I have no idea what type of data it is sending (that could be another interesting security-related topic), but have any of you seen or implemented this type of solution before? Having a catch-all for errors in the application that sends them back to you for analysis?
For MS, they know people are connected to the net when they use the app, but it could be more difficult with other apps. For applications used inside a corporation, it may be feasible, as well.
Netscape has this feature built in already... I don't know what it sends back, but it asks the URL you were at and what activity you wer doing at the time of the crash. It also allows you to not send the information at all.
I worked on a client/server application that would have event handlers to do a variety of things. The server had a monitor that checked internal status, log messages, overall environment... the monitor could be configured to send an email when a particular message was seen or any level 1 (critical) message. In some cases (when the customer allowed it/wanted it) we had the server automatically email us log files, config files, and dump files that our server created.
There are also 3rd party event monitors that look at OS log files and application log files then notify people or report the problem to its own log.
It is an interesting concept, but whenever data is being sent automatically via an application the customer/user wants to know what it is (or they dont send it). Theoretically the application can send anything on the system provided the permissions are appropriate.