diff --git a/ndsl/testing/comparison.py b/ndsl/testing/comparison.py
index ab4d881d..2feea615 100644
--- a/ndsl/testing/comparison.py
+++ b/ndsl/testing/comparison.py
@@ -70,17 +70,19 @@ def _compute_errors(
         near_zero,
     ) -> npt.NDArray[np.bool_]:
         if self.references.dtype in (np.float64, np.int64, np.float32, np.int32):
-            denom = self.references
-            denom[self.references == 0] = self.computed[self.references == 0]
+            # Rule number 1: Never touch the reference data!
+            denom = self.references.copy()
+            # Avoid division by 0. If reference is 0, we expect the computed value to be 0 too.
+            # (abs(computed - reference) / 1.0) is a good value for the error in this case.
+            denom[self.references == 0] = 1.0
             self._calculated_metric = np.asarray(
                 np.abs((self.computed - self.references) / denom)
             )
-            self._calculated_metric[denom == 0] = 0.0
         elif self.references.dtype in (np.bool_, bool):
             self._calculated_metric = np.logical_xor(self.computed, self.references)
         else:
             raise TypeError(
-                f"received data with unexpected dtype {self.references.dtype}"
+                f"Received data with unexpected dtype `{self.references.dtype}`."
             )
         success = np.logical_or(
             np.logical_and(np.isnan(self.computed), np.isnan(self.references)),