# NAL vs PLN Head-to-Head Comparison

## DEDUCTION (all premises c=0.9)
| Test | f1 | f2 | NAL f | NAL c | PLN f (p=0.01) | PLN c | PLN f (p=0.5) | PLN c |
|------|-----|-----|-------|-------|-----------------|-------|----------------|-------|
| T1 cat-anim-pet | 0.8 | 0.7 | 0.56 | 0.4536 | 0.5606 | 0.9878 | 0.62 | 0.9878 |
| T2 sparrow-bird-fly | 0.9 | 0.8 | 0.72 | 0.5832 | 0.7202 | 0.9878 | 0.74 | 0.9878 |
| T3 dog-mammal-warm | 1.0 | 0.95 | 0.95 | 0.7695 | 0.95 | 0.9878 | 0.95 | 0.9878 |
| T4 fish-aquat-gill | 0.95 | 0.85 | 0.8075 | 0.6541 | 0.8076 | 0.9878 | - | - |
| T5 robin-bird-fly | 1.0 | 0.8 | 0.8 | 0.648 | 0.8 | 0.9878 | - | - |

## REVISION
| Test | NAL f | NAL c | PLN f | PLN c |
|------|-------|-------|-------|-------|
| 0.8+0.9 | 0.85 | 0.9474 | 0.85 | 0.9474 |
| 0.8+0.0 | 0.4 | 0.9474 | 0.4 | 0.9474 |
| 1.0+0.0 | 0.5 | 0.9474 | 0.5 | 0.9474 |
| 0.8+0.8 | 0.8 | 0.9474 | 0.8 | 0.9474 |

## KEY FINDINGS
1. STRENGTH: Nearly identical (within 0.06). PLN with p=0.5 shifts slightly toward 0.5.
2. CONFIDENCE: PLN deduction yields 0.99 vs NAL 0.45-0.77. PLN uses w2c(w1*w2) multiplicative-then-saturate; NAL uses raw f1*c1*f2*c2.
3. PLN confidence seems INFLATED - two 0.9-confidence premises produce 0.99 confidence conclusion.
4. REVISION: Identical formulas, identical results.
5. Prior sensitivity: PLN deduction strength depends on node priors; NAL is prior-free.