Writing
How we build, what the evidence says, and what we are still guessing about. Every piece says which of those it is.
- The word we will not write
A student arrives having already been described in terms of what they lack, hundreds of times, in writing. We decided our software would not add to the pile.
- A null result worth reading twice
A pre-registered trial of personalised instruction found no average effect. Split the same data by where students started, and something else appears.
- What you hand the parent
Most tutoring produces no artefact. The lesson happens, the hour ends, and what changed exists only in the tutor's memory.