If I spend 10 minutes solving A, submit and then spend 20 minutes solving B and submit, I will get less penalty than someone who solves A in 20 minutes, submits, and then spends 10 minutes solving B. This doesn't seem right to me. Seems better to just give people time penalty based on the last AC'd submission (+ possible WA penalties). If someone spends a lot of time solving the first problem, but then quickly finishes all the others, and is done with all the problems 2 hours before the contest is over, seems like that is a stronger competitor than someone who solves all the first problems quickly but really struggles with the last two, and ends up submitting the last AC 3 minutes before contest is over. But with the current system, the latter could potentially fare better.

Is there a reason this isn't the way its done? Not criticizing, just curious if there is a rationale behind structuring the time penalty system this way.