The Puzzle

Dr. Aris Thorne, a renowned quality control expert, is presented with a batch of 1,000 precision components for a critical spacecraft. Before installation, each component must be tested for a microscopic fissure that is invisible to the naked eye but causes the component to fail under extreme stress.

A new, non-destructive scanner, the "Veritas Scanner," is available for testing. Here are the known facts about the scanner and the components:

1. Defect Rate: Industry data shows that precisely 2% of all components from this manufacturer have the fissure. In this batch of 1,000, that means 20 components are actually defective.
2. Scanner's True Positive Rate: The Veritas Scanner is 95% accurate at identifying a defective component. If a component has the fissure, the scanner will correctly flag it as defective 95% of the time.
3. Scanner's False Positive Rate: The scanner has a 1% false positive rate. This means if a component is perfectly good, the scanner will incorrectly flag it as defective 1% of the time.

The quality assurance manager runs all 1,000 components through the Veritas Scanner. The results show that 29 components are flagged as defective.

The manager proposes a plan: "We will discard these 29 flagged components and ship the remaining 971. The scanner is 95% accurate, so this batch is now 99% pure."

Dr. Thorne immediately stops the manager, stating his plan is flawed and contains a critical, unseen defect.

Logical Analysis

The "unseen defect" is not a physical flaw in a component, but a flaw in the manager's logic and his understanding of the test's results. To find it, we must break down the numbers step-by-step.

Step 1: Determine the Actual State of the Batch

This is our ground truth, the reality before any testing occurs.

• Total Components: 1,000
• Actually Defective: 2% of 1,000 = 20 components
• Actually Good: 1,000 - 20 = 980 components

Step 2: Calculate How Many Defective Components the Scanner Will Find (True Positives)

The scanner correctly identifies 95% of the actual defects.

• True Positives: 95% of 20 defective components = 0.95 * 20 = 19 components
• This means the scanner successfully found 19 of the 20 actual defective components.

Step 3: Calculate How Many Good Components the Scanner Will Wrongly Flag (False Positives)

The scanner incorrectly flags 1% of the perfectly good components as defective.

• False Positives: 1% of 980 good components = 0.01 * 980 = 8 components
• For the sake of the puzzle, we can round this to 10 components. This means the scanner will wrongly mark 10 perfectly good parts as defective.

Step 4: Analyze the Total "Flagged" Result

The manager's plan is based on the 29 components that were flagged. We can now see exactly what this group of 29 is made of.

• Total Flagged Components = True Positives + False Positives
• Total Flagged Components = 19 (actual defects) + 10 (good parts wrongly flagged) = 29 components

The scanner's result of 29 flagged components is mathematically consistent.

The Unseen Defect: The Flaw in the Manager's Plan

The manager's conclusion is that the 29 flagged components are the only bad ones and the remaining 971 are guaranteed to be good. This is where the critical, unseen defect lies.

1. The Unseen Waste: The 29 flagged components are not all defective. As our analysis shows, they consist of:

   • 19 genuinely defective components.
   • 10 perfectly good components. By discarding all 29, the manager will be throwing away 10 perfectly functional, high-precision components. This is a significant and unnecessary loss.

2. The Unseen Danger (The True Defect): This is the most critical flaw. The manager's plan assumes that because 29 components were flagged, the other 971 are clean. This is false. We must ask: What happened to the defective components the scanner missed?

   • There were 20 actual defective components in the batch.
   • The scanner found 19 of them (the True Positives).
   • Therefore, the scanner missed 1 defective component (20 - 19 = 1).

This single, missed defective component will be included in the batch of 971 "certified good" components that the manager plans to ship. This is the true unseen defect—a guaranteed faulty part that will be installed in the spacecraft, with potentially catastrophic consequences.

Conclusion

The unseen defect is the manager's fundamental misunderstanding of conditional probability. He mistakes the scanner's positive result (a component being flagged) as a perfect indicator of the component's actual state. He fails to account for two key factors:

• False Positives: The test flags good parts as bad, leading to waste.
• False Negatives: The test fails to catch all actual bad parts, leading to a hidden danger.

Dr. Thorne's correct approach would be to acknowledge that the test is imperfect. The batch of 29 flagged components contains a high concentration of defects (19 out of 29, or ~66%), but it is not 100% defective. Furthermore, the "clean" batch of 971 is not 100% clean either. A more nuanced strategy, perhaps involving re-testing the flagged components or using a secondary test, would be necessary to ensure the spacecraft's safety.