We show that the sequence of thinned uniform random counting measures converges weakly to the Poisson random measure. We call such measures 'orthogonal dice' due to their natural connection with certain counting problems and link the sizes of orthogonal dice with the primes by constructing the largest known prime orthogonal die. We give many examples of possible use of the construct of orthogonal dice in various areas of applications including gambling, cosmology, random matrices, approximation theory and circuits. The gambling example reveals for instance that fair six-sided die numbered $\{1,2,3,4,5,6\}$ generate negative covariance in point representations of hands across players, in contrast to seven-sided dice numbered $\{1,2,3,4,5,6,7\}$ that generate zero covariance and are orthogonal. The cosmology example suggests that a 'supermassive' orthogonal die underlies the galaxy point patterns of the Universe. The random matrix application reveals that the variance of the spectral gap of the Gaussian orthogonal ensemble exhibits non-monotone scaling with thinning for the Dirac (empirical) random measure, in contrast to the orthogonal dice where the spectral gap variance is monotone. Finally, the approximation application shows that the discrete Legendre polynomials converge to the Charlier polynomials in $L^p$ for integers $p\ge1$ whereas the electronic circuit/shot noise application identifies electric current as the Ornstein-Uhlenbeck process driven by an orthogonal die random measure., Comment: 31 pages, 6 figures, 1 table