Learning analytics established that learning environments produce measurable phenomena. It did not establish what those phenomena mean, how they relate, or what an institution can legitimately conclude from them. Those questions are open.
We treat them as research questions rather than product questions. That means stating what we do not know, publishing the limits of a finding alongside the finding, and being specific about the difference between what a system observed and what it inferred.
Signals become patterns. Patterns become intelligence. Intelligence reveals paths.
What can be responsibly anticipated about learning, and what cannot?
Which naturally occurring signals actually carry meaning?
How does understanding actually form inside a real environment?
What does it mean for an institution to understand itself?
Which decisions in a learning system carry the most leverage?
How should learning signal be governed, and by whom?
Where should authority sit between a system and a person?
We are interested in working with institutions, faculty, and researchers examining the same questions from different directions.
Every claim we publish carries what it does not support. A research programme that only reports what worked is a marketing programme.
A model that cannot explain itself to a professor has not earned a place in their classroom, regardless of its measured performance.
Models trained on historical outcomes learn historical inequity. This is examined continuously, not assumed away at design time.
The value of a modeled risk pathway is whether it changed. Predictive accuracy alone is the wrong success criterion in education.
We are looking for institutions and researchers willing to examine these questions seriously, including the parts where the honest answer is that we do not yet know.
