Comment by pinkmuffinere

At the very least I'm confident it's a power-law distribution. 90-9-1 seems like a reasonable first-order estimate (but I've also seen no data)