TruLearn

The research

Why it teaches the way it does.

Four decisions about how TruLearn teaches, each taken from research that has held up for decades. Here is what each study actually found, and where it stops.

Principle 1

It makes you explain it back

Every concept ends with the student saying it in their own words. Students who explain material to themselves while studying learn substantially more than students who read it twice, because explaining forces the gaps into the open where they can be fixed.

Where the evidence stops

Measured on physics and biology texts with small groups of students. It says explaining beats rereading; it does not say by how much for any given child.

Chi et al., self-explanation and learning, Cognitive Science (1994)

Principle 2

It asks before it tells

Being asked to retrieve something from memory does more for retention than studying it again for the same length of time. So a session opens by asking, review questions come back as questions, and the tutor waits rather than rescuing.

Where the evidence stops

The advantage shows up on delayed tests. On a test taken minutes later, rereading can look just as good, which is exactly why cramming feels like it works.

Roediger and Karpicke, the testing effect, Psychological Science (2006)

Principle 3

It brings things back later, on purpose

Practice spread over days beats the same practice massed into one sitting. What a student missed returns on a schedule rather than when they happen to think of it, in a question shaped to catch that specific gap.

Where the evidence stops

This review rates distributed practice and practice testing as the two highest-utility techniques, and rates highlighting and rereading as low utility. It compares techniques; it does not rate any particular product.

Dunlosky et al., which study techniques work, Psychological Science in the Public Interest (2013)

Principle 4

It teaches one student, not a room

Bloom's students taught one to one with mastery requirements performed about two standard deviations above conventionally taught students - roughly the top few per cent of a normal class. One-to-one attention is the strongest intervention in the literature, and the one schooling cannot afford to give everybody.

Where the evidence stops

The two-sigma figure comes from small studies with human tutors, and later work has rarely reproduced the full effect. We cite it as the reason one-to-one is worth automating, not as a promise about a score.

Bloom, the 2 sigma problem, Educational Researcher (1984)

What we have not proved

We have no outcome data of our own

TruLearn is in private beta. We can say what it does and why that design follows from the research above. We cannot yet say what it does to a student's grade, and we would rather say so than imply it. An external measure - AP exam results for students who learned with us - is the bar we intend to be judged on.

Until then, the only evidence we offer is per-student and specific: which concepts a learner has explained back correctly, against a rubric a person wrote. See how that grading works.

What TruLearn does not do yet

  • One course. AP Biology is fully authored, with a college introductory biology version of the same library; other subjects are being written.
  • No parent dashboard. Coverage maps, streaks and per-session digests are planned and not built.
  • No transcripts or grades you can submit anywhere. What exists is evidence of what has been explained back correctly.
  • It does not check written work step by step as it is being written, and it will not do a problem set for a student.
  • No mobile app. It runs in a browser, and a tablet with a stylus is the best experience for the whiteboard.

More on the learning science

All articles

The future

Will AI replace teachers?

No, and the reason is not sentimentality. Most of a teacher's job was never the explaining, and the explaining is the only part that is close to solved.

5 min read

Tutoring

What the research says actually makes tutoring work

Tutoring is one of the best-evidenced interventions in education, but the effect depends on who tutors, how often, and when. Most purchased tutoring misses on frequency.

5 min read

Homeschooling

How to homeschool a subject you never learned yourself

You do not need to know the subject to teach it well. You need a source that does, a way to check understanding, and the honesty to say when you are lost.

8 min read

Join the beta.

Create an account and we'll activate it as places open. Free during the beta.