, because you keep going around, you turn the corner, and there's something new. But not everybody gets into it in that kind of way, of course.

Rog-O-Matic

30:50 It does remind me, however, of a program called Rog-O-Matic, which somebody at, I think it was Carnegie Mellon, wrote in version 4.7 or 4.8 of Rogue, as a master's thesis project, I think. It was a Rogue-playing AI: they had a program that would play Rogue, and you could just watch it. Which I thought was great; that was actually probably the most fun I had since I had released Rogue, just watching this thing play. And I made it my job that with every release I would break Rog-O-Matic: do something that their system would just totally keel over for, and [they] had to whip around. Hallucination potion was my favorite one of those things. Hallucination potion: you drink it, and then every symbol on the map is random — not the floor, because that's just way too much, but everything else. Is that a piece of armor, or is it food? Who knows? And unfortunately they dealt with that by treating it like blindness: they would go in a corner and just wait for [it to wear off]. [I] thought that was cheating. But anyway, it's fascinating to watch. At a certain point — I think in the case [where] I hit O and then went on, that's interesting, but after a while I think I'd [say], "No, I want to do that part." But maybe for some other people it would make the difference between a boring game and not a boring game.

32:16 Yeah. I was wondering, how well did that AI do?

32:24 It was pretty good. I would say it's probably an above-average player — at least, not to the best of my knowledge; I should say I don't remember ever hearing about it winning, and I ran it a bunch. But really, seriously, I won once. It was designed to be very hard to win. So even if I'd been an excellent player, which I may not have been — anyway, I knew too much. At some point, if you know the game, you play by the rules it's supposed to [have], [not] sneaking around the side. Yeah.

Story and the roguelike

32:58 I know a lot of modern games are sort of story-driven, and that's why they're pretty much play once, and you know all the twists and all that stuff. I was wondering if you saw a place in more modern games for a story game that also had that randomly branching or randomly generated component, and if you may have already seen games that do this sort of thing.

33:20 Well, the narrative part of a lot of current games is, I think, its own interesting art form, and I don't want anybody to think I don't appreciate these other games in any way. But the roguelike universe is more like: you are in the game; you are part of the narrative. Instead of acting out the narrative, you create the narrative, in a roguelike game. It would be interesting. I could imagine running around the Final Fantasy world, but a lot of what those games have — I'm doing this on the fly in my head — a lot of what makes those games interesting are these little chunks of narrative, and the actual in-game combat is interesting, but it's not engaging in the same kind of way. So it would be great if somebody did that; it would be really nice. If you like a game and then you can go actually immerse yourself in it as one of the characters, to the extent that you weren't following a script anymore — that I can imagine being interesting. But I haven't seen anybody do that. To the best [of my knowledge] I haven't seen anybody do it; I'm not saying nobody else has.

34:38 How much do you think the popularity of Rogue was because of the ability for players to create their own stories? After a player finishes a game, they can tell their friends about this awesome experience they had, whereas when you're playing an adventure game, it's kind of like, "Yeah, that happened," or "Oh, spoiler." — Right, exactly. Yeah, if you're playing one of these standard systems where the narrative [is fixed], you can't tell your friends what you did unless they've already done it, because it's a spoiler.

35:09 In two ways, the answer, I think, is [yes]. There's the fact that when I have an interaction — because of the shared universe that I can't spoil, [I can] talk to other people; that makes it social in a way that the other games can't be. But also, while you're playing, you can be building this narrative in your own head. You're going into the room; you feel like you're the actor within that story, in a way that you couldn't if you knew that the script had been written by somebody else and you were just walking through the role. So I think both are true.

Winning

35:40 How much do you remember about the first time you beat — the only time you beat [it]? — The only time I beat it. Yeah, that was sort of an interesting moment, I suppose. I remember the office I was sitting in at the time, where I was being paid to develop Unix. One of the things about Rogue, by the way, is that when you get to the end and you win, there isn't some really big thing that happens, because it's hard to test. So you just win, and it says "Yay!", and it puts [you in] the score file. And yeah, I went out and I told people; I went out and chased people down and made sure that they knew. They weren't impressed, because I wrote it. But luckily it was a game we could play even when we were the authors, so we did enjoy it. Yeah.

Kickstarter and FTL

36:39 So for those of you that don't know, FTL on Kickstarter raised a little over $200,000. Does this surprise you, or were you like, "Duh, obviously"? — Well, I wouldn't say it surprised me. — This was in the 80s when you first released it, so are you surprised to see that people still yearn for this kind of experience? Or did you figure that the players would — like, point-and-click adventure games are having a bit of a renaissance from Kickstarter as well. Do you see it as just a cyclical thing that came back, or were you surprised at all when you heard about [it]?

37:17 I've been surprised by the whole thing. From the beginning, I thought people would enjoy it; I didn't think people would become addicted. And then when people stopped playing: what happened there? Why do people stop being addicted? What stopped being intriguing about this kind of thing? I actually didn't watch the FTL thing on Kickstarter; I wasn't paying attention to Kickstarter very much at that point — I had this job. But I guess my reaction when I found out was: it was interesting that people still want that kind of thing. But if they hadn't, I wouldn't have been surprised either. Of course, a lot of what happens in a Kickstarter is how good your pitch is, and that's all down to those guys; it's not anything to do with whether it's a roguelike or not. Did they do a good pitch? They obviously did a good pitch. Of course, they asked [people] to back something that people are really interested in. So even in retrospect I wasn't surprised. It was interesting. Yeah.

38:28 Did you find people's motivation to play dropped off after they beat it the first time? You talked about how it was those little victories that you hope to get —

38:37 No, not at all. There were people who won, and a few people — yeah, some people did drift off, but people drifted off without winning too. And then other people went in and said, "I'm going to be the first person to [unclear] points," or whatever; they just liked playing the game by that point. So I didn't notice that people's behavior changed a lot after they won. But that was in a big universe, and most of the people who won I never even knew. They'd mail us the score files or something, [and] we'd say, "That's cool." Yeah.

Unlocks and character classes

39:12 One of the mechanics of Binding of Isaac is that as you play through, if you meet certain requirements, you'll unlock new items, and so, in a way, even though each step is permanent, you're still gaining power as you go. What do you think of that element of —

39:27 Yeah, we were actually just talking about that in the car on the way over. I didn't play it long enough to know how much those things are different ways of being in the game, versus capabilities that make playing the game easier, and you can tell me which of those it is. For example, NetHack has different character classes, and obviously if you play as a librarian, you're going to have a different experience than if you play as a warrior. But it isn't to make it easy in either case; some things are tradeoffs. So it's more a different way of experiencing the same system, rather than "Oh, now you've reached level 33, and you can now use this extra special ability that you couldn't before, so now it's easier to win." In Binding of Isaac, does it have both of those kinds of things, or either? — I think that most of the items are just tradeoffs, just more things that you can get. I think a few of them are just better, but that might just be designed intentionally, like — oh, this is it, right: you don't unlock the BFG, but it turns out that playing as a warrior is a little easier than playing as a librarian. — Which it does, yeah. Okay. In general, that kind of character classing I think is great. I just wouldn't want this kind of "You've gotten this many experience points; you get double [unclear]." Oh, I'm sorry, I seem to be looking that way. I apologize.

Story, again, and MMOs

41:05 Do you feel that the lack of a story in Rogue actually helped make it a more addictive game? My thinking behind that is: today's games, you play because you have the story, gameplay or whatever. So you beat it once; you play it again, you already know the story, so you lose that much of a reason to play. But with this game it's only gameplay, right? — It's like — yeah, there's like a two-paragraph preamble in the instructions. — Yeah. Because it's only gameplay, that might motivate people to just keep going.

41:47 Well, I think that certainly part of the problem with the RPGs, where you follow the narrative through, is [that] you're done, right? Reading the same book twice — I've done it in my time, sometimes, but it's much less fun. So in some sense, though, that's bound up with the fact that it's randomly generated each time; it can't be the same story each time. Now, NetHack has — and I think several of these things have — a particular special level. There's a level of hell, and there's a level with the castle, where you go off on these other sides, and they're more fixed and less random, much less random; they're in some sense side games, or mini-games. One of them, in fact, is the Rogue level: you go down a level, and it says everything looks really old, and they have it look exactly [like], play exactly like, Rogue. So those things are — we're having fun, [and we] have to share with everyone. So in some sense I think it falls out of the fact that every game is different, and you can't have that narrative, for good or ill. So whether it's helpful or not helpful is almost beside the point; it's required. And it may be something some people like better than others. Obviously, if the narrative is put out for you, you can do a lot of rich narrative things, because you define them — sorry, the author can do a lot of rich narrative things — whereas if you're walking through the game yourself, how complicated can I make each interaction without making it obviously the same, something you pick out and play again, over and over again? So for me — I don't know how it was; it was nice, and it was different every time, and that story wasn't — I'm not sure how to compare it. Yes?

43:41 It's kind of like — a lot of elements [are] so popular, but obviously there are not as many roguelike games now. But I think MMOs have a lasting effect, and last a long [time], because of that feedback system and [unclear], and that's why people will play for so long, unlike any other game, I think. So I think it's nice to see that game.

44:06 Yeah, it's interesting, though: the MMOs I play — well, I haven't played a lot of them — [have] a lot of random chance. There's a lot of random chance, and then, yeah, what you encounter: the monsters you encounter or challenges you encounter are pulled out of a bag, because they can't be preset in any serious way, for the most part. Yeah, in that last slide, I think, I had a little mention of MMORPGs, because I don't quite know how to fit them into this universe. They're both random and not random, and some of it [is] sandboxing, like the sandbox games — Second Life or something, if anybody still looks at Second Life. But I don't quite know how that fits into the universe. The real answer to your question: I think that's a very interesting question, and I don't have an answer for it. It was interesting to hear your thoughts, but I can't give you back much more about mine.

45:25 Just a general comment on NetHack: I think it's as successful as it is because the in-game story just doesn't really make a difference; it's the stories that the [unclear] conflict, and [the] very high cost of death compared to other [games].

45:41 Yeah, well, that makes sense. But the high cost of death, I think — not everybody's willing to risk that, and people just can't lose that much. But then they play some of the game; amazingly enough, everybody [plays], pretty much. Yeah.

Money and the lost source

45:57 Since Rogue was included with Unix, did you make any profit off of it?

46:05 No. To give you some idea how old this was, this was before the notion of open source. Berkeley Unix actually had a lot to do with defining the open source license, not that they called it that at the time. But we tried to hold it back so that we could make some money. And in the end Jon did port it to the IBM PC, and he barely did better than breaking even, so he didn't have any profits to share, and that was really the only attempt. But unfortunately, it's a decision I regret a lot, because right now — I had this copy I was just showing you, 5.3 Rogue, and the only reason I have it is because somebody stole a copy of the backup tape for the system I was working on. And if they hadn't done that, and hadn't gotten it to me in time, before the tape deteriorated, I wouldn't even have that, because the source wasn't open; it wasn't widespread. Whereas if I want the source to any other Unix source at the time, they were pretty much all open-sourced, and so they're all out there somewhere. And I wish I'd made a different decision on that one, I'll tell you, because in the end we never got any money off [it].

A patent letter

47:33 Occasionally I got a dinner. [One time] I got paid for writing a letter in a patent fight. Somebody was trying to patent the idea that — this is just to show you how crazy patents are — you have the screen, and you want to move one space to the right without destroying what's there. And they had this brilliant idea, and patented it, that instead of using some escape sequence the terminal defines, you could just print out the character that's currently there again. So the dot becomes a dot again, and [you] move one space to the right. Isn't that cool? They patented that, and that is something that Rogue and curses did a long time ago, and [I] could write them a letter saying, "This guy's full of crap." It's so nice to help destroy a patent. [It may] have been invalidated; I don't know what happened in the end. The guys who were being challenged paid me, and then I didn't follow the case. Usually those things are settled out of court in the end. One side — let's say the people who are trying to enforce the patent — got this letter from them saying, "Look, we can prove to you that you're invalid," and they go, "Okay, never mind," and they walk away, because if they lose the literal case, then they can't write a nasty letter to somebody else who isn't smart enough to ask me for a letter, and so they get some income from the patent. Or, the other way around, if they know they're going to lose it, they would rather just give you some money now than fight it out. So usually those things are settled. [unclear] Other question?

49:07 Who's the most famous person you know who has played Rogue and liked it?

49:07 Bill Joy, off the top of my head. I can't think of — I mentioned Ken Thompson; that's pretty good for the day. Yeah.

Limited saves

49:23 I know that you feel the idea of permadeath is very crucial to this game. Would you feel that the game would become too different, and ruin the original [idea] of it, if you included even a very limited save system, where you could only save twice a day? Would that ruin it too much for you?

49:48 So that's interesting. If you could save twice, you're certainly going to wait until you have a pretty important state; you're not going to do it early. But everything's random, so you never really know — yeah, you never really know. But you can't just keep saving, and then try something, and save, and try something; that would totally wreck it. That's an interesting question. It probably wouldn't — as long as it was pretty limited, it probably wouldn't make that big a difference. I know how frustrating it is [to lose] a great game near the end, trust me. But something about it is [unclear]. But something like that might work. I could see giving people a [save], limiting it so that they have to take the risk: they save the second time, and then they can't do it again; or they have to [use] a save, and then they risk dying. Now that might be close enough. Yeah.

50:46 I think something similar to that is Geometry Wars, from the Xbox. It's basically like an arcade game: you fly around the ship and you shoot shapes, and they try to get to you, and it's permanent death. But the way that they did a kind of second chance is you have three bombs, which will clear the screen, so you can use them whenever you want, but you usually don't get more. So, a similar kind of thing: if you get in a tough spot — so instead of an extra life, you get an extra death. — Yeah: wipe everybody out.

51:28 Yeah, I'm not sure that in this particular case that would be so good: [if] you get in a tough situation, you don't have to fight your way out of it. I'm not sure that fits Rogue. But again, the fact that it's limited, and you have to think about whether this is the time where you need to save, fits in that same place. Yeah.

What he plays now

51:50 Do you still play [Rogue]?

51:50 No. I have moments of getting re-engaged, addicted to NetHack, but Rogue itself — the 26 monsters really don't do it for me anymore, I guess. And I don't play NetHack much anymore either; there are just so many different things to do. So, one more.

52:08 What have you been playing lately? — What have I been playing lately?

52:17 I suppose the thing I have been spending the most time on lately is probably Minecraft, building things. I mentioned mining; I find that interesting, and I get frustrated with myself, because I'm spending this much time just digging into an imaginary wall. But nonetheless I come back. So [I've] been playing that, probably more than anything else. [I] try out other things sometimes, but I do have this job thing going on, and this family thing going on; it does tend to eat into the time. So, yeah: don't get a job; don't have a family.

52:51 They had a major update to Minecraft. — Yeah, they just released it, but I haven't seen it. — It just came out today. — What? Yeah. Which is — he's my dealer on this thing, so he tells me what's in. Okay.

53:17 Ken, can you hang out a little? — Sure I can. — Yeah, so if you guys want to — I guess the meeting's over. Let's have applause for [Ken].