There is a considerable amount of good in HR3295, and I expect that its long term consequences, if enacted, will be generally positive, but there are several elements of this legislation that concern me. I am particularly concerned that this legislation will force the government and states to spend a considerable amount of money on voting systems that barely meet current standards, as ammended by this act, but do not meet any of the new standards that will be developed under this act. This seems irresponsible! It would be far more responsible to put the new standards in place first and then start spending money.
I believe that the standards committee and its governing boards are formulated in a way that is likely to lead to standards that are not agressive, but rather, respond to the capabilities of current vendors in the voting system marketplace. I also believe that the committees that does the actual work ought to be compensated. As constituted here, the actual standards are likely to be developed by consultants hired by the Election Assistance Commission in support of the boards and committee, without the kind of active oversight that I believe this work requires.
Finally, I am worried about the issue of handicapped access. If the legislation is enacted as worded, and if the courts interpret it literally, it may lead to the imposition of a very expensive legslative mandate. I have outlined my concerns about these issues more fully below, in response to the text of this legislationi; I should note, however, that in many cases, I have only commented on an issue the first time it occurs in the text; some clauses are repeated several times in later parts of the text:
Why are we in a hurry to get rid of punched card systems? The Datavote system, using long cards with candidate names printed on the cards and using a spring-loaded positive-action punching mechanism has generally been cited as a generally decent system -- it does not lead to problems with the interpretation of chad, and voters can easily interpret their own ballots after punching, to verify that the punches on the ballot conform to their intent.
The Votomatic ballot is, indeed, worth phasing out, but should we be funding a crash program to do so? The experience in Florida does not justify this. Rather, it points to two serious flaws in election administration that were not in any way confined to punched card voting systems.
First, there was the notorious butterfly ballot in Palm Beach County. Other counties using votomatic ballots did not make the mistake of designing such a ballot, and therefore, avoided the problems caused by this ballot design. Furthermore, other Florida counties using optical mark-sense ballots had analogous errors in ballot design that were equally serious. For example, in Jackson County, the presidential race was split into two columns, without notations such as "continued in next column" or "continued from previous column". This error in ballot design led 6 percent of the voters to cast unintentional overvotes.
The problem here was not a failure of technology, but rather, a failure to establish reasonable standards for ballot layout, independent of technology. We must establish such standards! Part II of the new Federal Election Commission standards, due to be released in draft form in January 2002, promises to address some of these ballot design issues. Unfortunately, the draft standard is grounded in intuition, not solid research; I am confident that even this intuitive set of ballot design guidelines will lead to improved election administration, but we really need to do the research to validate these guidelines!
Second, as the Supreme Court correctly recognized, Florida failed to establish uniform standards for what constitutes a vote. As a result, a bit of dangling chad that would be counted as a vote in one county might not have been counted as a vote in another. Again, this problem was not confined to punched card ballots! During the recounts, optical mark-sense ballots were also subject to hand examination, and there were many pencil marks that were counted as votes in some counties that would not be counted in other counties.
In fact, it is quite practical to establish standards for a hand recount of punched card ballots, where these standards are both uniform and honor voter intent. The key lies in distinguishing between chad that was partially dislodged because the voter intentionally withdrew his or her punch prior to completing a punch, and chad that was partially dislodged because, for some reason, the voter could not dislodge the chad. As is illustrated in my web page on chad (see: http://homepage.cs.uiowa.edu/~dwjones/cards/chad.html) the bits of chad resulting from these two circumstances look quite distinct, even if they remain attached to the ballot by all four corners! Early withdrawal of the punching stylus produces chad that is slightly folded in the center, while punching against an obstruction produces chad that with an almost sphericalbulge.
Votomatic voting machines are particularly prone to obstructions in only a few positions on the ballot, and if these machines remain in use, these ballot positions should simply be avoided. These positions are documented in the above-cited chad web page.
I want to emphasize that the Votomatic voting system should be phased out. With the procedural safeguards I have outlined above, fair elections can be conducted using the Votomatic system, but I am only arguing for a deliberate phaseout of this technology, as opposed to an emergency program of immediate replacement. Even with appropriate the safeguards I have listed, the Votomatic system still poses problems. The most serious of these is that the voter cannot easily check his or her ballot to make sure that the punches present reflect the voter's intent. When in the voting machine, the machine itself obscures the ballot so that the voter cannot easily see where it is punched. When out of the machine, the ballot is just a card with a cryptic pattern of holes. With an appropriate key, these holes can be interpreted, but all other voting systems based on physical ballots allow a voter to read the votes from the face of the ballot without use of such a key.
Is it realistic to ask for the replacement of all punched card systems by November 2004? One thing the Florida 2000 experience teaches fairly clearly is that the first use of new voting systems is prone to serious difficulty. Neither the voters nor the polling place workers are experienced with the system, so errors abound. Therefore, new voting systems should see their first use on elections with low turnout, so that polling place workers do not have to deal with crowds at the same time they are learning what problems the voters have. Only after working through a few such elections should the new voting system be put into use for statewide and national offices.
Typically, municipal and school district elections have the lowest turnout. The off-year general election typically draws a lower turnout than the congressional elections on even years, and the school elections many states hold in the early fall are among the lowest turnout elections of all. This leads to the following idealized timetable for bringing a new voting system into a jurisdiction:
The new voting system should see its first use in school elections or special elections in the spring, summer or early fall of an off year. Many states hold school elections in September, so I will assume this in what follows. After such a use, it can be used in the off-year primary and off-year general election for state and local offices, and finally, it can be used in the national elections that occur in even years.
In terms of the current election cycle, this means that new voting systems for use in the 2002 general election should have been in place by the end of August 2001, and new voting systems for use in the 2004 general election should be in place by the end of August 2003. Having the system in place means that the machines are in polling places and the election workers are properly trained. This means we need delivery of some machines and completion of the ballot design by July, and the county election administrators need to be trained by June if they are to complete the ballot design by July. This, in turn, means that the purchase decision needs to be made in May. If we use competitive bidding, we are talking about a several month process, with the request for bids being released in January or perhaps February of 2003.
If we assume that HR3295 will pass this January, this gives state and local jurisdictions just one year to commit to the purchase process and determine the standards they will set for the new voting systems they purchase -- presuming that the request for bids is the document that ultimately sets the standards to which the vendors will be held. If we read the November 2002 deadline for obligating the funds (as set by 101.c.1.a) as setting the date for publishing the request for bids, we push the process forward a few months, leading to a more relaxed phase-in of the new system. If, on the other hand, we interpret this as the date for signing the purchase order, the bidding process must begin in July or August of 2002!
Therefore, HR3295 can be interpreted as requiring the states and counties to adopt reasonable standards governing the purchase of new voting systems by some time between July 2002 and January 2003! Assuming that the bill is on the fast track through the House, this gives the states and counties only 6 months to a year to set such standards. This seems very brief, and it certainly precludes waiting for an improved FEC standard or any standard developed by the Election Assistance Commission established by Title II of HR3295.
Given the weaknesses of the current voluntary standards, promulgated by the Federal Election Commission in 1990, I certainly hope that any new standard will be significantly more rigorous than this standard. I therefore expect that many of the voting systems currently on the market will require significant changes to meet the new standard, and because of this, I would discourage a widespread emergency purchase of new voting systems built to the old standard.
The first 55 members of the proposed Standards Board are mostly Secretaries of State or other elected state officials, since in most states, it is the Secretary of State who oversees the conduct of elections. The second 55 members will mostly be county auditors, county clerks or county commissioners of elections, since in most states, counties conduct elections and these elective offices are the ones most commonly charged with the conduct of elections.
In sum, the Standards Board is composed of 110 elected officials, none of whom are expected to have any experience in the formulation of standards and none of whom are expected to know anything about the actual internal mechanisms of voting systems.
Generally, the county election officials will have first-hand experience with the administration of elections, while the state officials will typically rely almost entirely on appointees and civil service staff in an office of elections. While the county officials typically have excellent experience in election administration, most of them have little technical experience; instead, they rely on close working relationships with the vendors who have sold systems to their counties. As such, I expect the resulting standards committee to be very trusting of what they are told by the vendors, and as a result, I would expect the standards they promulgate to pose very few challenges to the leading vendors in the marketplace.
Because all 110 members of the Standards board are elected officials with full-time job responsibilities, I expect that none of them will have the time to offer more than nominal oversight over the standards for which they are nominally responsible. Therefore, most of the work will be done by the Board of Advisors and the Technical Standards Development Committee.
If we cannot rely on the Standards Board to take a leadership role in the formulation of the standards that govern our voting systems, we must hope for leadership from their Board of Advisors. Of the 25 members of this board, 2 are required to have technical background (item 12), 5 will most likely be lawyers (items 1, 2 and 13), and the remainder are appointed without any legislative constraint on their qualifications. In effect, the two boards provide massive representation from the customers who buy voting systems, that is, the states and counties.
It is fair to guess that some of these appointees will have significant expeertise in elections, as judged by the agencies appointing them. The National Association of State Election Directors, for example, played a pivotal role in the development of the 1990 FEC Voting System Standards, and the Election Center has a long record of searching for and publishing the information on the best practices in the field. Therefore, I expect 4 members to have significant expertise (items 6 and 10), with 4 more with at least some expertise likely from the associations of county election officials (items 8 and 11).
This gives us the following likely pattern in the membership of the proposed Board of Advisors:
The diversity constraint is interesting in that it asks for more than mere partisan diversity, but the requirement for geographic diversity is a puzzle What kind of geographic diversity is relevant? Does it matter if the standards are all formulated by southerners or by people on one or the other coast? I can't think of a reason to worry about this. I would be concerned, on the other hand, if all of the members of the Board of Advisors came from urban areas, or if all came from rural areas. The election technologies that are most appropriate for large urban areas like Los Angeles, New York, Atlanta and Dallas are likely to be quite different from those most appropriate in a rural place like Barry County Michigan or Cedar County Iowa.
What we really need, though, is a way to constrain the balance between people with expertise in the formulation of technical standards and those with experience in the conduct of elections. With too many of the former, the product of this process is likely to be unworkable. With too many of the latter, the process is likely to be too biased toward the status quo.
I seriously wonder how we can expect the board members to devote much time to the standards process if they receive no compensation. My first reading of the 1990 Federal Election Commission Voting System Standards took several days, and my attempt at a serious review of Part I of the 2001 draft standard took several weeks. Many of the people likely to serve on these boards will have little free time to devote to these jobs unless their employers are willing to donate their services, and in the case of the Standards Board, all or most of the members will be elected officials with constitutional duties that preclude their devoting much time to this project.
If we want serious standards, we must be willing to compensate those who direct the work! If we expect the board to vote on a standards proposal, we must at the very least be willing to compensate the board members for the time it takes to read through the proposal! As things stand, essentially all of the time consuming work will be done by the staff serving the boards, and the board members cannot be expected to be much more than rubber stamps.
As it stands, we have a pair of boards, one ill prepared to evaluate any technical content, and one with a little technical expertise and a fair amount of expertise in the conduct of elections, making essentially a volunteer commitment to their boards. They won't have the time to develop standards, and they will hardly have the time to do a serious evaluation of the standards proposed to them by others. The only ones with the time to devote to serious effort on standards development will be the vendors, and they are unlikely to develop standards that do anything but favor their current products. Who will counter this?
Given the make-up of the boards, as already noted, there is very little guarantee of any technical expertise. As a result, it is very odd to ask them to evaluate and adopt engineering standards!
While I agree that correction at the precinct or what some have called second chance technology is desirable, it offers no protection for absentee voters. Thus, all of Oregon is left unprotected, as are a growing percentage of voters in states that permit unrestricted absentee voting.
In fact, the primary reason to require second chance technology is to permit the voting system to detect overvotes and give the voter a chance to correct them. In the 2000 general election in Florida, it is true that this would most likely have changed the outcome, but if we had adopted reasonable standards of ballot design, the vast majority of the overvotes encountered in Florida would never have occurred. It was serious ballot design errors that led to overvote rates of as high as 11 percent (in Gladsdon County) and averaging 5 percent for all users of central count optical mark-sense systems. If we had not made these errors, the overvote rate would have been far lower and the need for second chance technology would have been far less serious.
If we are to protect absentee voters, we must adopt far more stringent ballot design standards, and we must adopt vote counting standards that assure that the overvotes that remain among absentee votes are subject to human inspection. so that voter intent can be discerned wherever it is expressed. The 'what is a vote' legislation currently being drafted in Iowa attempts to offer such protection, and I have outlined a somewhat more ambitious model for this in http://homepage.cs.uiowa.edu/~dwjones/voting/optical/. I believe that the combination of such vote counting standards with good ballot design rules will offer substantially the same protection as is offered by second-chance technology. Because of this, I am suspicious of a legislative mandate for second chance technology, and I would perfer that the legislature legislature mandate results and not the specific means used to achieve those results.
The requirement that new voting systems provide "practical and effective means for voters with physical disabilities to cast a secret ballot" is wonderful but in its broadest interpretation impractical. Does it mean some disabilities, the common disabilities, or all disabilities?
I would hope that we will be able to accomodate the needs of most disabled voters, because there are, indeed, broad classes of disability for which reasonable accomodation can be made. The most common disabilitiy category includes those with a mixture of mild to moderate motor problems with mild to moderate vision problems. Think of the elderly voter with a severe tremor and tunnel vision. The needs of blind voters are fairly well understood as well, and I would hope we can meet these.
What bothers me is the possibility that we might be required to meet the needs of all disabled voters. The problem is, we cannot anticipate all of the problems nature or accident may saddle a voter with, and as a result, we can never guarantee that any technology or combination of technologies will meet the needs of all disabled voters. There are some disabilities that I suspect will always require an assistant to cast a secret ballot. Among these are combinations of severe motor problems with speech severe speech impediments. Those with such a disability will be unable to use manually operated devices such as a pencil or a touch screen, and they will also be unable to use speech input to a voice activated voting system.
As already noted, the Standards Board and Board of Advisors, as constituted by Sec. 213 and Sec. 214 will be effectively rubber stamp boards. As constituted, they will have neither the time nor the expertise to develop the standards called for in this legislation. As a result, the real work will fall on the shoulders of the Technical Standards Development Committee. Given the preliminary work done by the FEC in proposing a revised voting systems standard that should be available in early 2002, the request for a preliminary recommendation from this board within 9 months is reasonable, but the question is, how does this fit in with the remainder of the legislation?
Sec. 101 asks us to put in place replacements for punch card voting systems by November 2004. As I pointed out in my response to that section, the latest responsible date to issue a request for bids to purchase these new machines will be in February 2003. It is this request for bids that will set the binding standard that the new voting systems will have to meet, so that standard must be available for incorporation into the requests for bids by the end of 2002. If we want any part of the new standard to be applicable to the new voting systems purchased under this legislation, the 9 month period means that the Technical Standards Development Committee must be fully in place by April 2002!
If this legislation passes in late 2001 or early 2002, this gives us just 3 months to organize both boards and the standards committee. If we cannot do this, we will release the new voting system standards proposal after it is too late to incorporate it into the request for bids, and as a result, we are very likely to be purchasing machines built to the 1990 FEC standard, with all of its shortcomings! If the Technical Standards Development Committee rushes its work, it is unlikely to do a good job.
As it is, the timetable is so tight that there is no room for a public comment period on the preliminary standard before purchase decisions for new voting systems are set, and there is no time for the vendors to develop new systems to meet the new standard. Today, the certification of a new voting system usually takes at least 6 months and frequently as much as a year! This timetable may require accelerated certification if the vendors are to produce systems that meet the new standard in time for delivery in the summer of 2003, and my experience with the certification process suggests that this would be a mistake! In fact, certification delays make me suspect that even if the new FEC standard expected in January 2002 is adopted without change, it will be difficult to certify new systems to this standard by the summer of 2003!
As a result, I conclude that the voting systems purchased under Title I will most likely be bound not by any new standard, but by the 1990 standard, as ammended by the restrictions imposed by this bill. If this is destined to be the case, there is no rush to develop new standards, and the important thing will be to build an enduring institution for continuing the development of standards!
The committee structure required is quite interesting because it allows several different balances in the membership. The clause requiring "An equal number of each of the following" could fix either 3, 6 or 9 seats on the 14 member committee, depending on whether we take 1, 2 or 3 from each category. The director of the NIST and the representative of ANSI make 2 more, so this leaves us with either 9, 6 or 3 from the "other individuals with technical and scientific expertise" category.
I expect that the members of the boards appointed to this committee will be among the more technically knowledgable, so I expect them to be able to contribute meaningfully to the debate, but I am worried that they will not have the time to devote much effort to this committee, and this committee is where the major work will be done! I also expect that the director of the NIST will have very little time to devote to this committee. Therefore, I expect that the majority of the work will be done by the 4 to 10 other members.
This committee is going to do most of the work of establishing the new voting system standards. Since they are not to be compensated for their service, but only for their travel and time away from home, if we expect them to devote significant time to this project, we must expect them to spend lots of time traveling and away from home! I believe that this is not a responsible way to spend money!
The World Wide Web was developed to facilitate collaborative writing efforts between geographically dispersed teams of researchers, and a decade of experience has proven that this works very well! This funding structure discourages this kind of long distance collaboration, and instead encourages wasteful travel. Instead of a few brief meetings, with long hours spent between meetings developing and refining standards proposals, I expect this funding structure to lead to several weeklong or even longer working meetings at expensive hotels because those kinds of meetings are compensated, while any work done from home by internet, phone and fax is entirely uncompensated.