Here are the answers to some of Fred's questions: On Sat, Aug 21, 1999 at 11:17:38AM -0700, Fred Smoot wrote: > > What is the name and version of the voting software that was used in our > runoff election? The software has no name or version number. Tim wrote it from scratch between the first election and the runoff, with three goals in mind: accuracy, efficiency and easier voting. It is written in Perl and runs on a Unix server. The Perl scripts are short (they total about 300 lines of code) and are straightforward for a Perl programmer to understand. Tim would not have any objection to posting the source code if it would help allay anyone's concerns about whether it accurately tracks each vote. The `back-end' database software -- the server software that actually kept track of each person's e-mail address and how they voted -- is a server package called MySQL. MySQL is an industry-strength open source SQL database package: more details about it are at http://www.mysql.org/. > Why was the software changed from that which was used in our initial > election. Who made the decision? Who first suggested that the software > be changed? Quote from Tim "The reasons were to benefit everyone (not just the committee but also the candidates and the voters). I didn't like having to make excuses for not running a single-candidate election, when knew perfectly well that better software would make such a thing trivial. I also didn't like running software that was unforgiving about the format of submitted votes, and simply could not process some of the votes that were sent to us. These were the most important reasons (in my mind) for revising the election process." (The Election Committee worked with Tim to debug the system. We tried hard to break it. When we found something that we thought could be confusing or that just didn't work, Tim fixed it. We did this before we announced the new system. We decided it was the greatest thing going and asked Tim if it would be ready to use. The rest is history) > Who, other than employees of RootsWeb.com Inc, actually monitored the > entire voting process? Who was your on-scene teller? There was no `on-scene teller', just a Perl script that confirmed votes and put them into a database. The only things that Tim monitored along the way were the Web server's error logs, looking for any sign of malfunctions in the script. Tim did not monitor the vote submissions themselves, and only looked at how an individual voted when they asked us to confirm their vote manually. > Can you explain how you personally arrived at the election result > numbers (copied below in your original message)? > > What kind of verification did you have in hand to prove that the > election result numbers are correct? Quote from Tim "To do these things, I ran this SQL command against the vote database: SELECT VOTE, COUNT(*) FROM VOTES ORDER BY VOTE" > Who actually holds the all the computer logs of the election? Tim does. vote.rootsweb.com is an alias for his personal workstation. The Web server logs that record each vote that was submitted to us are stored on Tim's machine. Each vote is physically stored on RootsWeb's database server, which is housed in a network center in Anaheim, California. The vote database includes this information about the voter: * their e-mail address * who they voted for * the confirmation code they were sent * the date and time they voted * the Internet host (i.e. ISP) from which they voted * the date and time they confirmed their vote * the Internet host from which they confirmed (Each of the Election Committee Members now holds a copy of email address and their respective vote. We could individually count them. I hope that is not necessary. Which we have now examined and certified to be correct - Jim) > How many unconfirmed votes (not accepted or otherwise rejected, etc.) > were cast in the runoff election? 33 people submitted 39 votes that were never confirmed. There are more votes than people because a few of them submitted two or three unconfirmed votes. During the vote, we added e-mail addresses to the list of eligible voters whenever a SC notified us that someone's address had changed. So some of these unconfirmed votes may have come from people who subsequently voted successfully using their new address. In any event, we have examined the vote counts for these 39 votes and confirmed that these votes break down roughly the same way as the confirmed votes. (The committee did check the totals. Even though we believe they can not be added, because they were not confirmed. We would like to let you know that the percentages are roughly the same in the favor of the same candidate. Jim) The Election Committee