At a technical panel, researchers and practitioners debated whether forensic experts should present Bayesian likelihood ratios in court, focusing on juror comprehension, model uncertainty, instrument differences and the need for clear documentation and empirical validation.

Dr. Song of NIST presented a method called congruent matching for forensic image correlation and reported validation experiments (including 95 cartridge cases and 4,465 image pairs) he says yield cumulative false positive and false negative error rates of less than 1 in 1,000,000. He described workflows for cartridge-case, firing-pin and bullet‑profile comparisons and said the approach could support federal ballistic identification efforts.