The Problem with Parsers

Like many I held out hopes for Scribblenauts being revolutionary, but I think the big problem is it didn't learn the lessons of its predecessors, the text adventures of old.  Put succinctly: Scribblenauts got the nouns down.  What it forgets is the verbs.
 
Parser commands in those old games were difficult because they not only had you worry about nouns (that were, granted, already in the game world in some fashion, not summoned by the player), but also about how you could interact with the environment, other things, things with other things, and things with the environment.  Lots of combinations to work out, and eventually you just had to hope people wouldn't be so lacking in creativity, or too creative, to get themselves stuck.  Some of the puzzles were deliberately diabolical ( Douglas Adams' games spring to mind), but other times it was a function of the game makers and the game players sometimes being in a different mindset.
 
That different mindset is dangerous if you expect players to play a game fluidly.  While you can expect some stumbling blocks in a game so ambitious, it seems like Scribblenauts wound up letting the actions that objects and you can have with the environment go by the wayside, and it apparently suffers for this huge oversight.
 
Others aren't so rash as to ignore the power of the verb.  Veteran game creator Chris Crawford is working on a system that seems more like a build-your-own-dialogue box rather than set choices.  It's sort of like a hybrid between the old, unmanageable parsers and the staid dialogue options we often get in RPGs now.  Take a look, I'm hoping more people will capitalize on the power of this idea, as it may finally help bridge the gap between immense player creativity, and a designer's limited means to harness it.
 
The openly licensed system is available on the project's main page:  

http://www.storytron.com/

4 Comments
5 Comments
Posted by ahoodedfigure

Like many I held out hopes for Scribblenauts being revolutionary, but I think the big problem is it didn't learn the lessons of its predecessors, the text adventures of old.  Put succinctly: Scribblenauts got the nouns down.  What it forgets is the verbs.
 
Parser commands in those old games were difficult because they not only had you worry about nouns (that were, granted, already in the game world in some fashion, not summoned by the player), but also about how you could interact with the environment, other things, things with other things, and things with the environment.  Lots of combinations to work out, and eventually you just had to hope people wouldn't be so lacking in creativity, or too creative, to get themselves stuck.  Some of the puzzles were deliberately diabolical ( Douglas Adams' games spring to mind), but other times it was a function of the game makers and the game players sometimes being in a different mindset.
 
That different mindset is dangerous if you expect players to play a game fluidly.  While you can expect some stumbling blocks in a game so ambitious, it seems like Scribblenauts wound up letting the actions that objects and you can have with the environment go by the wayside, and it apparently suffers for this huge oversight.
 
Others aren't so rash as to ignore the power of the verb.  Veteran game creator Chris Crawford is working on a system that seems more like a build-your-own-dialogue box rather than set choices.  It's sort of like a hybrid between the old, unmanageable parsers and the staid dialogue options we often get in RPGs now.  Take a look, I'm hoping more people will capitalize on the power of this idea, as it may finally help bridge the gap between immense player creativity, and a designer's limited means to harness it.
 
The openly licensed system is available on the project's main page:  

http://www.storytron.com/

Posted by Claude

You can play the Hitchhiker's Guide to the Galaxy for free here. There's also a second edition here. It's an online Web game. I've been meaning to try them out. When I get time, I'll check out your suggestion.

Posted by RagingLion

From what I read it sounded like Scribblenauts worked by having a whole load of properties assigned simply to each object in the game world.  I think just generally that there was not enough of these properties present for all the interactions you would expect.  They probably spent most of their effort on getting all the words (nouns) into it since that has a huge impression on how you think about the game and how you can then market it.  Maybe with some more time then can refine the game and add more verbs/properties to everything which will increase the believability of the interactions.
 
Btw are using this Storytron thing or just monitoring what's going on with it?  I think I stumbled onto it myself from something you wrote about Chris Crawford in an earlier blog or maybe from somewhere else entirely.  Seems interesting - haven't looked into it fully.

Posted by Von

Spellcasting 101 and similar text adventure games solved this quite brilliantly. You could either type in the words yourself, choose from a list of words, or even click in the graphical window and interact with some objects that way.  Not to mention that they were hilariously funny to play too.

Posted by ahoodedfigure
@RagingLion: I'm sure that if they had expanded the interactions, while maybe sacrificing some of the nouns, they would have been in a much better position.  I still think the engine they have is capable of something to that effect, but with the critical dings they took I don't know if they'll get the chance.
 
I fiddled with Storytron for a bit, but I guess I wasn't in the proper mindset.  I've messed more with the latest version of Inform than Storytron.  Since Storytron is free it wouldn't hurt to dive in and take a look.  It does NOT follow traditional adventure paradigms with regard to place.  Each location doesn't really focus on where it is relative to the other places, and the only thing separating a place is the time it takes to move from point A to B, which may bother folks who want to make mazes.  Crawford justified this in that he was concentrating more on what actually happens in a given place.
 
To be honest, the engine was just daunting enough for me that I focused elsewhere, but I always have it in mind as a possible tool to make a game, should I ever do that. :)
 
There's a political simulation that shows how the system works.  It's non-graphical for the most part.  It has a series of screens that allow you to follow a train of thought to its conclusion, with pull down menus for the various variables of an action (as opposed to guessing words or picking from a whole list, as with a parser).  You can fiddle with that to see some of the system's potential, although I don't think it's bringing the engine's full power to bear.
 
 @Von: Thanks for the head's up on Spellcasting.  It's easy for me not to pay attention to standard sword and sorcery themes, but of course themes are only one part of design.  Some games had lists of words, but they were sometimes hints, because to say every word that would be useful would often spoil puzzles.  The click-and-move sorts of solutions at least got the "use object" style of verbs out of the way, but of course they're not nearly as versatile or creative as a parser can (when it's comprehensive enough).