NATIONAL ACADEMY OF SCIENCES

Five detectors fired for the National Academy of Sciences. Benford's law analysis found the leading-digit distribution of award amounts significantly deviates from expected frequencies, with digits 1 and 2 appearing more often than natural patterns predict. Modification growth on a single contract vastly outpaced peer norms in the scientific consulting sector. An isolation forest algorithm flagged the entity as a multivariate outlier based on its combination of high dollar volume, award count, low agency diversity, and extremely frequent contract modifications. Sole-source concentration showed the entity receives nearly 94 percent of awards without competition, far exceeding the median of 1.3 percent for peers in the same industry. Award velocity surged sharply in the past 90 days, with 60 recent awards versus a historical baseline rate that would predict only 25.

The worst-performing contract modification grew 387.66 times in value compared to a peer average of 12.61, yielding a z-score of 9.24. The entity received 279 sole-source awards out of 297 total, a fraction of 0.939. Benford's test produced a Kolmogorov-Smirnov statistic of 0.122 with p-value 0.0003 across 297 transactions. Recent award velocity reached 0.67 awards per day versus a baseline of 0.28 per day, with a z-score of 6.89. The isolation forest raw anomaly score was negative 0.744 under 5 percent contamination.

These statistical signals warrant scrutiny but do not confirm fraud. The National Academy of Sciences is a congressionally chartered nonprofit with legitimate advisory and research roles across federal agencies, which may explain high sole-source reliance and modification frequency. However, the Benford deviation and extreme modification growth on a single contract remain difficult to reconcile with normal operations and merit closer examination of contract documentation and justifications.