This paper examines the general percolation problem of cutting randomly centered insulating holes in a two-dimensional conducting sheet, and explores how the electrical conductivity sigma decreases with the remaining area fraction. This problem has been studied in the past for circular, square, and needlelike holes, using both computer simulations and analog experiments. In this paper, we extend these studies by examining cases where the insulating hole is of arbitrary shape, using digital-image-based numerical techniques in conjunction with the Y- [nabla] algorithm. We find that, within computational uncertainty, the scaled percolation threshold, xc=nc=5.9±0.4, is a universal quantity for all the cases studied, where nc is the critical value at percolation of the number of holes per unit area n, and is a measure of nI-1, the initial slope of the sigma (n) curve, calculated in the few-hole limit and averaged over the different shapes and sizes of the holes used. For elliptical holes, Leff=2(a+b), where a and b are the semimajor and semiminor axes, respectively. All results are well described by the universal conductivity curve: sigma / sigma 0=[(1-x/5.90)(1+x/5.90-x2/24.97)(1+x/3.31)-1]1.3, where x=nLeff2, and sigma 0 is the conductivity of the sheet before any holes are introduced.