LCARS24, I'm one of those who downloaded and used your program, and I liked what I saw
, (what features worked on my XP system). The problem of WIMPless computing is one I've given some thought to as well, having written a logical universal language based on a seven-number system (called 7qb, or Sevenqube), which could feasibly be utilized somehow to access computing without complicating grammatical syntax from around the world.
One problem with voice rec is that computers have a hard time pinpointing words from streams of vocalizations. The actual issue of linking sentences to commands is actually not that difficult in terms of starting apps; the freely-downloaded ALICE AI from a former Carnegie Mellon professor is one such example. It utilizes an XML-parsing technique based on a statistical usage database, to filter general queries into reasonable, scripted or preprogrammed responses. Imprecise, but if centered around common desktop applications, it should be pretty easily programmed.
The challenge would be in getting it to recognize spoken vocabulary, from presumably a world of dialects and sentence patterns.
The next challenge would be in entering and combining applications, in order to command the computer to present a certain kind of output. The computer should figure out how to run its own damned applications. And inside the applications, commands ought to be accessed the same way as on the desktop. And accessible from the Web without a lot of complication. By now general computing has become pretty well figured out. We all send email, we all browse, we all boot up, can we please, please have computers that are already prepared to do these things? And don't tell me Mac, because I'm not talking about Mac, I want something better. And more, such as be ready to adapt text to the internet posting in a far more efficient way than tapping Stewart Little-sized qwerty keyboards? How hard would it be to run Word, and have all the format options easily accessed from a small string of submenus, or just a simple voice command?
Why must it take so much computing to accomplish these things? I'm no programmer but it seems to me we ought to have some common functions figured out by now, and made so blindingly fast that it's ridiculous. But by comparison, the home computing experience hasn't changed that much! A computer is still a thousand bucks, it's still got sucky graphics-heavy idiotproof GUI's, it still takes five minutes to boot up, (and ten minutes in six months), it still attracts viruses and spyware, it still requires constant patches, it still does a ton of invisible tracking and updating when you use things like Google desktop, that accumulate until surfing becomes moe like slogging through heavy underbrush. It still has only one display/sound format, and the most common applications are not improved by much needed Human Factors design and elimination of, to paraphrase LCARS24, a crapload of redundant, tedious maneuvers that you've already done a thousand times.
For example, the qualitative experience of MS Word computing has improved not one bit in a decade. I still have to wait for the thing to boot, I still have formatting problems when others change the settings, and every computer I get on has menus in different locations, and not even the same commands! Meanwhile, the simplest format problems have the toughest solutions and waste so much time trying to solve them. What do you do when you get on another office computer and suddenly all your docs have no header or footer and an annoying inability to remember you are trying to type in English? You spend a half hour trying to figure the damned thing out because of someone's whack settings are five menus across. Then you open your doc on another computer, and your own doc format has changed!
A word doc ought to have a basic menu system, each menu opening a ready submenu easily scanned in 2 seconds. Whether it's iconic, textual, WIMP or touch. I realize a program like Word has a ton of features; but it ought to be simple and fast to be able to name your document problem and fix the damned thing - instead of opening that god-awful paperclip help program. The beauty of TNG computing is its simplicity and clarity.
The TNG voice interface has its advantages; you can use qualitative verbal inputs without finding applications. The computer is continually on standby for commands, and it can access external controls for home management.
But anyone who has used a verbal interface, as with speech-to-text typing, knows that while a great invention, it hardly flows, and impedes too much to be any fun. It's like the old days of constantly stopping typing to get out the white out. No thanks.
But qwerty, and the global language English, do not seem like protocols we would want to limit computing to. There may be more efficient systems, such as numerical. Another point? Talking to a computer, say to type a document, is actually tiresome. I believe the most efficient input device is the DTMF pad; a dial of 5-10 buttons, easily known by touch. The mouse is imprecise. But 7 keys alone, are very hard to miss. Watch an accountant hitting a calculator to see what I mean. While I have becomed accustomed to the qwerty, there has got to be a better, more logical way, to learn how to input language. And not necessarily by laying out the alphabet.
One more feature I would like for computing is for the computer to access a database of applications, choose which ones are needed, and combine them into modules automatically; so you could feasibly create a computational model from the inquiry alone, for example: "How much rainfall does Bogota receive?" Computer requests: "State time frame by day, week, month or year." It accesses Bogota: rainfall and relays the most probable answer. Then you say, "Computer, run a comparative analysis of Bogota rainfall and Barcelona's, and tell me which is most likely to have a sunnier climate this June. Now predict flight delays and cancellations and which city is more likely to have fewer cancellations due to weather." The computer figures out how to do this on its own, perhaps from an expansive database of application modules which can be readily mixed-and-matched. Compare this to googling "probability of flight cancellations Barcelona + Bogota". It wouldn't be just googling any old static SEO-content from the topic; it's actively creating (predictive) data from a reliably secure source. It's accessing information you didn't think to ask. And answering in multiple modes of output: vocal, graphical, textual, formatted for a report, a newspaper article, an email, or any other requested format. Then you say, "mail this to Karl" and it bloody well knows who Karl is by now.
Anyway, the LCARS model is simpler than WIMP and that's a good thing. However, the logic is superior and more powerful. Faster and more versatile, and more hidden. But one key is the ability to access information or applications that the user doesn't know about. In order to do this, then the "desktop" and the "application modules" had better be pretty much speaking the same language, and all modular, and all accessible from any other point within the system, and all already running.
