Peer Cross-Examination
Why This Matters: James and the Borrowed Connections
James was reviewing his cascade map one last time before the peer exchange. Three drafts. Dozens of arrows. Feedback loops he could explain from memory and connections he'd added from AI that still felt borrowed.
"Nervous?" Emma asked.
"Not about the parts I built myself. Those I can defend. But some of these AI-sourced connections..." He tapped the regulatory branch. "AI added a whole section on international regulatory precedent. I included it because it seemed important. But if my partner asks me to explain why the European Banking Authority's response matters for this scenario, I'm going to freeze."
"So what do you do about those?"
"Remove them?"
"Or understand them. You have fifteen minutes before the exchange. Pick the two connections you're least confident about. Trace the mechanism yourself. If you can explain why A leads to B without checking your notes, keep it. If you can't, mark it as an open question."
James pulled up the regulatory section. He read the AI's explanation, then closed the document and tried to reconstruct the reasoning from scratch. The European precedent made sense once he thought about it: when one jurisdiction acts, others face pressure to harmonize. He'd seen the same pattern in trade compliance at his old company. One country tightened import standards and three others followed within a year.
"Okay. The regulatory chain holds. I can explain it through trade compliance, which I actually lived through."
"Now you own it. The source doesn't matter. The understanding does."
Exercise 4: Peer Cross-Examination
Layers Used: Layer 3 (Live Defence)
James is about to defend a map full of borrowed connections under live questioning. So are you.
Defend Your Cascade Map
Pair up with another student. Exchange your final cascade maps (Draft 3). In a live 15-minute session, each student must: (a) identify at least 2 missing connections in their partner's map, (b) challenge at least 1 feedback loop as unrealistic, and (c) defend their own map against their partner's challenges. No AI access during the session.
Submit your final cascade map to AI with this prompt: "You are a critical peer reviewer examining my systems thinking map. Identify 3 gaps; missing connections, implausible causal chains, or feedback loops I missed. Then challenge one of my feedback loops as unrealistic and explain why. I will defend my reasoning in writing." Conduct a multi-turn defence and submit the full transcript.
Your written preparation notes identifying gaps and challenges in your partner's map (written before the live session). Your partner's written feedback on your map. A post-session reflection (200 words) answering: What did my partner find that I missed? What challenge was hardest to defend? What would I add to my map based on this exchange?
I just completed a peer cross-examination of my systems thinking cascade map. My partner identified gaps and challenged my feedback loops. Below is my final cascade map, my partner's critique, and my reflection.
Please: (1) Evaluate the gaps my partner identified -- are they genuine gaps or were my original connections actually sufficient? (2) Was my partner's challenge to my feedback loop valid? (3) Based on all the feedback (from AI in previous exercises and from my partner), what are the top 3 improvements I should make to my systems thinking approach? (4) Rate my overall systems thinking development across this entire chapter from Beginner / Developing / Proficient / Advanced. (5) Give me a personalized practice recommendation for improving my weakest area.
My final map:
Partner's critique:
My reflection:
Finally, complete the Thinking Score Card for this exercise: Independent Thinking (1-10), Critical Evaluation (1-10), Reasoning Depth (1-10), Originality (1-10), Self-Awareness (1-10). For each score, give a one-sentence justification.
Discuss with an AI. Question your scores.
Come back when you have your BEST evaluation.
What Happened With James
James sat with his partner's feedback and his own notes spread across the table. His partner had picked the healthcare scenario and attacked James's map from a completely different angle. She'd found a feedback loop he'd missed entirely: the bank's AI system, trained on historical loan data, would encode the biases of the very loan officers it replaced. The system would automate discrimination while appearing objective. That loop connected technology, regulation, and community trust in a chain James hadn't drawn.
"I spent three drafts on this map," James said. "AI reviewed it twice. And a fifteen-minute conversation with someone who picked a different scenario found a gap none of us caught."
"Why do you think that is?"
"Because she was looking at the same type of system from a different entry point. Healthcare triage and bank lending share the same trust dynamics, but I never would have seen the bias loop if she hadn't brought that lens from her scenario."
Emma was quiet for a moment. Then she said something James didn't expect.
"I made a version of your mistake once. Early in my engineering career. We had a database query that was bottlenecking a reporting service. Took thirty seconds to run. I optimized it. Got it down to three seconds. My manager was thrilled. Ten times faster."
She paused. "Two hours later, three downstream services started throwing timeout errors. The reporting service was suddenly pulling data so fast that it was saturating the connection pool that six other services shared. I'd made one node ten times faster and broken the network."
James watched her. She'd talked about getting something wrong before, but it landed the same way each time: not as a confession but as a map. A cascade she hadn't traced before she changed the variable.
"What did you do?"
"I rolled back the optimization. Then I spent a week mapping every service that shared that connection pool, every downstream dependency, every resource contention point. I drew the system before I touched the system. That's when I stopped thinking in components and started thinking in systems."
She let the silence sit.
"Three exercises ago, you looked at a bank decision and wrote three lines. Problem, solution, outcome. Now you've got three drafts, peer feedback, AI comparisons, and a change log. The map got bigger, but that's not the point. The point is that you can see the connections now. You can ask 'and then what?' and actually trace the answer across boundaries."
James looked at his three drafts side by side. The first was a list. The second was a web. The third was a web with scar tissue, where connections had been torn out and rebuilt because the world changed. The difference between the first and the third wasn't sophistication. It was honesty about how many things connect to how many other things.
"Ready for Chapter 4?" Emma asked.
"I think so. But I keep finding new connections in things I thought I already understood."
"That's not a problem. That's systems thinking."
The Lesson Learned
Systems thinking is tested by other humans, not just by AI. A peer who approaches the same problem from a different entry point reveals blind spots that AI, which tends to converge on similar analyses, cannot find. The live format tests whether you understand the connections you drew well enough to explain and defend them. If you can trace the mechanism from memory, you own it. If you freeze, you borrowed it.
A Systems Thinking Portfolio containing: (1) the original cascade map (Draft 1, before AI), (2) the comparison document and merged map (Draft 2, with attribution), (3) the revised map after variable shift (Draft 3, with change log), (4) peer cross-examination notes and feedback, (5) all AI feedback responses, and (6) a final reflection on your systems thinking growth.
Grading Criteria
| Component | Weight |
|---|---|
| Original cascade map depth (feedback loops, multi-domain effects) | 20% |
| Merged map quality and attribution honesty | 20% |
| Variable shift adaptation quality | 20% |
| Peer cross-examination performance (attack and defence) | 20% |
| AI feedback integration (did you improve based on feedback?) | 10% |
| Reflection quality | 10% |