Bugs, like features, are “work”, and no one likes to think that there is work that cannot be estimated. However, Considering the difficulty in estimating a feature, how could anyone possibly expect an estimate of time for a bug in that feature? There is an assumption that features are “big” and bugs are “small”. Granted, if a feature was implemented correctly, there should be relatively few bugs, and the bugs that are found are quick to fix. This is practically the definition of a “well implemented feature”.
What happens if a feature was not implemented correctly? This results in the proportion of time needed to implement the feature, and fix a bug-in-that-feature’s estimate can be vastly off. I have seen a hand-full of bugs scrap an entire feature, because the bugs were unfixable, thanks to the way the feature was implemented.
As bugs cannot be estimate, implementing a feature correctly needs to be top of mind, if hitting deadlines is important.