Bug prioritization suggestions/discussion

Hi all,

As Olmy and others have pointed out, asking us to submit bug fix requests and prioritize them using the same tools as new feature requests really doesn't make sense for a number of reasons. Olmy recommended that LL give us their existing bug list to work from. I second that request. I can see no value in having us start from scratch in building a bug list to prioritize.

If LL needs our help in prioritizing bugs, it may be useful to offer some guidance in the process. In software development jobs I've been involved with, we have a matrix of frequency vs. severity, which we use to determine priority. Here's a version that might work for Second Life:

Frequency:
1 - Every login, or unavoidable (9 points)
2 - Every day, or easily reproducible (3 points)
3 - Sometimes -- less than daily, requires specific circumstances to reproduce (1 point)

Severity:
1 - Client crashes or hangs, possibly even causing system crash or hang, or irrecoverable data loss occurs (9 points)
2 - Client/user suffers extended bad performance, or recoverable data loss occurs (3 points)
3 - A feature does not work as intended, but the client continues to function normally otherwise and no data loss occurs (1 point)

Multiply the two numbers together - that's priority. To combine the input of multiple people, take the median of all Frequency points and multiply by the median of all Severity points. Note that this does not require you to limit the number of bugs someone can rate. Anyone who wants to can rate any bug.

Alternatively, if LL has already classified the severity of the bugs (which I hope they have, but I have not assumed so), they could ask people simply to identify bugs which they have encountered, possibly rating frequency of encounter, and multiply the total by their own severity rating. This allows users to specify how often the bug occurs, both in their rating and by the fact that they bothered to rate the bug at all. Bugs which occur very rarely will not be rated simply because no one will care enough to rate them. Again, it is not necessary or desirable to ask users to limit the number of bugs they can rate in terms of severity.

I would still expect, however, that LL's internal bug prioritization will involve some additional criteria, such as cost to fix and possibly market vulnerability. I, at least, accept that their final prioritization needs to be internally driven, and I'm happy to have an opportunity to at least participate in bug prioritization.

What do the rest of you think?

neko

I agree. I also think creating some form of compensation for reporting bugs that are subsequently fixed via the existing tools would be an additional plus and incentive to track them down.

"Voting" on bugs is a touch counterintuitive, especially when we have incumbent tools to report them. I agree with Neko that priority should be placed on frequency and severity... not popularity.