Pros/Cons to text parsers

[Snake]

Hey, everyone,
I've came across the idea a long time ago to do a text parser for The Legend of Leitor's Edge, but when I got thinking of all the work to include one, I decided not to.

Now I'm thinking I might want one.
The reason why I'm thinking this is because text parsers can open up a wider range of interactions, puzzle-wise, opposed to the basic, TALK/LOOK/INTERACT cursors.

I'm thinking I could very easily avoid a parser and include the other options a different way in a menu, like I was going to do, but something about that parser is very intriguing.

I'd like to know what everyone else's input is. Pros/cons. Who opposes and who doesn't.


Thanks in advance,


--Snake

[nulluser]

-

[Radiant]

This GUI may be a reasonable compromise?

</shameless plug>

[accolyte]

Me personally, I wouldn,t object to a text-parser as long as it,s at least reasonably flexible. Good parsers can add a lot both to game depth and puzzle-complexity. (Whereas a bad parser is a major pain in the ass ) Another thing I like about them is that if you get stuck and bored, you can waste your time by trying all sorts of irrelevant and crazy commands, such as JUMP, SCREAM, SING, FART, PEE, etc to see if you get an answer. So if you think a text-parser could add to your game go ahead and implement one is my advice.   

[EagerMind]

I think a properly-implemented text parser could result in a more immersive experience than the typical point-and-click interface. But my vision of this is basically a complete melding of an IF game with a graphical adventure game (AG) - you type everything in with the keyboard like in IF, and watch the events play out on the screen like in an AG.

In my limited experienced with adventure games using text parsers, I've noticed the following shortcomings:

1. Most parsers are fairly limited. Take a look at some IF games made with Inform. You can pretty much type anything in and the game understands what you mean.

2. Moving around is extremely clunkly. Switching between the keyboard and mouse isn't really a "true" text parser experience and is a nuisance for the player. But I also have issues with using the arrow keys. It still takes your hand out of the typing position, and as a result I find it just about as bad as using the mouse.

Of course the problem stems from the fact that there's no clear "north, south, east, west" in an AG. Some way needs to be found to label the exits so you can simply type something like "exit <exit 1>" or "go to <exit 1>" or even just "<exit 1>".

3. For similar reasons as #2, interacting with objects in the room also becomes quite clunky and tedious. Normally you have to move the character into "position" or move the character's gaze around to "focus" on the object you're trying to interact with. Also, there's the potential problem of actually identifying what it is you want to interact with. In IF it's literally spelled out for you, but in an AG it may not be so obvious unless you're somehow able to bring up hotspots.

Of course, in IF there really isn't any notion of movement within one room. By reading the room description, you identify an object of interest and can immediately start acting on it. Consider how easy the following actions unfold in IF, and how much work is required in a typical AG with a text parser to do the same thing. Maybe I'm being too fussy, but it'd be nice to somehow make the experience be similar.

"You're in a room with a desk."
> l at desk
"There's a sheet of paper on the desk."
> l at paper
"The paper appears to have something written on it."
> read paper
(picking up the paper first)
"The paper says 'See how easy this was?'"
> n

[Vel]

Albeit I told you my opinion on IRC, I shall elaborate upon it.
In theory, parsers can give you much more freedom and you can do whatever you wish in a game.
Actually, however, creating a parser that understands more elaborate things than verbs plus nouns is a very hard thing to do, and even more so with the limited parser possibilities of AGS.
While it gives the freedom of using  obscure actions such as say, kneel, jump or pronounce, their implementation is something that is more likely to lead to exasperation than to a rewarding puzzle. Some silly and ridiculous examples of this are the ridiculous "say the word" and "hold gem in mouth" puzzles in SQ2.
So what I would suggest is that you use a GK3-like point and click system, showing only the available actions when a hotspot is clicked over. That would allow you to use various actions while sparing the player the frustration from non-obvious actions.

[Radiant]

Perhaps someone should write an INFORM plugin for AGS.

[Kweepa]

I was thinking about how that would work...
Probably the way to go would be to have an Inform module that modelled the game world (via a .z5), and translated any complicated commands into simple verb-noun commands that could then be parsed by the game.
Keeping things synchronized would be a problem though.

[mkennedy]

Text parser are fine, The ability to combine the robust parser ability of something like PAWS or ADRIFT with the graphical ability of AGS would be wonderfull!

[Rui 'Trovatore' Pires]

Pros/cons are the same as to any interface. I guarantee you this - if it's well implemented, people won't *care* what interface you use.

Musts with text parser:

- PLENTY of synonims
- be ready to script many possible interactions, many many many. EVERY possible interaction should have some sort of response, ideally.
- Avoid guess-the-verb situations like the plague
- Be consistent, avoiding the obligatory use of the mouse
- Shortcuts (x = examine, inv = inventory)
- "Retype" function
- if possible, add an "it", and a way to use the PARSER to interact with the INV items in your possession. In my experience, this is the hardest bit to code.

I wouldn't suggest Radiant's template, because as cool as it is it's overwhelming. And if you want to use a text parser or are in any way going for the "multiple interactions" route, overwhel is what you should NEVER do. A text parser interface shouldn't be hunting for commands - the whole point of the text parser is to make playing as intuitive and flexible and UNcomplicated as possible.

No, it ain't easy, but I guarantee you it's extraordinarily rewarding. The best parsers are the ones you don't know are there. I reccommend to you Anchorhead (Inform game) for a brilliant use of the parser, and even for the brilliant conversations you can have with the "ask X about Y" format.

[blueskirt]

Rui, while I agree with most of what you said, "Ask X about Y" can easily be a major turn down and a guess-the-verb situation when it is badly designed and doesn't have enough synonyms. Often in games with text parser I just wanted to discuss about something very mundane and revelant to the person I was talking with, but easily spend some minutes just to guess the exact word required to ask about it.

I wonder if it would be feasible or better, rather than having tedious "ask X about Y" fest, to just program a "talk to X" which would open a dialogue mode, similar to the SCUMM games where you would be given 3 or 4 conversation options, each possible choices with a number attributed to them, and all you would have to do would be typing a single number or letter to converse with persons.

Same thing with Sierra kind of game, rather than ask X about Y, you would simply have to type a number/letter to pick a subject from a list of possible conversation subject that you have unlocked so far (kinda like GK1), a conversation tree with a "back" option (like the point and click QFG games), or simply a list of subject that are revelant to the character you are currently discussing with. I guess it would have to be designed further to get some sort of standard dialogue mode, one that is easy and simple to use, so the conversation mode do not change from games to games.

If well coded, it could also allow you to manually type conversation subjects, may it be optionnal conversation options that are useless but just help building a mood, or key important conversation subjects, which, once discussed, would be added to your list of subject you can now discuss with everyone.

[Rui 'Trovatore' Pires]

Blueskirt, re "ask x about y", I agree with you totally, and that's why I suggested Anchorhead. I have NEVER seen such a fluid conversation system. It doesn't feel like one, it feels like actual conversations, it's quite amazing.

[Snake]

I also like that idea. I haven't messed with parsers at all, but I'm guessing that it is possible to simply call up a dialogue box or gui with options on it and go from there with the shortcuts. Why woudn't it be possible?

Vast synonyms for sure. Can't go wrong there. I hate it myself when you've got to get it exactly right or you get one of those, "I didn't understand the word, X".

I agree also with having just as many possible interactions as you can fit in there with an equal amount of responces. Even if something isn't right or doesn't work, I like to watch the character at least try first rather than say right away that he/she can't do that or it doesn't make sence. Well, did you try, jackass? No, you didn't. You're just standing there like a penis.
That in itself will add to the game world and make you feel more involved and not just controlling a player character.

Shortcuts are an amazing thing and take away the frustration of having to type a full sentance again if you didn't get it right the first time. The "repeat last..." option can also be a life saver.

There was going to be something else I was going to say, but I've forgotten.


--Snake

[thewalrus]

     I'd like to see a really good parser mixed with a graphical adventure game. Although all the points that everyone brought up are very valid. There would be a lot of work to get to act just right. I guess it comes down to if your willing to put in the work and time to make a really great parser and have tons of interactions..... 

[Erenan]

Am I the only one who deeply enjoyed Quest for Glory 2? I suspect not. I found its combination of graphics and parser to be rather well implemented.