Emily Short -- Down with Parsers!

Authoring General Design Discussions
21

Well, OK. I guess I don’t care about getting rid of flailing, because flailing in games is commonplace: the inevitable and unavoidable consequence of requiring skill that the player may not possess.

I play 3D shooters like Modern Warfare 2 and flail around, firing at bushes and trees while unseen snipers gun me down. I play RTS games like Starcraft 2 and flail around, trying to decide what to build while fending off a zergling rush. Same deal with World of Warcraft raids, where the content is designed to be unfair and unforgiving until your group invests a large amount of collective flailing.

I don’t see why the parser should function like an auto-aim bot in a FPS, or a build order wizard in an RTS.

Where IF differs is that once I start flailing, I am far more likely to quit forever than to doggedly stick with it. I think that has at least as much to do with the single-player nature of the games, though. Flailing is more fun in a group setting: if I spend five minutes uselessly firing into trees and drawing sniper fire, and my brother jumps out and guns down the enemy with a single precise shot, I feel as though we have both accomplished something meaningful within the game.

For me this is the charm of the annual IF Comp: the sense that other people are playing the same games at approximately the same time. That we are all flailing on the games together. And if I begin to waver in my commitment to a particular game, I can glance at reviews to see if people think the end justifies the pain.

The core failing of the modern IF parser is not in game mechanic terms but in technical ones: it presupposes a keyboard that is pleasant to use. There are many potentially great IF platforms - Android devices, iPads, iPhones, etc - that are downright horrible to type on.

Some way to identify significant nouns and auto-complete appropriate verbs would go a long way toward alleviating this, but it is not the only way. Voice input is another possibility; my father-in-law uses that for Google searches on his iPhone. The results are often hilariously inaccurate but he prefers it, only switching to keyboard input after repeated parser failures.

Overcoming this failure with assistive interfaces would be a fantastic step forward, but I would rather see it done in a way that retains the need for the player to compose a coherent command as the dominant interaction model.

22

One of the things (in my opinion) that would make for more compelling stories for a broader audience would be stronger character content: more interactions involving people, fewer involving locks and keys. And that is something I think a different interface would directly help with, because it would make it less necessary to rely on the same conventional verb set.

I don’t think anyone is arguing that the general public is too dumb overall to learn to play IF. I do think that, at the moment, the parser offers enough of a challenge that you need one hell of a hook to get novices to regard learning it as a good use of their entertainment time – and, in particular, because it does a poor job of communicating what sorts of things the player is going to be doing in the game, it makes it very hard to offer the player a gameplay hook.

I had trouble with that same race in Assassin’s Creed 2 (though not quite two hours’ worth) – but I did have a pretty clear idea almost from the beginning of what kind of game experience was being offered, and that was part of what hooked me. It wasn’t the story intro, which is frankly terrible. But “game where you parkour around a surprisingly accurate version of Renaissance Florence” was something I was willing to invest some effort to play. What’s more, I was always clear on what exactly I was supposed to be doing with Assassin’s Creed 2; I was just having trouble executing the plan of “be faster”. An IF prompt doesn’t offer nearly that much information: it’s easy to get in a situation in IF where you have no idea even generally what you’re supposed to be trying to accomplish, what sort of input is desired, and what you should attempt next in order to attempt to break the log jam. In the AC2 case, you have a process you can iterate with some expectation that it will eventually lead to success. Not being sure even of that much makes it even harder to justify the time.

Anyway, that was partly a tangent: my main point was, IF in its current format doesn’t necessarily give the player much of an idea of what the game is going to be like to play.

So if we’re stuck for communicating what the gameplay experience is going to be like, we’re left with setting and story as ways to get people invested a given game. Not impossible – Peter Nepstad has, to the best of my knowledge, done a better job than anyone else in actually selling IF games recently, and I’d guess it’s because he picked a strong setting that was of particular interest to a small but active community, then devoted himself to rendering that setting accurately and marketing the results to the relevant group. But this is not easy, especially when working with a setting that isn’t a historical reproduction with an existing core of followers.

If you don’t have something external to show to players, then you have about two paragraphs of text in which do the job – not only to get them to keep reading into the third paragraph, which is what the first two paragraphs of a book in the bookstore have to do – but get them to persevere at the task of learning HOW to keep reading.

So I do think there’s a lot to be gained by certain kinds of marketing effort: showcasing our games better, giving them art and a website that sets the tone and builds interest, getting reviews and articles published in more mainstream places. All that is important if we want to get people to want to play, and we want to give them more to go on than they’ll get by reading the intro text. I just don’t think it’s enough.

23

I read your post at work, then came home and found Emily has said what I was going to say above, but just to be clear: this is not my opinion. People - even people who have never played IF - often have an idea what commands they can type. They know the syntax, they know some of the standard verbs, they can see the nouns in the text. Their complaint is not “What are valid commands to type?” but “Given the enormous number of potentially valid commands, how the hell do I figure out what I’m supposed to be doing?”

The verbs in Metal Gear Solid are: run, crawl, wrestle, shoot, punch. But the game also makes it clear through the story, through level design and through the interface, that the game is about avoiding combat. (And before you mention the way a parser allows unexpected interactions, MGS also features sections where you have to do something completely different - a direct firefight, resisting torture, abseiling down a building, figuring out how to defeat an opponent who can read your mind…)

24

This. I started the below comment and never posted it, but to pile in:

The learning curve for games like Assassin’s Creed is a lot different from the IF learning curve, at least what Emily’s talking about. In reflexy video games, you usually start being able to do something, and even when you fail, you can see why you did and how close you came to not failing. I’m currently stuck at the final bosses for Cave Story (medium ending), but I can at least do something in this fight, and before that I had lots of running around and jumping and stuff before the difficulty started to kick my ass. Which gave me the incentive to keep playing, so that I was eventually able to learn how not to get my ass kicked on some of the earlier fights. – In fact, what almost made me quit at the beginning was a sort of IFfy moment; before any of the really arcadey stuff happens, you have to push the plot along by finding something and giving it to someone, and I had no idea what to do. Actually, what really almost made me quit was that on the loading screen I couldn’t figure out that you had to press “Z” to start.

Anyway, in IF, if you start from scratch, you don’t know what to do at all. It’s not that you eventually get your butt kicked by the Babelfish puzzle, it’s that (to take an example from Brian Moriarty’s talk) you have someone type “Where am I” and give up when they don’t get an answer. Not to mention the games I’ve stalled out on because I forgot about “search.”

Another thing – going back to reading and video games – is that there’s a huge audience that’s spent a lot of time practicing video games, and a lot of money that’s gone into helping them practice. Even more so for reading. IF doesn’t have that luxury, so saying “Learning IF is no harder to learn than learning platformers or reading” is neither here nor there. IF doesn’t start with those games’ built-in advantages.

25

I take your point, but you only need that kind of hook once per player: to get them to invest the time and energy overcoming the shallow learning curve in pursuit of content that they want. At that point, they are initiated into the ways and perfectly capable of engaging other IF works, which need not be compelling to the same degree.

FPS players are not born knowing how to mouse look and keyboard strafe. It may take a game like Half Life or Counterstrike to force them to acquire that skill, but once acquired it serves them well in any FPS game available. Most of which will not be very good, and few of which could compel the uninitiated potential player to bother learning how to play.

That’s OK; not all games have to be accessible and amazing, so long as some few are.

:blush: I eventually asked my brother how in the world he had beaten that part. He had no idea what I meant and had to struggle to recall it.

It turns out you can just completely ignore the other guy and run straight to the church and climb up, instead of chasing him up ladders and across rooftops.

Really? I felt like the tomb sequences were fairly opaque. Like the one in the church where the only goal you are given is to climb to the top. There are random, unmarked checkpoints along the way, but there is nothing in the way of specific guidance.

I don’t see this as a parser issue so much as a game design issue: how and when is information presented to the player? It would be weird in AC2 if the controller tried to guide you toward the objective by vibrating when you deviated from the path. The parser is the IF player’s controller; why should it be responsible for telling the player what to do?

I understand that there are aesthetic objections to cluttering game prose with cardinal directions, or out-of-world references to verbs. But there are also aesthetic problems with popping up a flashing red RT trigger icon during Mass Effect 2 cutscenes, to alert the player to the possibility of a special action, and commercial video games don’t shy away from that behavior.

It may not be artful but it is practical, because the Xbox controller is rather complicated. It has a lot of buttons and they do different things in different games. The player accepts the complexity because it offers greater potential for fun interactions, not because they enjoy discovering new things for the blue X button to do.

Well, that doesn’t have to be the case. More windows and more graphics can help with that. Alabaster uses a changing graphic of Snow White to draw the player’s attention to her and to highlight her emotional state. That seems to be a good way of focusing the player’s efforts. If you can get them thinking, “She looks angry - I should try to calm her down” then you have nudged them away from the standard IF verbs into a more conversational mindset.

Erik Temple sent me a demo with a very nice graphical map, with a blinking light marking the player’s position in the city. In a game about exploration, that suggests to the player that if they’re not sure what to do, they can always wander to a different location. Add a second blinking light to the map, and you allow the player to visualize a destination and by implication a goal.

Both of these are ways to suggest a game experience in ways that mainstream gamers will readily grasp.

Absolutely. But none of that hinges on a better parser, and I don’t think a better parser necessarily serves that goal.

26

My son can pick up an Xbox controller and input random commands. I am confident that none of the games I play would reward this interaction such that he would continue to do it for more than a few minutes.

In your example, typing “where am I” is much like hitting buttons at random. It is not a command that anyone experienced with the activity would bother trying, except as a way to abuse the simulation. Catering to this sort of abuse is counterproductive.

Do people put searches like that into Google? Undoubtedly. Does Google cater to them? Sort of. But Google has an advantage in that it can tell how popular a search is, and it can roll out a targeted service if it deems it worthwhile.

It would be more productive to adopt this practice than to meddle with the parser. Give authors a way to track all commands that players enter, and some way to capture whether or not the game responded appropriately. Then give them a way to add new responses to those commands in a seamless way, without disrupting the progress of current players.

This doesn’t help players that hit the wall before the update, any more than Google’s new mobile locator service helps the hapless person that queried “where am I” five years ago.

Yes, but so? It doesn’t matter that there’s a giant audience with FPS skills, because IF does not use FPS skills. If you had an illiterate child that excelled at playing the drums, would you entice him to read by making it more like playing the drums?

Even if you were in a village in rural Africa where no one could read and everyone could play the drums, that still does not mean that the best way to introduce reading is to turn it into drumming, or suggest that you should borrow concepts from drumming to make it more palatable.

All I am saying is that IF is a distinct activity with its own learning curve. Like any activity, it succeeds or fails on its own merits. The existence of other, more popular activities is at best irrelevant and at worst a distraction, even if they appear to the outside observer to occupy the same conceptual space. People learn new activities all the time. Learning is not the problem. Motivation is the problem.

27

On another thread a discussion arose about the “use” verb. Most people think that “use” makes the game too easy. But while following this discussion it occurred to me that “use” might be one of the tools to make the parser more accessible. Especially since I see other games mentioned that pretty much provide “use” as the only interaction. In FPS games (or most RPGs, even), you press the “use” button to open doors, take items, flip switches, etc. There are no buttons for “open”, “close” and “take”.

Could a game that only needs a few general-purpose verbs (say, “use”, “take”, “talk” and “go”) to be playable (the rest of the verbs would be optional) be the “Beginner Friendly” piece of IF this discussion is about?

28

Again, I see this as a game design problem, not a controller problem: the character and the setting should channel the player toward a specific set of actions. If they don’t, more explicit suggestions in the form of UI elements or text in the game window can prod them forward.

If the player sees a painting in the room, the game is responsible for communicating the appropriate set of actions. Characterizing the player as a master art thief suggests one approach: steal it. Casting the player as an investigator searching for clues suggests another: examine it.

You could even add a status line that read “Paintings: 0/1”. Tacky and heavy-handed, to be sure, but a player that cannot be bothered to come up with at least one valid input under such circumstances is not a player worth having. At least, I’m not interested in writing IF for players like that. It’d be like playing an RPG with someone who refuses to talk or roll dice. What’s the point?

I am more sympathetic to Emily’s argument, since the space of the possible interpersonal actions is so much greater than a few noun / verb pairs. But you seem to be making a more general argument against the IF parser, and I don’t really accept it.

29

“More explicit suggestions in the form of UI elements” is pretty much exactly what I’m advocating.

30

I don’t see it as a problem for either, but I do think that changes to either the design or the method of control (or both) could be the key to changing this from a challenging element of a game, into a transparent element - depending on where in the game you want the challenge to lie.

I’m assuming we’re not simply talking about using the verb ‘steal’ here. Maybe alarms have to be disabled, so it’s a game about meddling with gadgets? Or it’s an Ocean’s Eleven style heist, and we have to deceive the NPCs? We’re a cat burglar and the aim is to move unseen? Or it’s an armed robbery and I have to defeat the guards in combat?

In each of those cases I could come up with standard or semi-standard verbs to use, but I think it would very difficult to distinguish between them purely based on the in-game text. The game’s responsibility to communicate the appropriate actions becomes a rather direct one, I think (unless, as above, you want figuring out the action set to be part of the challenge), and why shouldn’t that be a part of the UI rather than something in the About text?

Funnily enough… From an early prototype of Walker & Silhouette:

I think the parser is fine. I think games with parsers are fine, and so are the people who want to write and play such games. But there’s a very stark line between parser IF and more static CYOA on the one hand, and completely graphical games on the other. I think this should be a fuzzy line all round.

To be absolutely clear: my stance is not a stance against the parser, but in favour of additional modes of all-text game.

I find it interesting that both you and RealNC seem to feel that IF must have a parser to be IF and have all its benefits, but you don’t blink at incorporating graphics into the game at a fundamental level (RealNC suggested even making them interactive). To me, it’s this that’s moving towards a different genre.

31

I’d say W&S is “less” of a text adventure than Gateway because of the keywords vs traditional parser approach. (This isn’t meant as criticism; W&S was awesome.) Putting graphics in text adventures was a very popular practice in order to attract new players. Graphics doesn’t mean a move away from IF; it means enhancing it (it shouldn’t replace the textual description but rather complement it.) I believe that whatever the new “IF for the Masses” would end up to be, media would play an important role. Even for me - and I consider myself “text adventure hardcore” - some text games tend to get boring because there’s no artwork and no music (to this day, I still have to find a piece of IF that that I enjoyed as much as Gateway.) I can imagine J. Random Gamer getting bored much faster than me. Making IF popular would, IMO, involve both accessibility as well as presentation.

32

Ah. I tried your Jade demo, but when I typed something the prompt just disappeared. Based on that, I thought you were trying to dispense with the command prompt, and by extension the parser behind the command prompt.

After resizing my game window so that it actually shows everything, I am much less confused.

The references to parsers seem misleading, though. If I understand correctly, you are arguing for a better prompt, one that provides more explicit suggestions.

I like that idea.

33

It seems to me that you don’t even really need the verbs. The direct objects are pretty much exclusive – “use” applies to objects that are fixed in place or in your inventory, “take” applies to portable objects that aren’t in your inventory, “talk” applies to people, “go” applies to directions and rooms. So you’ll have a keyword-based game. In other words, you’ll have Walker and Silhouette.

Now, this will limit the puzzles you can do. Sometimes you don’t want it to be absolutely obvious how you use this thing you just picked up, or you want to make sure the player uses the thing on the correct other thing. But that’s part of making it a beginner’s game, I guess.

(Now, in FPSes, I’m guessing that “use this” is mostly flavor in between and around the real gameplay. There’s a nice bit in Cave Story that uses some “go here and press the down arrow” play as pacing before a big boss fight. In IF using things would be pretty much all the gameplay. But point-and-clicks effectively have “use” as the only thing you can do with fixed objects and objects in your inventory – they basically only have “use,” “take,” “go,” “examine,” and sometimes “combine” – and that is all of their gameplay. So that could be another model for “use”-based IF.)

34

How do we reconcile that with “I prefer not to have the window of text forced into a small corner of the screen while the rest of the territory is taken up with interaction helps.” I agree with both, but…

(I tried Jade, but is there a particular interpreter I should use for it? I found it bewildering, but used it on a fairly old version of Windows Gargoyle.)

On the thread as a whole: the parser should out-of-the-box recognize when the player is asking a question and instruct him to phrase all his questions as commands: “READ SIGN” not “WHAT DOES THE SIGN SAY?” Even if the programming is no more complicated than the equivalent of a one-liner – Understand “who/where/what/are/is/can/should…<etc.>…” as a mistake… – it alone would go a long way to improving new player experience. It is better than a menu which has to be explicitly consulted because as soon as the player’s attention breaks from the prose to reach for a UI widget, the CLOSE PROGRAM button is just as accessible as the HELP button.

35

I was actually thinking of something like USE X WITH Y. Well, now that I think about it, the only thing this does is emulate point&click adventures with text, so meh :stuck_out_tongue:

36

Make the UI elements small and textual, and put them with the prompt. :slight_smile:

I don’t know of a reason that it shouldn’t run on that, but several people have mentioned that their windows were sized small enough that they didn’t get the lower pane clearly, in which case it wouldn’t make sense. So maybe try resizing the window to be larger?

I may try rereleasing it in a less-easy-to-screw-up form when I have a little time.

37

(This is not a reply to anyone in particular.)

In this thread “Gateway” has been mentioned a couple of times. I suppose some people here have never played it and therefore don’t know what it is all about. So here goes:

youtube.com/watch?v=zXgh2dQf3bM

Note in particular how well cut-scenes were presented (well, for 1993 standards anyway.) “IF for the General Public” could draw some ideas from that one game alone, but with a more modern approach, like less clutter in the GUI (endless lists of verbs aren’t that helpful) and making sure the presentation keeps players interested. All that combined with a more helpful parser prompt could “bring IF back”. Of course I’m aware that something like that isn’t something a single individual can attempt at his spare time; it would either have to be several people collaborating or a commercial entity doing it (/me looks at Textfyre…)

38

Bah, when I was talking about point-and-click games I somehow forgot that they’re usually about USE X WITH Y rather than just USE X. Anyway, I like point-and-click games, but I worry that some of the innovations Emily is suggesting could push us to basically emulating point-and-click games with text. (Even the social verbs – as I said at Emily’s blog, Raising the Flag on Mount Yo Momma implements the social verb “INSULT,” and it could easily be done as one of those point-and-click where you get three verbs when you click on something – click on a person and you’d get “examine,” “talk,” and “insult”.)

39

…come to think of it, this is why in the past I’ve occasionally given up on xbox games after a few seconds. Braid had those nice things showing you which button to push where. Flower was completely incomprehensible until someone told me what to do. I don’t actually have a console, as you may have gathered.

Which means that I was thinking of Flash-based games, and in a lot of those I can pretty much start from scratch with the arrow keys, and if there’s some other command I do expect them to let me know what it is.

Well, it’s actually Brian Moriarty’s example, which went like this:

“I remember going to a trade show to exhibit Beyond Zork back in fall of 1987. The story was always the same. Some curious retail executive would sit down in front of a computer loaded with BZ. I would explain that you typed ‘real English sentences’ to interact. They would nod, turn to the keyboard and type WHERE AM I? End of demo.”

And, as Emily pointed out in this discussion, the moral is not to say that you type “real English sentences” to interact. I mean, as recounted that’s basically a demonstration that it hurts when I do this.

(Here’s where I switch sides on the argument for a little bit.)

So, shouldn’t we really try a better explanation of what you do? Say that you can go directions, examine things, pick them up, and if you can think of some other stuff you might want to do it’s worth trying that verb – e.g., if you see a button you press or push it. But that not every verb will work. I understand that people don’t want to read the instructions, but we could do a better job of presenting this than we do.

The point about the giant audience is that it doesn’t matter if IF is no harder to learn than FPSes or reading. People are motivated to get over the hump for those, they aren’t for IF. And you can’t separate motivation from learning – it takes a certain amount of motivation to get you through a certain amount of learning. If FPSes were eleventy-billion times harder to do anything in, people wouldn’t be motivated to learn them.

Also, as someone who teaches a subject that a lot of students aren’t very good at or very motivated for, I think making it more like something they’re better at would be a great approach. Specifically, I teach logic, and I’ve realized that a big problem with all my textbooks is that they don’t introduce the new elements gradually enough, the way a game will introduce elements one at a time. I think logic textbooks could learn a hell of a lot from level design in games. (Though I’m still relying on “you have to pass the class” for the motivation.)

40

I think the problem is simply that the text input window is proportional rather than fixed. On my laptop, I couldn’t expand it to any greater than about 1.3 rows. If you set it to a fixed height of 2 or 3 rows (is that enough to fit all of the features of the demo?) it should be easier to work with.

However, to get somewhere polished with it, you may want to consider doing an extension/hack of one of the javascript interpreters–this would give a lot more control over the behavior.

–Erik