Discovering a feature could result into a sudden shift of product behaviour and thus implementation. When this feature is under a tight deadline, this shift will result in a rushed implementation, not giving enough time for technical discovery.
This is very similar as implementing unknown tasks, but here the unknown is the expected behaviour and not discovered work, then end result is exactly the same though.
Discovering a feature
Discovering a feature is not discovering work.
Discovered work is work you find whilst working on the imagined work. This discovered work is the most volatile and sometimes even the majority of the work. Besides bringing uncertainty, it’s harder to report on this discovered work.
Link to original
This might also clash with estimating imagined work.
When you do need to cut scope, it is not a failure, you are just ensuring software quality.