Why don't you know what caused the problem?

This is a common (and fair) question that I'd like to take a moment to address at some length.

For the most part, each unique problem results in a unique Bucket ID that can be found in the Windows Error Reporting Log; however this isn’t always true. While it is apparent where the crash occurred, the actual situation that led to the crash (the root cause) may not be apparent or available. There are a large number of failures for which appropriate blame cannot be assigned without additional data, and secondary analysis of that data once received.

This results in situations where one distinct bucket can actually contain crash events that are caused by completely unrelated problems. Since the bucket is the only link between the client and the server, the bucket is the only avenue for providing more information about and solutions to problems. As a result, those buckets that are contributed to by more than one problem become very difficult to respond to in a fashion that is meaningful to everyone who might see the response.

Traditionally, buckets that do not correspond with one and only one crash event have very rarely been responded to, due to the aforementioned difficulty in posting a precise and helpful response. As we work to augment our responses with this blog, improve bucketing, and change the response system, we are attempting to tackle the problem of responding to a broader class of failures.

If you would like to read more about how Windows Error Reporting works, you can visit this site.

If you are an ISV (Independent software vendor) interested in the types of failures that your application reports, you can get more information on this free service here.