A system should learn from reality, not agreement
Every decision leaves behind evidence. What was understood at the time, what was decided and what followed can all tell an organisation something about itself. But the decision itself is not evidence that the reasoning was right. Protu keeps the two separate: what people chose becomes part of the record; what followed becomes new evidence against which that understanding can be reconsidered.
Outcomes are evidence, not truth
What follows a decision is never a perfect verdict on the decision itself. Organisations change, people move, circumstances shift and an outcome rarely has a single cause. Later evidence can strengthen an earlier understanding, weaken it or raise a different question entirely. We treat it as another piece of evidence, not proof that Protu was right or wrong.
The harder problem is learning the wrong thing
If a system learns from what an organisation chooses, it can slowly learn to reproduce those choices. An organisation that repeatedly hires externally could teach it that hiring externally is preferable. One that rarely develops people internally could make development disappear from consideration. The system would become better at reflecting the organisation without necessarily becoming better at understanding it.
Patterns should be visible, not mistaken for truth
Repeated choices are still worth remembering. They can reveal patterns in how an organisation tends to act. But repetition does not make a preference evidence that the preference is right. Where the organisation’s history and the behavioural evidence point in different directions, that difference should remain visible rather than one being allowed to quietly overwrite the other.
Every organisation develops patterns. A system worth trusting should help make them visible, not quietly learn to agree with them.
The past should remain revisable
New evidence shouldn’t erase what Protu understood before. It should sit alongside it. That makes it possible to see what changed, what remained consistent and where an earlier conclusion no longer holds. Organisational memory is useful precisely because the past remains available to be questioned.
Learning takes time
One outcome should rarely be enough to change what the system believes. As decisions accumulate and later evidence becomes available, patterns can begin to carry more weight. Contradictions matter too. Learning should be gradual because the evidence behind it accumulates gradually.
A system that learns only from what you choose becomes a mirror. What follows the choice is where the organisation begins to learn something new.
Keep reading