The single most important skill that can be taught in school is autodidactics, learning to learn independently. That can be taught just as well with everyone in the same classroom as otherwise, because it is not done by lecturing the class as a whole, or otherwise teaching the same content to the whole group at the same time. That's the mistake this experiment made and why it failed. It is done by requiring and allowing each student to pursue their own study at their own pace. That is, by replacing the regimented assembly line Prussian model of education with a model of individuals breaking their own paths through trial and error.

This is not a new model but the default of the old one-room schoolhouse, where the toddler learning to read sat next to the teenager learning to read Latin. In this model the teacher's job isn't to directly implant knowledge into the students' neural nets, it is to guide them to the correct trail head when they are ready for it, and nudge them back on track as needed. Doing this for each of thirty kids is harder than making one lesson plan for the same group and insisting that they work in synchrony. So it can take more teachers per student to do well and looks inefficient to the Prussian, until you examine the results. Particularly the results of the most capable students who would otherwise be paralyzed and repulsed by boredom. But also of the least capable who would otherwise learn little unless the whole class is taught to the least common denominator.