I agree, but the two are very similar if you read 'defects in a program' as bugs. Bugs can be hard to cause, and thus not be noticed by user acceptance testing

I agree, but the two are very similar if you read 'defects in a program' as bugs. Bugs can be hard to cause, and thus not be noticed by user acceptance testing. So for catching them, tests are more useful.