SAN MATEO (01/21/2000) - Waiting for innovations to hit the mainstream: What about natural language?
HAVE YOU EVER wondered why some great ideas haven't taken off as well as you might have expected? I've been thinking about a handful of them for a while, and this week I'm going to wonder out loud.
To start with, I wonder what ever happened to the dawn of the natural language interface? Do you remember the product Q&A? It was last owned by Symantec -- I don't recall who originally published it.
It had a very nice natural language interface to its database module. It worked under plain old DOS, although you really had to free up as much of your 640K of memory as possible for it to work at all.
But the interface worked surprisingly well for a DOS product. I loved it, especially because you could train it to recognize slang.
My favorite query went something like, "Where is John Doe and how do I get that turkey on the blower?" Q&A recognized that to mean I wanted to get John's phone number.
There were a few other DOS products that incorporated a natural language interface.
But the only the only product I've seen to make good use of a natural language interface since the bad old days of DOS is Lernout and Hauspie's Voice Xpress.
Why is that the case? Aren't there enough people interested in typing natural phrases into their computers? Will the natural language interface have to wait until voice recognition becomes more commonplace?
Don't point me to things such as Ask Jeeves or the Microsoft Office paper clip.
They hardly live up to the kind of questions I was expecting computers to be able to handle by now.
By the year 2000 I fully expected to be able to walk up to a box and type (or say) "What is the best-selling brand of soda pop outside the United States? Who buys more of it, men or women? What age group?"
What ever happened to the object-based user interface? OS/2 came closest to delivering a useful object-based desktop, although NeXT fans tell me their baby was better.
Microsoft even promised to have an object-based desktop for Windows NT, but only until OS/2 no longer presented a threat to Windows and they could safely drop the idea.
Fans of OS/2 may remember that at least one iteration of the IBM C++ development system had an interface that was based on the desktop folder. In other words, instead of managing your files in an integrated development environment (IDE), you simply dropped the project files into a special folder.
The folder itself was the IDE. If I recall correctly, you compiled a project by right-clicking on the folder and making a choice from the pop-up menu that appeared.
At the time, it sure seemed as though this was the way for software to go. But even IBM backed off the idea and delivered a more traditional IDE with the next version of its C++ compiler. I have no idea whether IBM thought the folder concept was too hard to grasp or if IBM simply wanted to make its OS/2 compiler work more like a Windows product.
How about template-based application development? The idea was simple and sound. You define a few database tables, push a button, and out pops an application. You aren't likely to get exactly what you want. But you can fine-tune the template, add custom code to the application, and then edit the final code base when you're done.
Topspeed's Clarion is the last product I've seen that did a good job of implementing application templates.
Clarion is still around, and I doubt if Clarion is the only product left that uses templates, although I am not aware of any other product that uses application templates. There was a time when pundits were predicting that template-based application development would be the wave of the future.
So where are all the products that perfected these techniques? It's not as though I'm asking for a self-aware HAL9000 computer.
I simply expected some innovations to have panned out a little better than they did. Granted, I wasn't blessed with the power of omniscience. So there may be products out there that have exactly the features I think have disappeared or were left undeveloped. If you know of any then by all means, please enlighten me.
Nicholas Petreley is editorial director of LinuxWorld (www.linuxworld.com).
Reach him at email@example.com, and visit his forum at www.infoworld.com.