How well intentioned requirements paved the road to election hell
By Jim Willard, President, Votation Corp.
Table of contents
Washington, DC, as a case study
Research generated surprising results
Defective procurement requirements behind the problems
The problems with parts
Reducing competition by requiring prior use
Can we get beyond today's model?
Big firms opt out
The immediate future remains bleak
About the author
Now everyone knows that, while "every vote counts," not every vote gets counted.
The burning questions of the day include: In this day of computers and the Internet why can't we build a good voting system? Why is voting the only operation still using punch cards? Why aren't these systems accurate?
It is not technology, or the lack of it, that caused our recent problems. Rather, a defective voting-business environment brought us to the brink of political disaster. Well-intended procurement requirements have served to ensure that the modern equipment desired and needed for public elections will never reach the voting public.
Modern life is full of personal and business automation and, by and large, most of it works pretty well — mostly because billions of dollars have been spent on making it work well. We generally expect to have a good outcome from the technology we use every day such as our cars, televisions, and telephones. Certainly, you would not use a bank that randomly lost deposits, miscalculated your funds, or couldn't determine from their records whether or not you were a customer.
But would you enter into a business relationship in which you knew from the outset that the product you were buying was defective? That the product didn't do what it was supposed to do? That the product would be hard to operate and expensive to maintain? Would you buy a product knowing full well that once the sale was complete you may never see that salesman or his company again and would have few options if any, for warranty claims, repairs, parts, and service?
Welcome to the world of voting-system procurement where such a defective business relationship is the only relationship available. Practically everyday, some election administration somewhere is purchasing a voting system knowing full well that "it doesn't count all the votes."
Washington, DC, as a case study (Back to top)
Witness the Washington, DC, Election Commission recently announcing that they just purchased a new "optical-scan" voting system for the District of Columbia at the cost of $1 million. The District is replacing their antiquated punch-card voting system with "new" but equally antiquated optical-scan technology. If you followed the evidence presented in the Gore v. Bush Florida court testimony, you would know that the District has improved their ability to count (some) of the votes, but they still will not be counting all of the votes. The fact is that the District Government has purchased a system with a documented failure (of its one and only limited task) of 1 to 1.5 percent.
What is the problem here? Can you believe that the DC Election Commission got together and said, "Hey, let's spend a million dollars to replace our defective voting system with another defective voting system." Why would they buy that system? In the District's case and throughout America, election officials right now are buying, or preparing to buy, defective voting machinery because they have no other choice available. Even the new computerized voting machines have the potential of failure because they do not have the quality system design required to absolutely guarantee the truth of the votes collected.
Election commissions, politicians, and the voting public at large today are stuck with clearly unsatisfactory and actually defective vote-counting systems because well-intended system and procurement requirements drove the business of supplying voting equipment to hell and kept it there. Those requirements actually precluded the very equipment development and technology advancements needed.
Research generated surprising results (Back to top)
My in-depth research and engineering into the world of elections and voting machines began in 1988 as part of my defense company's efforts to locate business opportunities for mission-critical applications in areas other than our main business of developing naval weapons systems. Our initial look at the voting business produced astonishing information. Here was a business process virtually devoid of automation from the back-room operations to the actual customer interfaces (voting machines) in use. There was not even one major supplier in the business and virtually no technology companies were involved in providing voting machines. Like many people, we imagined that there was a complete automated infrastructure in elections and that it was all efficient and accurate. We were amazed to find that such was not the case in 1988. Now in 2000, the world shares our amazement.
Our research found that there were major flaws in the machines and devices being used for voting. We also turned up a report by Roy Saltman at the National Institute of Standards and Technology (NIST) that confirmed what we had discovered independently. The Saltman report (NBS Special Publication 500-158) showed the deficiencies in voting systems right down to a documented election disaster in Palm Beach County, Florida, almost identical to the recent problems there – only the names have changed.
If governments knew as early as 1988 that the voting machinery in use was defective, why was this problem untended until now?
Defective procurement requirements behind the problems (Back to top)
The answer we found was in the business environment. A business environment that evolved over a number of years and is furthered till this day by a set of defective procurement requirements. These business-related requirements commonly published as part of voting system (machine) procurements actually tend to exclude the advanced systems and technology that "we all thought existed already." These business requirements also explicitly mandated a defective business relationship between the seller and the buyer.
When we began interviewing election officials they were in full and vocal agreement about their problems. The voting machine business was defined, they said, by a set of small companies that they characterized as very undependable, unreliable, and available "only when they were ready to sell you something." By and large the comments regarding the salesmen were uncharitable at best. Further complaints: Service — "forget it." The voting machines — "junk". "They don't count all the votes, they frequently have troubles with the software counting programs, and there is nothing new on the horizon to buy."
As we looked closer at the procurement process, we quickly saw that the business-related requirements in the requests for proposals were the root of the problems. The buyer's own specifications were ensuring the non-availability of the very systems they wanted. Requirements such as:
- The machine must have a service life of at least 20 years.
- The machine must be capable of being serviced by county employees without assistance from the vendor.
- The machine must have a paper backup (i.e.; paper ballots).
- The machine must be capable of accepting (paper) ballots from any supplier.
- The machine must have been used in a public election elsewhere.
- The supplier must provide a performance bond.
Assessing these requirements from a business perspective one can readily see why the voting business failed to keep pace as the rest of our society moved to more and better automation.
The problems with parts (Back to top)
In the late 80s and early 90s, every machine requirement that we saw carried a mandatory service life of 20 years. Electronic systems and machines may indeed last 20 years … but parts don't. It is not practical to guarantee a customer that you can have a ready supply of electronic components available for twenty years. Missed from this consideration was the fact that electronic systems should cost less and therefore very-long-term-use requirements would not produce the economic benefit intended. What this requirement did achieve was the total elimination of any computerized or electronic systems because the only way to meet it was to propose a mechanical system that (hopefully) required little maintenance. By the mid-1980s, the only companies that would offer machines to meet this requirement were the punch-card suppliers.
As best we could determine, this requirement came from a vendor that sold the large mechanical voting machines that were advertised to last at least twenty years. It was adopted by a number of jurisdictions as a "good" (well-intended) standard and incorporated into new voting-machine requirements. What's ironic is that this vendor established a set of requirements in the industry that led to its own demise — the vendor went bankrupt waiting for the next 20-year round of voting-machine purchases.
The tendency of vendors to disappear between elections largely motivated another unfortunate requirement: that voting machines must be independent of vendor support after the sale. Maintenance, parts, and supplies had to be "easily obtainable from a variety of sources." The unwanted effect of this requirement was to further lead buyers to deal with an ever-shrinking circle of smaller and smaller companies that had less and less reason to establish any kind of reasonable support services for the products sold. The stated objective of this requirement was "so that we don't get locked into any one vendor" for ballot production and follow-on support. This practically required the seller not to offer follow-on support and has resulted directly in one of election official's biggest complaints — lack of vendor service and support after the sale.
Reducing competition by requiring prior use (Back to top)
Another requirement that made sure that the door was closed to new development was the requirement that "the proposed machine must have been used in a public election elsewhere." This absolutely killed any chance of new technology ever seeing the light of day in the voting machine business. The original idea for this requirement came from voting machine salesman; it was meant to exclude new competitors and new systems from entering the market. It apparently was easy to convince election officials that this was a "self-protection" requirement that assured them that they would only be buying a "safe" system. Like the other requirements, this one certainly worked … to everyone's detriment.
To get around this problem many elections officials will allow new companies to conduct a "mock elections" or service one of their small local or town elections where the stakes aren't considered as high as electing a governor. While this satisfied the "used elsewhere" requirement, it failed to improve the situation. Even existing suppliers didn't want to spend money to develop new products or they'd be out wasting even more money doing free elections to get around their own anti-competition requirement.
Another requirement typically seen in procurement packages is the "paper" standard. "The system must provide a paper backup for the votes cast so that effective recounts can be conducted." Thus implying a paper ballot system.
Clearly the fallacy of this requirement was seen in Florida and broadcast to the world. It is a false sense of security to believe that we can rely on marked paper ballots to ensure the true results in an election. Both punch-card and optical-scan systems require someone's interpretation to establish the "voters intent." As was seen in the Florida fiasco, "interpretation of the voters intent" is subject to much interpretation of its own.
Can we get beyond today's model? (Back to top)
One of the hardest problems to overcome is the current sales model. In most states, voting-machine purchases are the jurisdiction of county governments and major municipalities. A voting system supplier has to go to each county separately to make sales. This is rather impractical, mainly due to the small size of the voting-machine companies involved, the long lead-times between sales, and the amount of work necessary to achieve even a single sale. What generally happens is that the larger counties get all of the sales attention and the adjacent smaller counties tend to follow along — if they can afford to follow along.
Once it is known that a major county is looking for new machines, all of the sales reps tend to head there. If there is more than one large county in an area looking for machines, sales tend to go to different suppliers. This occurs simply because the salespeople have difficulty working two major sales at once — these small voting companies just don't have the financial means to fund the long-lead-time procurements and work necessary to cover so much territory.
Even if they could put enough salesmen in the field, another requirement raises its head to stop large-scale sales — bonding.
Voting-machine procurements normally have a requirement to post a performance bond. This strictly limits the number of sales that can be made by any one company in a given year. There is just no way a typical small voting machine company can have the kind of capital tied up in performance bonds that would be required to supply an entire state larger than maybe Rhode Island or Delaware.
Big firms opt out (Back to top)
Which brings us to another business issue about the voting industry: Where are the big companies? I've often been asked by potential investors, "If this business has so much potential, why isn't IBM or one of the other big companies in the voting-machine business?"
Well, aside from the above problems with the business operation, there is an even bigger chasm to cross — business risk and the perception and risk of failure.
IBM developed the earliest version of the punch-card system only to exit the business when an election didn't go quite as planned. Unisys was also an early provider but they also exited the business when an election didn't go well. These companies decided that it just wasn't worth the headache and potential public scorn when they had no operational control over the system and couldn't guarantee against its misuse.
I'll submit also that it appeared that both companies approached the business a bit too casually. People who have not researched elections commonly assume it to be a simple business. Such is not the case. Much work is required to get the solutions engineered correctly.
If election systems procurements continue to require performance bonding and other promises from voting system suppliers that the business situation just doesn't support, then this business will continue with a variety of small companies with even smaller technology — companies who have little to lose and less to offer.
The immediate future remains bleak (Back to top)
And, it appears that the situation will get even worse before it gets better. Today's voting-systems suppliers do not have enough capacity to meet market needs between now and the congressional elections in 2002 let alone the next presidential election. Interestingly, most of the voting-system suppliers capable of handling large-market sales are still providing only the defective punch-card or optical-scan systems. Any new player is going to require years to get up to speed on the real "voting requirements" necessary to field a useable product, and as I've seen, finding the significant capital necessary to bring a new system to market is probably still going to be difficult.
The fact is that election officials are some of the most dedicated public officials I've met. They work hard because, frankly, they have to work hard. With what they have to operate with they have no choice. It really is amazing that elections come off even as well as they do. Over and over again, I have heard the plea, "Please build us a system we can buy that really works."
Before we can do that, the business system must undergo major changes. The states will have to lead these changes. A sampling of state legislative bills already introduced shows that this problem has their attention. Here are some changes that will aid the process.
Voting systems should be procured at the state level and provided to counties equitably. This will help ensure that the quality of the votes recorded and counted are consistent from county to county and town to town. It also ensures that there is enough money to purchase a system that can be used by everyone and does not bias the vote towards rich or poor. And, it offers the vendors a legitimate business opportunity to provide on-going service for their products and maintain a long-term business relationship with their clients.
The other issues to overcome are the onerous requirements that make the environment so impractical for today's business operations and capital management. Any company capable of performing full statewide sales is going to have to be capable of raising huge amounts of capital. This is going to require state assistance at some point. Whether that assistance is merely a reasonable display of support for their chosen supplier or something more substantial, the private capital required to fix this problem cannot be had without some effective support from the equipment buyers. The old business model that made a company "bet itself" to make a sale has brought us to where we are now and it has been proven to be a defective business approach from both the buyers and the sellers point of view. It's a two-way street. The business requirements provided in voting-system procurements must create an acceptable business climate or we are going to be voting on punch cards for a long time to come.
I was somewhat amused when Cal-tech announced that they had teamed with MIT to develop a new voting system. It reminds me of the work in voting systems that another of our national technology treasures, Sandia National Laboratory, sought to achieve a few years ago. Sandia did good work, but there was not a business climate to bring their work to market. If Cal-Tech really wants to solve the problem, I hope they hand the research over to their business department. That's where the work has to start. A good system can be built by a number of organizations, but it has to be offered and purchased to be used by voters and that requires an acceptable business climate.
A lot of people are going to have to step up to the plate to fix this problem. We'll step up to the plate once more. Certainly it appears that state governments are going to step up. Hopefully, the venture capital community will also step up to the plate and at least ensure that this vital public service has the funding necessary to provide America with the fundamental means it needs to conduct fair elections where all of the votes are counted.
I think it's fairly safe to say that the political will does exist now to make the changes that will give companies like mine the opportunity to bring our advanced systems to market. I look forward to participating in the process.
About the author (Back to top)
Jim Willard has been in the research and development of advanced computer systems since 1974. Since 1988, Jim has been involved in the research and development of advanced voting systems. He holds patents on critical election's technology including the "electronic absentee ballot" and the "Audio Ballot." He has other patents pending including the patent of his optimal "electronic voting system."
Jim incorporated Votation Corporation in 1994 to market and sell his voting system products. Jim and his company have been invited to a variety of elections-related events, such as the California Election Summit and testifying before the House committee on Government Affairs – Elections Subcommittee. The problems of the recent presidential election have spurred Jim into bringing his advanced voting systems technology to market. In 2001, Jim will release his new business plan in an effort to see if there might be new venture capital interest in this high growth market space. Jim's present company Paper Computer Corporation (PpC) makes a single-use, disposable Internet access device derived from his original Electronic Absentee Ballot design. PpC will produce the Paper Computer configured as an absentee ballot or an Internet voting appliance. The Paper Computer Absentee/Internet Ballot solution will be offered through Votation on license from PpC. To learn more about Votation, contact Jim via e-mail at votation@bellatlantic.net.