Yesterday, I was watching this rather long, but very interesting video about Dasher - how it was created, what it has become, and what it could be.
(Skipping forward to 25m and 40m will reveal some instant coolness).
Before watching the video, all I could really conceive dasher to be useful for was trippy text entry and perhaps as an alternate (but weird) way to zoom through an abstract hierarchy - a replacement for right click menus, perhaps, and certainly an improvement (or welcome alternative) for most consoles' virtual keyboards.
But then the video started going into what was happening behind the hood: The prediction algorithms are simple, but could be effective at destroying TXTing's malformed literary artifacts. While an abc-123 style mobile phone text input encourages you to write like a... well... like a cunt (I got dumped over TXT once, incidentally), Dasher naturally encourages and enables you to spell better, without being a game of Thumb-Twister*.
Dasher could have just had endless letters nested within nested nests, with no weighting on the size of nests at all... and it would have completely failed. Since it weighs predicted/deducted letters heavier than other letters (without removing those letters), it suddenly becomes very fast to use... but it's more than that. It bends the possibility space implied by the alphabet to encourage valid words while discouraging illegible ones. In essence, if you're as bad a speller as I am, Dasher can improve you simply because it's giving you a helpful nudge in the right direction by making spelling errors harder, and correct spelling easier.
In a funny kind of way, it's a microcosm for how games designers work (or how they probably ought to work): when we play a game, we're flying along a path through possibility, and creating a story (or, strictly speaking, a "set of events". Stories are edited with dramatic impact in mind... but this is a whole other fiasco of a discussion which I'll avoid for now). This story can be as boring or as interesting as the system abhors, but as designers, we can do a lot to encourage the interesting paths, without disallowing the boring ones - if we were to do the former without the latter, we'd not only compromise a consistent, complete, believable, immersive, systemic world, but also disallow the player's ability to create their own pace.
So anyway, as I said in my last post, I had thinking about doing both a "radial" version of dasher, but also a music game. When I heard Mr. Dasher talk about how he was using many different "grammars" and "probability models" to drive the weightings in Dasher in different ways, my mind opened a bit more, and my love of it increased.
Sleeplessly trying to shift a cold, I played a bit of Aquaria. I ended up just laying at the bottom of the ocean, playing out a tune on its radial singing interface. By almost sheer luck (or could it be Alec and Derek's foresight to keep music in the same key as the sung notes?) I was able to play out a tune which sounded really nice. Now, this was a completely alien musical instrument to me, and I don't have a great amount of musical talent myself, but somehow I was jamming along happily and without any noticeable error. How the hell does that happen?
Then I considered a documentary I had seen about how a computer was taught to create their own variations on famous piece of music, using music theory, mixed in with chaos theory to provide the variation. This got me thinking - if music has its own grammar**, and Dasher is, from one point of view, a way to explore grammar (without compromising your ability to explore stuff that lies outside of the grammar), they might be combined.
A "Music Dasher" (or what I'm dubbing "Beet Root Jam") might be a way to give a novice musician at an abstract instrument a way to produce music that sounds pretty "correct", but unlike something like Frequency, Amplitude, Guitar Hero etc. doesn't constrain a player's expression to a prescribed "correct" one. I.e. Jamming made easy.
I don't know yet how I'll fit a game around this... at first it needs to work as a toy - totally abstract, just to see if the concept even works. Once the "toy" is both proven and also "fun", I can layer on higher level mechanics which will encourage (but not force) players to develop their songs with respect to a higher level grammar.
Here's kind of how the interface may work (but this is a slightly more generic one, for non-music stuff as well - menus or whatever):
I'll have another quick picture up tomorrow which explains how, no matter what arc you move into, the discarded ones always slide around to the left hand side, so that going "up" the hierarchy isn't a case of twitching all over the place - you just hold "left".
*I know Sony have been in the dog house for a while, for a wide range of reasons. It sort of upset me, because I want everyone to make cool stuff more than I want a schaden-freude-hard-on. However, it only took one mistake to fix my schaden-freude-erectile-dysfunction. That reason is this: the PSP text interface. What kind of blind-in-the-thumb SonyEricsson-whipped producer decided that a virtual mobile phone pad was a good idea? Look at it. About all it's got going for it is "familiarity" (a dirty word where interface innovation is concerned. Consider the story of The New Suit), but even that's destroyed as you have to manipulate a cursor, one step at a time before you can actually select a key, which you have to press multiple times before you get your desired letter! The thing is, I actually talked to a coder responsible for the interface, and, poor sod, it wasn't his choice at all - some mentalist higher up the chain of command probably decided was a safe bet. Shit rolls downhill, and blame rolls straight back up.
**Music Theory as a be-all, end-all theory of "what sounds good" became pretty punctured by Jazz, which proved that music could defy the so-called rules but still sound right, somehow. Luckily, the approach I'm proposing (much like dasher) doesn't stop you playing the "wrong" notes if you happen to think, despite the music model, that they sound about right.
14 hours ago