From: "cfb" <cfb@nirai.ne.jp>
> .... If you're going to use technology to
> solve business problems in Japan, you're going to run into the brick
> FAX machine wall very quickly.
Tell me about it.
Over a year ago, I proposed to a client that they supplement their paper
mail-order catalog with a wireless ordering site -- a synergistic
combination, it seemed to me. I did a demo, saw some initial enthusiasm,
but then ... well, they changed their minds and said they wanted to do a
"real" website first.
Last I looked at their "real" website, you ordered through it by printing
out a form page, filling it out by hand, and, yes, faxing the order form in.
Yes, there is some value added by this site. Customers can listen to music
samples before they buy, for example. This is cheaper if delivered through
an ISP, and provides better quality sound than keitai, depending on their
PC's speakers.
Pivotally, however, the "real website" approach created more work for the
client's employees. That's what really made it ever so much better. They
now do as much or more manual data entry as they used to, typing in orders
from scrawled faxes, just as before, plus they have a website that needs to
be updated in parallel with their paper catalog.
Who could ask for more?
(No. I. Am. Not. Bitter.)
All sarcasm aside: the main barrier isn't the technology. The fax machine
itself is a brick wall, but it's a wall propped up by people. The main
problem is that a large percentage of office workers in Japan traditionally
(and to a great extent, legally) own their jobs. They don't lose these jobs
piecemeal -- those jobs have to go down with a whole sinking ship. In the
meantime, automation doesn't get started because there are always plenty of
deck chairs to rearrange. (Gee, I started this with "all sarcasm aside,"
didn't I?)
> I'd say that OCR at the FAX machine (both inbound and outbound) would
> be an interesting attempt... (obviously, someone out there has done it,
> but has anyone done it in a data warehousing manner on centralized,
> network accessible file stores (and I'm not just talking about FAX
> image storage)?... Ok, I'm sure it been done too... but what about
> the mobile, company keitai using part of your work force?).
If the source of the FAX data is other employees, there's potential for
discipline -- the boss can crack down on the sloppy ones, or shuffle them
off somewhere else.
(If the source of the FAX data is customers, that's another story.)
Having a mobile workforce deliver data by FAX instead of keitai, and using
keitai interaction to clean up any OCR misunderstandings on the spot, seems
a more likely proposition. This is a case where the sender knew what he/she
wrote, AND is under the gun to get it right. If they don't write clearly
and inside the lines, they are "punished" by having to thumbtype
corrections. Whereas a customer might simply respond to such "punishment"
by going to another vendor ...
On the other hand, if you're going to go that far, why not just use
palmtops? While I see mobile phones eventually taking a big bite out of
what would otherwise have been palmtop market share, I still suspect that
the palmtop market as a whole will survive, in part because of vertical
market applications involving mobile data entry by company employees.
> With regard to M. Turner's original observation, if you do the math,
> I believe you'll find that the cameras that are starting to appear
> in handsets would need to have resolution an order of magnitude (or
> more) better to be able to provide images of text documents suitable
> for OCR purposes (esp. considering how tiny some of the office type
> 'A's write... not to mention the fact that most cameras these days
> have lenses with artifacts that are mutually exclusive to applications
> such as OCR)....
[snip]
I haven't done the math, and I should. If you carefully read my post,
however, you'll see that my not-entirely-tongue-in-cheek proposal isn't for
reading arbitrary pages -- and definitely not for blindness-inducing
bureaucratic forms. Nor would it be text-only input: I suggest a parallel
speech recognition pass for correlation and correction.
The user could adapt somewhat -- writing larger characters, with a wide-nib
felt-tip, if nothing else. Those who like to write tiny characters would
opt for some other system. Enthusiasts might buy purse-size whiteboards,
erasers and magic markers, to save on paper. (And learn about the "whisper
mode" mic setting on their keitais.)
The real question is, Is this worth it compared to thumbtyping, for enough
people? If the answer is Yes, you might even see the camera-makers adapting
to it. I suspect, however, that the answer is No.
-michael turner
leap@gol.com
Received on Mon Jan 7 11:48:47 2002