27Bstroke6 wrote:One is the use of too many total input buttons to choose from, if button sharing is taking place in any one input mode (e.g. alpha, numeric, etc.) Research suggests that when you can't provide a separate button for every letter (e.g., a real keyboard) it's best to to use the absolute minimum -- ideally only one in conjunction with some other modality of input. An example is a mouse, which can select in 2-D space plus one "button" (if we consider left click only).
But only so of course if you want something that is easy to get into.
The Dasher model does away with the button entirely by controlling all input exclusively through mouse movements (or in the PSP case the analog pad).
And I do really like that. It's just that the analog pad is the one key, preferably along with the R and L buttons, that I would like to keep separated from the keyboard entry so that the analog pad and shoulder buttons can be reserved for mouse emulation.
Also, after testing Dasher, I have some concern that the analog pad might not be sensitive enough for Dasher - it's not all that easy or comfortable to control, especially when you need mouse-like behaviour and precision. Nevertheless, I think with some tweaking you might still get a workable setup. and I definitely think it's worth to port Dasher to the PSP, even if only to wow people with.
This is certainly not the only viable model, of course. But multiple buttons are to be avoided, and this problem becomes even more intense if the meaning of any given button can change in the course of input.
Agreed. I'm trying to avoid that as much as possibe. But in this case, there's something to be said for the system I'm using. First of all, you can basically know the two button combination for the keys from looking at the user interface directly. Right now I'm barely reinforcing that - in fact, the redistribution of letters over the buttons might just harm it more than it helps - there's a lot to be won there.
Also, you're mostly not moving your fingers, just like on a real keyboard, you can hold your fingers of both hands over the buttons, and they won't generally have to move at all. So you can start to memorise the button-finger combinations. And you can put down your PSP on a table and it is almost like a real keyboard in that respect.
Worst case: multiple buttons with changing input definitions in a single input mode.
Agreed. But my system doesn't do that all that much.
Dictionary-based systems (you must be careful of existing broad patents once you get into this) basically come in two flavors, static and learning. The former is simple to implement and fairly limited, while the latter is much more powerful and of course more difficult to implement (and to do right, since you don't want to "contaminate" the learning dictionary with errors).
Yep. But I know enough about linguistics and such to pull that off well if necessary. However, my first goal is to make a very light and easy, non learning system. Not that many systems really benefit from learning anyway - that can, after all, also interfere with the changing buttons concept, of which MS Office's automatically adjusting drop-down menus are a very good (or bad) example.
It's worth noting that another advantage of minimizing the number of input buttons in the input choice set is that it allows for more "modeless" input to applications. E.g., if an input scheme requires choosing from among four buttons, those buttons are not available for the application when in input mode. On the other hand, if the input mechanism uses a totally different part of the device (e.g. analog pad) the buttons are still free in a modeless operational sense. Again, other models are of course possible.
True. But like I said, it's the analog button and shoulder buttons that I'd prefer to keep free (mouse movement, word selection, left click, right click - it's worth noting that Dasher doesn't do a lot of these things, and my system would be a lot more natural when, say, filling out a form or editing text - but you could get used to switching to Dasher and then back to an editing system also, I suppose).
I will probably make the system configurable - in a situation where the left and right analog buttons are available, the dictionary system would win a lot from being able to use them to get into the dictionary directly and autoselect the first suggestion with a double-click for instance, rather than through the (SELECT) button-menu, and get faster to the CAPS and Numbers/Symbol option.
I still have to work out a little which features will be used most, and in which setting (obviously, a phase one/no dictionary implementation can do without the dictionary options, making it much simpler)
Have you tried the latest version yet at all? I'd like to know what you think. I'll be polishing it up, but then I'll soon start working on a basic phase 1 version on the psp.
