|

 
|
| | Name : | Ron Crane | Organization : | N/A | Post Date : | 9/30/2005 |
| Section : | .20.4.5 | Page no. : | | Line no.: | | Comment : | 10d.
d. Vol. II, §4.5 allows test labs to use a software-based "simulation device" to test "data
detection and control paths that are used in casting an actual ballot."
If the simulation device cannot test all the paths, "then an independent method and test procedure must
be used to validate the proper operation of those portions of the system not tested by the
simulator." If the system is not subject to rigorous code review, this kind of testing makes it possible for a crafty cheat to escape detection.
Imagine that module A is the
touchscreen I/O module, and module B is everything else involved in collecting, collating,auditing, etc., votes. B could be configured to cheat only when module A is also present.
Thus, if the test lab uses a simulator to drive B, it won't find any problems with it, and it wont find any problems with A at all, since it is not programmed to cheat. Cheating
occurs only when A and B are integrated. | |
|
|