I agree with the 12 drive solution to the 6+6 problem, but that is only if the drives fail where you want them to, a theoretical maximum. Simply put, you cannot afford to have 3 drives fail in each of the RAID 6 subsets. You CAN afford to lose 3+ in two of those subsets, but only two in each of the other two subsets.
Therefore, we are back to a practical maximum of 10 drive failures.

With 10 drive failures, you can guarantee that two of the RAID subsets will have two or less drive failures, thus preserving your data.