And? This is like MapReduce, which scales well for certain problems, particularly those involving counting. If I am counting up the words in 25 TB of text in one case and counting up the words in 1250 TB of text in another, but in the first case I have 25 servers available to me and in the second case I have 1250 servers. How long does it take given in both problems each server is counting up the words in 1TB of text? Does it take 50 times longer to count up the words in the second problem? No, because there are 50 times as many servers. There are other issues which make the Indian elections complex, but counting the votes isn't one of them, and neither the counting nor the other issues mean that voting machines are "pretty much required" as was claimed.