Well, it's easy to demonstrate. Imagine there is a rubric with three criteria all having levels 1, 2, 3 and 4. The reviewer gives grades 4, 3 and 2 for these criteria, in order. Then the final grade given by the rubric is:
((4 - 1) + (3 - 1) + (2 - 1)) / ((4 - 1) + (4 - 1) + (4 - 1)) = 67%
(In other words, the submission got 4 + 3 + 2 = 9 rubric points. The maximum possible number of points in that rubric is 12. The minimum number of points is 3. Those 9 points are at two thirds of the interval 3 - 12).
If you used the similar setup with the accumulative strategy (three criteria, each of them having max grade 4 and all having the same weight 1), then the grade is
( 4/4 + 3/4 + 2/4) / 3 = 0.75 = 75%
because the accumulative strategy automatically includes the zero grade as an option. While in case of the rubric, the level with the lowest numerical grade (1 is this case) is considered as the "worst" level performance, regardless the actual value of the level.
To make calculations in rubric more obvious and intuitive, it is strongly recommended to always include the level with the zero mark. With rubric sheets printed on a paper, teachers tend to have an "implicit" zero mark if they simply leave the criterion empty without any level checked. We realized this would have bad consequences when it comes to web based rubrics.