Avaya Blog by technoverse


The Blog
About Andy
Andy Green is Global Managing Editor at
Avaya. A seasoned technology writer,
Andy was a senior editor at CMP Media's
Communications Convergence
Magazine, where he covered the early
days of Voice-over-IP. He has written
about telecommunications topics for
market research firms, high-tech
companies, and B2B publications. Prior
to Avaya, he was a marketing writer for
Numara Software, a leading maker of
help desk software. Andy is a former
software developer with degrees in
mathematics and computer science.
February 28, 2008 | Author: Andy Green
Crowd Control: Crowdsourcing the Enterprise
In The Wrap-Up, Andy will be your guide
to all things contact center. The Wrap-Up
is news with context, perspectives,
real-world advice, and, we hope, a
starting point on your journey through
the blogosphere. You can reach Andy at
wrapu p@avaya.com.
• •
Net everything can be automated. Sorry, it's just not possible. We expect contact center agents, net
a software application , to sort out a misplaced delivery order. There's an interesting computing
trend called crowdsourcing that taps into the unique skills of real people. Turning the
human-versus-computer relationship on its head, crowdsourcing uses software to organize human
problem solvers to take on tasks too difficult for silicon. Think of it as an on-demand contact center
that matches customers with a virtual expert community.
In my last post, I mentioned that Amazon has developed a kind of distributed contact center
technology. Known as Mechanical Turk, Amazon's software allows companies to divvy up work
among hundreds of human agents (the crowd), collect the results, and return the answers to an
existing business workflow.
Amazon developed Mechanical Turk to solve its own internal, CPU-thrashing task: finding and
eliminating duplicate product descriptions in its vast web site. Jeff Bezos, Amazon's CEO, said that
existing Amazon software could algorithmically get very close without being completely certain that it
had identified a match. With Mechanical Turk, a real person - not a software subroutine - is called
upon to close the gap.
In reality, a request for help, known as a Human Intelligence Task or HIT, is posted on a web site.
Humans do the analysis of the multiple descriptions, entering their yes or no responses into a web
form that's read by the software.
Mechanical Turk has obvious applications for audio transcription services (legal industry), copy
editing (publishing), debugging (software), and photographic analysis. On that last point, Mechanical
Turk was used to help in the search for Steve Fosset's missing plane.
.. .i.
I . . lU. - I. . -
" —- _

To top