Free State Project Vote Count, Certification Process, and Results
6507 Jester Blvd. Suite 511
Austin, Texas 78726 USA
Free State Project Vote Count, Certification Process, and Results
Alan R. Weiss, Chairman and CEO of Synchromesh Computing and ECL, LLC
Free State Project Vote Count , Certification Process, and Results
Its actually an interesting problem: how do you assure that a healthy, contentious, very public vote of a private organization dedicated to liberty and freedom goes smoothly? How do you make sure that all the votes are counted according to a publicly-available process, that the votes are counted fairly and honestly, and that the entire process can be certifiable (that is to say, repeatable and trustworthy)? Can you establish the creation of free-market solutions to what was previously the purview of Government (often-times poorly), counting votes?
When ECL volunteered to create a "certifiable process" for counting The Free State Project's votes, Free State President and Founder Jason Sorens asked the Author, innocently enough, if we'd also be willing to count the ballots. We agreed, and together we decided to make history.
ECL, the EEMBC Certification Laboratories, has had over 5 years of experience creating and executing benchmark certifications for microprocessors, digital signal processors, and micrcontrollers as well as a operating systems and software tools. As the certification company for an industry-standard consortium of almost 60 semiconductor and software companies, all ardent competitors to each other, we have the sort of background you need to be able to create certifiable processes. With a rigorous background in engineering, a charter and mission explicitly stating fairness and honesty, trustworthiness and equality of treatment, ECL has successfully certified hundreds of benchmark scores. In the semiconductor industry, the results of benchmarking can, at times, be worth literally hundreds of millions of dollars in sales, so a lack of guile is considered necessary, to say the least. Companies, and individuals, trust ECL, and for our part we have never had our fairness questioned or in dispute.
The first step, obviously, would be to establish a Certifiable Process. If you want to know who won, however, you can jump to Page 5 and find out.
Creating the Certifiable Process - and the Results of the Process
The key would be to write up a process that included the following attributes, and publish that on the Free State Project website. The membership would have to "buy in" and trust it.
The Vote Count and Tally must be repeatable to a level of only 2 defects in the entire vote count. Furthermore, the actual ballots will be preserved for the future so that a vote count can be done at any time.
During ECL's Quality Assurance procedures, we sampled over 600 ballots and found only one, very minor error (which was immediately corrected). We estimated that there might be as many as five errors in the vote count, which we knew would not affect the outcome (but each defect would, of course, have to be corrected immediately).
In fact, our defect count showed that, after completing voting, there were essentially no defects in the vote count. After posting the double-checked database to the FSP website (to allow individuals to verify their own vote count once they entered in their FSP member number and their last name), not a single person contact ECL or the FSP, and reported that their vote was counted incorrectly.
ECL will make copies of all votes, and also scan them, providing a permanent record. Copies will be stored both on-site and off-site. ECL's physical security system is very secure, and has passed scrutiny of companies that have billions of dollars at stake.
In practice, we tried very hard to quantify the amount of work required to:
a.) open the envelope or retrieve the fax, or email.
b.) input the vote itself
c.) input the demographic information
d.) calculate and process any money donated (in fact, ECL processed over $13,000 and turned every federal reserve note, check, money order, silver and gold warehouse receipt, and other form of currency to the Free State Project).
We found that we could process a ballot at a rate of 1 per 2 1/2 minutes, but that scanning each and every ballot would add at least 5 minutes to the process. We decided, in consultation with the FSP Board of Directors, to instead do the following:
a.) Make a physical photocopy of each ballot, and store them off-site.
b.) Commit the spreadsheet database where votes were recorded to a Concurrent Version System (CVS) often used in software engineering, so that versions of the spreadsheet could be retrieved at any time.
c.) Back up the data every day to a second machine (a server), and burn a CD-ROM as well every two days.
If anyone questioned their vote, we'd have at least five copies (two paper, three electronic) and could always scan and send via email their ballot at the time of challenge. In practice, this was never required.
Checkpoints of the vote count showed a complete absence of defects as well, and we could retrieve any arbitrary set of ballots, double-check them, and calculate any defects. There were none, which we attribute both to good processes as well as the diligence of our ECL Free State Project Coordinator Erin Decatur Silkenson, a dedicated worker with a Bachelors Degree in Economics from Southwestern University in Georgetown, Texas. Erin was used to dealing with large amounts of data. Background checks on Erin showed she was reliable, dedicated, and amazingly tolerant of people's curious penchant for wanting to convey their opinions to the Free State Project itself by somehow forwarding it, along with their votes and occasionally money, to ECL.
At any time, the FSP Directors (or invited guests) can statistically sample the vote count themselves, and at any time the FSP can request a list of who has voted. A statistical sample should yield at most 1 error in 2500 votes. In addition, we will send all of the ballots to the FSP. ECL will do a cross-check after they have audited and compare our results. The results must match 100%. ECL will employ technology to make sure the ballots we have sent to the FSP are the same as the ones we counted.
ECL only received one request to come audit the vote count, and that person (also located in Central Texas) decided that she was too busy to actually come watch us open envelopes and enter text into the spreadsheet. The best way to audit is to have each member be able to double-check their own votes online after voting was completed, and FSP Information Services expert Matt Cheselka put the database online with alacrity after ECL finished the final vote tally and quality assurance procedure, certifying the vote.
Fair and Honest
Each ECL staff member having access to the votes (and it will only be two people) will sign affidavits and have them notarized of our vote count, honesty, and that we followed this CP.
Only two people actually had access to the votes, the spreadsheets, and the money that came in: Erin, and ECL / Synchromesh Computing Chairman and CEO Alan R. Weiss. Neither Erin, who is not a member of the FSP, nor myself (who joined rather late in the Project's history) had any axes to grind, and by profession both of us were intensely interested in what could only be described as "the truth." Economics is a profession that rests on hard data, and Benchmarking Certification by definition is designed to ward off corruption and report "the real results." Erin's family was from Ohio and New York, and Alan, born in NYC but raised in California, moved to Austin Texas 11 years ago. None of those states were on the selection list, and as Alan put it in a message to the Free State Project membership, "all y'all look alike - all cold weather states!"
ECL/Synchromesh Computing hereby certifies the results of the Free State Project balloting and voting process as fairly conducted, results honestly polled, providing fair access to FSP members, and repeatable under all circumstances.
By This Seal, ECL Certifies the Vote Count as Accurate
Balloting Process, Timetables, and Turnout
If there was a flaw in the process, it was that ballots were mailed late to many members (being sent via US Postal Service Third Class instead of First Class from Henderson, Nevada). Combined with other delays, it truncated the voting time period by a couple of weeks, which in theory should not have mattered a great deal since the voting interval was still over a month. In practice, it caused some confusion, and furthermore the announcement schedule was fixed as a hard and fast end-date for Press relations reasons.
ECL was concerned that if a lot of members waited till the very last minute, the incoming flood would have proved to be uncertifiable given a "hard stop" date. In actuality, this was avoided because about half the members voted (reducing the incoming flow considerably) and members reacted with (mostly) timely responses.
Ballots were mailed out, and could be returned via US Mail Postal Express, Federal Express, Airborne Express, UPS, or other common carrier. The number of ballots returned by these means, costing each member a few dollars each, was staggering - well over 200 came in that way, almost 1/10th of the received vote. This showed, clearly, that of those that voted, they really cared about making sure their vote came in on time, and was counted. Because of the initial ballot dissemination snafu's, ECL decided to accept a fax of the ballot, or an Adobe Acrobat (tm) .PDF file. Later, ECL agreed to accept a JPEG file as well. In practice, about 1/10th of the ballots came in using these electronic methods (and towards the end, a higher percentage.
5000 ballots were ultimately mailed to FSP members, and ECL's final count of 2388 constitutes a return of 47.7%. While its tempting to be depressed about that, we believe the following factors are important to keep in mind, observationally:
· A government-sponsored vote (for example, an election or propositions) that saw almost a 50% turnout would be considered extremely newsworthy and be deemed a "very successful election."
· The Free State Project has been in existence for awhile, and doubtlessly a number of people had moved, didn't leave a forwarding address, or otherwise lost contact with the FSP.
· A certain percentage of people, realizing that signing up is fairly easy (although they had to make a Pledge), voting was quite another matter and might constitute even more of "a contract." This fear of really committing is to be expected in any movement that asks its members to sell their home, quit their jobs, pick up their lives, and move them to a state that may be quite alien or foreign to them and then "get to work" setting their lives back up as well as working for liberty and freedom and democracy. Given the magnitude of the basic decision, it is utterly astonishing that almost 50% even returned their ballots at all.
· The average dollar figure donated to the FSP was over $5 per voter, and would have been much higher had the FSP not encouraged people to fax or email in their vote (of course, the primary purpose of the vote was not fundraising, but rather to vote. ECL fully concurred that was most important). Further, the number of people calling and sending email to ECL to verify that their votes were cast and recorded was impressive, as were the number of votes returned by expensive common carrier rather than simply the US Postal Service regular first class mail. Those that voted, were very serious about their voting and it can be said with clarity that no one took it lightly.
The answer to "which state is the Free State" is New Hampshire. New Hampshire not only beat 2nd place Wyoming by over 250 votes using the Condorcet Method, it also won if you just weighed "the number of first place votes granted to a state."
As Jason Sorens commented, "One interesting factoid is that preferences are very stable and
"well-behaved": not only is there a Condorcet winner over the entire 10 candidates, but if you eliminate the Condorcet winner sequentially, there is a Condorcet winner at each iteration, yielding the final ranking:
- New Hampshire
- South Dakota
- North Dakota
As the following table shows, New Hampshire received 251 more first place votes, and 15 more 2nd place votes, than runner-up Wyoming.
# of 1st's
# of 2nd's
The total number of first and second place votes is greater than the total vote count because the FSP allowed people to vote for more than one first or second (or any place) entry. Many people, for example, gave a particular state a "one" and gave two or more states a "two", and sometimes gave many states a "ten" (indicating no interest at all in selecting that state).
Interestingly, it appears that the so-called "western Libertarians" divided their votes between Wyoming, Montana, Idaho, and Alaska, leaving the so-called "eastern Libertarians", who were much more unified in their first selection, clear access to win with their favorite, New Hampshire. If you combine, for example, the first place votes of Montana and Wyoming, you get 242 + 498 = 740, which would not have been enough to topple New Hampshire. But if you notice the spread between Idaho, Montana, and Alaska, it is much closer than the spread between New Hampshire and the next most popular eastern state, Maine. Montana seemed to be a very popular 2nd place choice, beating even New Hampshire as the second-favorite state. Clearly the industriousness, organization, and marketing of New Hampshire had some effect, though, because it won, and it was also a popular second place choice, even amongst so-called "western Libertarians."
The Members of the Free State Project have spoken, and rather clearly at that, in their selection of New Hampshire as The Free State. It may be the case that this particular project spawns a second effort to select a western state (or even a western province of Canada) as a relocation settlement.
If this occurs, it will not be due to any fraud or abuse of the process during voting, but rather because some liberty-minded individuals decide, quite simply, they prefer a western environment to establish a new libertarian society. If so, Synchromesh Analysts would be honored to again conduct the vote (if there is one) and certify the results.