Natural-language UI in the wild
Wednesday, March 25, 2009
Tagged: bot colony, starship titanic, hit list, if, natural language, interface design
Not directly IF-related, but interesting: the developer of a Mac task-mananger app tries out a natural language prompt for setting up repeating events.
Not being satisfied, and after throwing away all of my mockups and even code, I went back to the drawing board. I'm glad I did because here is the end result:
There is no myriad of buttons and fields to choose from. All the user has to do is directly type in what he wants.
(--from Better Software Through Less UI, Andy Kim)
The comments discussion that follows is good. People hit all the topics familiar to I7 debates: discoverability of underlying syntax, feedback (Kim's UI repeats its understanding back to you as a form as soon as you hit Enter), initial learning curve versus long-term friction, non-English versions.
We also have an announcement of Bot Colony, an upcoming game from North Side Inc.:
In Bot Colony, the player speaks to the characters and they answer intelligently, providing clues that help the player advance through the game. The English used is completely unrestricted: the player simply speaks and the characters answer naturally using speech, asking questions, seeking clarifications or offering comments in return.
(--from North Side's initial press release
It's not clear at all what this means or how well it works; we're supposed to get more details at GDC this week. But they do contrast it with an earlier natural-language game that didn't work:
Douglas Adams' Starship Titanic is perhaps the first game where the player chats with the game's characters (also robots).
However, the level of "understanding" of the Spookitalk engine powering the game is very different from what we target in this proposal. To quote Wikipedia, "Spookitalk had the ability to converse with the player in an almost lifelike manner, partially because it incorporated over 10,000 different phrases, pre-recorded by a group of talented voice actors. The recorded phrases would take over 14 hours to play back-to-back."
In our game, the response will not be pre-recorded, but rather a result of a parsing, reasoning on a fact base, and generation (which means turning a logic formula back into English).
(--Eugene Joseph of North Side, in an interview at gamesetwatch.)
This is somewhat unfair to Starship Titanic. That game did parse the player's input, matched it against topics, and tried to choose an appropriate response. I'm not sure how much "reasoning" it did, but you can't characterize it solely by its output mechanism. As we know from IF, output is the easy part, or at least the technically tractable part.