The Score measures improvement over QWERTY using fixed reference points. A score of 0% means the same as QWERTY, 100% means the theoretical best. Negative scores are possible for layouts worse than QWERTY.

Step 1: Normalize each metric

Each metric is normalized to measure improvement over QWERTY, where:

• \( M_{\text{layout}} \) = the layout's metric value
• \( M_{\text{qwerty}} \) = QWERTY's metric value

For lower is better metrics (SFB, SFS, LSB, Scissors, Redirect, Pinky):

$$\text{normalized} = \frac{M_{\text{qwerty}} - M_{\text{layout}}}{M_{\text{qwerty}} - 0}$$

The numerator is the improvement over QWERTY; the denominator is the maximum possible improvement (QWERTY to 0).

For higher is better metrics (Rolls, Alternation):

$$\text{normalized} = \frac{M_{\text{layout}} - M_{\text{qwerty}}}{100 - M_{\text{qwerty}}}$$

The numerator is the improvement over QWERTY; the denominator is the maximum possible improvement (QWERTY to 100%).

Step 2: Apply weights

Each normalized value (already in "higher is better" form) is multiplied by its weight:

$$\text{contribution} = w \times \text{normalized}$$

Step 3: Calculate final score

The weighted sum is converted to a percentage of maximum possible improvement:

$$\text{score} = 100 \times \frac{\sum w_i \times \text{normalized}_i}{\sum w_i}$$

Note: This QWERTY-fixed approach ensures stable rankings — adding or removing layouts won't change the relative order of existing layouts. QWERTY's metrics are language-specific, so scores may vary by language.