In education, the classic bell curve represents the distribution of grades that occurs given a relatively large sample of students: a small proportion will get very low and very high marks and most students will get average marks. A J curve (or J-curve) distribution implies that most students can occupy the rising part of the "J," which means most students can get above-average marks. This theory seems to have come about based on research presented at various "Outcome-Based Education" conferences in the mid-90s. (The forerunner is "mastery learning theory," which hails from the 1970s.) I found a number of citations from 1994, including the following earliest citation:

"The principal said the old view of the bell curve and testing 'to sort people out' is becoming less important. Educators now lean toward a 'J' curve philosophy that strives to move all students to the high end of the curve, based on the belief that all children can learn." —Sharon Henson Pope, "District Uses Test Scores to Help Shape Curriculum, Principal Says," St. Louis Post-Dispatch, August 29, 1994