What Did Not Work
What did not work in the Acacia Initiative's pilot programmes and field experiments — the practices, tools, and approaches that did not deliver results.
The Core Argument
Not everything worked. This page documents what did not — the practices, tools, and approaches that did not deliver results.
Success is not the only signal. Failure is also a signal. When something does not work, it tells us something important — about the context, the approach, or the assumptions.
Failure is data.
This page documents what did not work so that we can learn from it. It is a record of our mistakes — and the lessons they taught us.
What Did Not Work
The practices, tools, and approaches that did not deliver results.
Generic Frameworks
What did not work: Applying generic frameworks without adaptation.
Why: Generic frameworks do not capture the specific context, constraints, and relationships of each situation.
Signal: Context matters. Generic solutions fail in specific situations.
Over-Reliance on AI
What did not work: Over-reliance on AI-generated answers without human evaluation.
Why: AI can generate plausible answers, but it cannot evaluate them. Human judgment is essential.
Signal: AI is a tool, not a decision-maker. Human judgment cannot be outsourced.
One-Size-Fits-All Training
What did not work: Providing the same training to all staff without differentiation.
Why: Different roles require different skills and different levels of judgment. Generic training does not address specific needs.
Signal: Training must be tailored to the role and context.
Quick Fixes
What did not work: Trying to implement the Acacia framework as a quick fix.
Why: Judgment infrastructure takes time to build. Quick fixes do not address the underlying problems.
Signal: Building judgment infrastructure is a long-term investment, not a quick fix.
Lessons from Failure
What we learned from what did not work.
Context Is Not Optional
Generic frameworks fail because they ignore context. Context is not optional — it is essential.
AI Is Not a Decision-Maker
AI can generate answers, but it cannot make decisions. Human judgment is essential. Over-reliance on AI is a mistake.
Training Must Be Tailored
One-size-fits-all training does not work. Training must be tailored to the role and context.
Patient Implementation Is Required
Quick fixes do not work. Building judgment infrastructure takes time. Patient implementation is required.
The Acacia perspective:
What did not work tells us what to avoid. The signal is clear: context matters, AI is not a decision-maker, training must be tailored, and patient implementation is required.
Reading path: