I've seen courses using some variant of this. It's fine for the most motivated students, but many other students will do better with more traditional grading systems, for two reasons:

1. Grading things in the middle of the course provides a separate incentive for early course corrections. This is especially important for students who are heavily conditioned by the gamification of learning.

2. Cramming hurts learning very badly, and this system increases the incentive to cram. In the extreme, you have to choose between learning (through a more effective approach of distributed practice) and passing the exam.

Gamified learning in general doesn't seem to produce very good results, but they do seem to be better than the results most people achieve left to their own devices.

Final-first exams is a version of this idea that works even better: https://slimemoldtimemold.com/2020/09/14/hindsight-is-stats-...

I think there is a more important concept embedded in this essay. Don't only assess something once and provide useful feedback during those assessments.

One of the criticisms I have of midterms in university is that you could end up getting the results back weeks or even months later. How was that useful for students to learn?

Did a 4 year EE engineering degree where almost all subjects 95% exam or more, two hour exams.