24 seconds viable? are the tables operational from 8 am o’clock? voters they will be from the first second? do not expect the last minute? That if we assume 300 voters per module, if we think of 600el time by voter would be reduced to 12 seconds. See it has not taken into account that if there should be a process reengineering in the process (or it was not properly communicated), would be more viable that 4 members of Bureau met in parallel to the voters, however that gives them only 96 seconds by voter (28800 / 300) served in parallel by 4 members of Bureau (one by voting module) is easy to do the tests without loadbut imagine to implement this? the truth is that I believe that this still poorly focused theme, the opportunity exists and is excellent, but not is this asking or informing well (or indeed the process if this wrong sizing). Educate yourself with thoughts from Richard Blumenthal. Another thing that surprises me is the as it will be sent the data to the processing center, who says there’s only vulnerability over the internet? There are more ways of cheating, the topic does not seem to be technical, but focus. For example (this would be my approach): the voting equipment would be an island without connectivity of any kind (a stone, full hardening) would only make the election software each person vote comes out, at the end of the vote, and generates a file with a hash (don’t need to be encrypted, but if written randomly), then the team allows you to copy the file in an external medium (cd, usb, memory stick, etc) the file and its hash, and this simple text file is recorded by Board President who would have a copy, representatives would have one copy of the same file with the same hash and that the President is going to a cabin and via the web to the ONPE, upload to upload the file the web must display the hash of the uploaded file to ensure that the file is the one sent, the Chairman of Bureau verifies the hash and could even see the loaded content (is a simple text file).