Opinion: Connectivity and content fall short of users' continued desire for live interaction.
Early in the story of Neal Stephensons 1995 novel
The Diamond Age, a wealthy client is mildly peeved to learn that an interactive educational
project that he has commissioned must be delivered partly in the form
of a service, rather than a completely self-contained device.
Regretfully, the project leader explains the problem: "After all of our
technology, the pseudo-intelligence algorithms, the vast exception
matrices, the portent and content monitors, and everything else, we
still cant come close to generating a human voice that sounds as good
as what a real, live ractor can give us."
On the bright side, that engineer continues, "At any given time
there are tens of millions of professional ractors in their stages all
over the world, in every time zone, ready to take on this kind of work
at a moments notice." Stephenson envisions a vast pool of talent,
ready to read a script thats generated on a moments notice by what we
would call today a Web service, with the interactive actor -- the
"ractor" -- needing little or no actual knowledge of the subject
matter. All thats needed is the ability to read, and speak, and create
a verbal illusion of being involved and interested in the resulting
conversation. Even the face of the speaker is synthesized, if needed to
support a video link, to follow the spoken words while being tailored
in appearance to the customers personal preferences.
This isnt a huge leap of techno-fantasy beyond some of the things
that we do today. When you call a toll-free telephone number, you
really dont know what time zone is at the other end of the
conversation. You dont know whether the person who helps you is really
an expert in that area, or is merely well-supported by a
keyword-searchable database of frequently asked questions and standard
procedures. A telephone call center in India is staffed by people who
dont merely speak excellent English -- theyve
even been trained in the differences, for example, between Canadian
and U.S. dialects and accents.
Combining
text, video and speech into presence awareness is now only a
mildly challenging piece of the problem. You may or may not consider
this an advance, but you can now be approached
by a helpful salesperson while shopping online. Adaptive learning
algorithms guide
live representatives in timing their approach. Electronic gaming
increasingly involves interacting
with communities of other players youve never met.
Two things have to happen, though, before reality can close the
remaining shortfalls from Stephensons vision. The first is that
script-based customer interaction has to become much better than the
one-size-fits-all compromise that we tolerate today. Its always an
exercise in patience to use any kind of telephone-based technical
support, for example, knowing that its going to take either a lot of
time or a delicately tailored exhibit of annoyance to get the voice at
the other end to skip the baby talk and start actually solving the
problem.
It seems as if the current state of the art in writing tech-support
scripts has a No. 1 goal of making sure that the customer isnt confused.
Id urge people who build such services to change that: to make the No. 1
priority a
clear understanding of what the customer already knows, and what the
customer believes. This would mean putting the state of the
customers mind ahead of the state of the product or the problem, but
any good salesman already knows the importance of doing that.
The second thing that has to happen is a massive transfer of
knowledge from the minds of people into the databases and decision
trees of customer relationship management systems. Remember "expert
systems"? They turned out, back in the AI 80s, to be enormously
difficult to build and maintain -- but perhaps their time has finally
come.
What are the other missing links between todays call center and tomorrows "racting"? Tell me at peter_coffee@ziffdavis.com
Application Development 
