Superior systems operating under maximum control, such as those of FedEx, currently
aspire to over 98% accuracy. A more usual wellproven, wellmaintained system of laser
readers can boast of 96 to 97% accuracy, which still means that 3 to 4 bags out of 100
go to the misread pile or the wrong destination.
The effect of misreads can be great, both on cost and on performance. At Denver, the
usual problems of identifying bags properly are compounded because the baggage
system requires two distinct kinds of readings: the destination of each bag must first be
read by lasers and then this information must be transmitted by radio to devices on each
of the baggage carts. This duality compounds the errors. The reliability of two devices
working accurately together is, roughly, the multiplication of their individual reliabilities:
that is, always less than either alone.
To deal with the inevitable misreads, the most important thing is to have a backup
system. This is standard, but costs money and time, and degrades performance. The
complementary solution is training of personnel and continuous improvement of the
system this comes with experience and may take years to acquire.
Managing the information accurately is also difficult. The database needs to track tens
of thousands of bags, going to hundreds of destinations, all in real time. The problem is
further complicated at Denver because it uses a distributed system of about 150
computers. The software must, in addition to the usual error checking codes that guard
against electrical disturbances in the communications, have multiple levels of
redundancy and be able to recover from errors very rapidly. Getting this right can take
many expensive programmers a lot of time