How CramRocket is built

Lessons start with recall, not review

Every CramRocket lesson opens by asking you to recall the previous one before showing you anything new. This is retrieval practice. Roediger and Karpicke's 2006 study found that on delayed tests, prior testing produced substantially greater retention than restudying the same material, an effect they named test-enhanced learning.

We put it first on purpose, and it is the least popular part of the lesson. A quiz is a worse opening than a fresh page of content, which is why most study apps skip it. That trade is backwards.

The teaching follows a fixed six-step sequence

Lessons are built on six steps: a curiosity hook, concrete before abstract, one idea at a time, active construction, naming the trap, and a transfer task. The sequence draws on the worked-example effect, which Sweller and Cooper demonstrated in 1985 when they found that students learning algebra gained more from studying solved examples than from working the equivalent problems themselves.

The order carries as much weight as the content. Concrete before abstract and one idea at a time both exist to keep working memory from overloading, which is the failure mode when a lesson tries to explain too much at once.

Lessons end with you explaining it back

Each lesson closes by asking you to write the concept in your own words, and the explanation is graded against a four-criterion rubric. This is the self-explanation effect, documented by Chi and colleagues in 1994: students prompted to explain material to themselves as they studied understood it substantially better than students who studied the same material without explaining.

This is also what the Feynman technique is. The difference here is that something checks the explanation instead of you deciding it sounded fine.

The tutor asks before it tells

When you are stuck, the tutor questions you rather than supplying the answer. Being handed an answer feels efficient and teaches almost nothing, which is the central problem with using a general chatbot as a tutor. You end up watching someone else do the reasoning.

It is slower. It is occasionally irritating. It is also the difference between following an explanation and being able to produce one.

Flashcards are scheduled around forgetting

Cards are generated from concepts you got wrong and scheduled with FSRS, a modern spaced-repetition algorithm. The underlying finding is the spacing effect: reviews distributed over time produce more durable memory than the same amount of review massed together. Cepeda and colleagues confirmed this in 2006 across 839 assessments drawn from 317 experiments.

Ebbinghaus mapped the forgetting curve in 1885. The interesting part is not that we forget, it is that forgetting is predictable enough to schedule around.

Practice exams are a learning event, not a measurement

Timed practice exams generated from your own material are not only there to score you. Testing is itself one of the strongest learning interventions available, and the difficulty of retrieving under exam conditions is what makes it work. Robert Bjork's 1994 work on desirable difficulties covers why practice that feels harder produces better long-term retention than practice that feels smooth.

Most students treat a practice exam as a final check. It works better used early and repeatedly, while the results are still useful.

The readiness score exists because confidence is a bad signal

Rereading material until it feels familiar produces confidence without competence. Roediger and Karpicke found this directly: repeated studying increased students' confidence in their ability to remember the material, while producing worse retention than testing did.

The readiness score replaces that feeling with a count of concepts you have actually demonstrated, across three separate gates. It is deliberately unflattering.

References

1. 1.
   Bjork, R. A. (1994). Memory and metamemory considerations in the training of human beings. In J. Metcalfe & A. Shimamura (Eds.), Metacognition: Knowing About Knowing. MIT Press.
2. 2.
   Cepeda, N. J., Pashler, H., Vul, E., Wixted, J. T., & Rohrer, D. (2006). Distributed practice in verbal recall tasks: A review and quantitative synthesis. Psychological Bulletin, 132(3), 354–380.https://doi.org/10.1037/0033-2909.132.3.354
3. 3.
   Chi, M. T. H., de Leeuw, N., Chiu, M.-H., & LaVancher, C. (1994). Eliciting self-explanations improves understanding. Cognitive Science, 18(3), 439–477.
4. 4.
   Ebbinghaus, H. (1885). Memory: A Contribution to Experimental Psychology.
5. 5.
   Roediger, H. L., & Karpicke, J. D. (2006). Test-enhanced learning: Taking memory tests improves long-term retention. Psychological Science, 17(3), 249–255.https://doi.org/10.1111/j.1467-9280.2006.01693.x
6. 6.
   Sweller, J., & Cooper, G. A. (1985). The use of worked examples as a substitute for problem solving in learning algebra. Cognition and Instruction, 2(1), 59–89.https://doi.org/10.1207/s1532690xci0201_3