Washblog

---

Thanks to Jason Osgood (Reckless Plan series) and Timothy White (Timothy White and Allan Rosato vs. San Juan County Auditor's Office) for help & info with this story.

---

Jason Osgood, in his Reckless Plan series on Washblog, has also outlined many of the flaws and risks in this plan.  As reported on Postman on Politics on September 18, his testimony lead to the rejection by the County Council of one part of that plan, the purchase of VoteHere's MiBT.  

Jason testifed that Dategrity, the company from which the Council members expected to purchase the MiBT, appeared to have gone out of business.  Its phone was disconnected, its website down, and its office apparently vacant.  (1)  Evidently, REALS had not informed the Council that The Election Trust, LLC had announced on August 31, 2007 that it had secured reseller and development rights for the MiBT  from Dategrity.  Councilmember Ferguson informed Sherill Huff, REALS' Director, that this change in product ownership was a significant development that should have been relayed to the Council.  (2)  The Council unexpectedly passed a unanimous resolution directing that the "VoteHere software should not be purchased at this time."

Prospects for the MiBT suddenly appeared much diminished.  But in a 9/19/07 phone conversation, REALS Director Sherill Huff told me that she expects to proceed with the purchase of this equipment anyway.

SUMMARY AND CONTENTS
The purchase and system overhaul proposed by King County Records, Elections, and Licensing Services (REALS) contains several components.  This story focuses on one of them, the VoteHere Mail-in-Ballot Tracker.  But much of the information and most of the documents referenced here are applicable in relation to the entire business case.  

There are three sections, below.  Click on the hyperlined headings to get there.

I.   COST
A discussion of what the MiBT system may be expected to cost King County, both initially and for years into the future, based on:

1. Ballot Tracking and Accountability Report, REALS, 5/15/07 (Business Case)
2. Response to Questions, Diebold Election Systems, 4/18/07
3. Staff Report, prepared by Nick Wagner, Principle Legislative Analyst for the King County Council, 7/9/07
4. VoteHere Software License Agreement, pp 7-12, a contract between VoteHere and San Juan County, Washington (provided to me by Timothy White who noted that it was signed after the Sept 05 Primary and before the Feb 06 Special Election. The first 6 pages of the contract are in a separate pdf).

II.  FUNCTIONALITY AND SECURITY
A discussion of the private companies offering the software and the expectations that this software will deliver on its promised security, functionality and costs savings.  This discussion draws on:

1. Media accounts of the software and of Dategrity and The Election Trust
2. My conversations with Sherril Huff of REALS and Dale Miller of VoteHere
3. A reading of the citizen, expert, legislative, and peer reviews listed as #1-1 - 1-4 in the footnotes below and
4. A reading of Timothy White and Allan Rosato vs. San Juan County Auditor's Office.

III. PROCESS
A discussion on process.  This includes the process by which REALS has reached its determination to make these purchases and conduct this overhaul.  It also includes a discussion of the process that the King County Council has followed to review REALS' work.

I think the record shows that, in this case, REALS' process is inadequate to the task of providing a solution to the voting system needs of the county that serve the public interest. On the other hand, the King County Council conducted a thorough due diligence review which slowed down the purchase of the new equipment and documented that the proposed plan was a mistake.  However, the results of the Council's review were not translated into policy or even, it seems, heeded.

 

---

COST

---

I haven't found a consolidated summary of initial and ongoing costs associated with the proposed purchase of VoteHere's MiBT by King County.  I suspect that one may not exist or be publicly available.  The summary I've put together below is a citizen best-guess meant only to draw a "big picture" of the cost potential of this proposal - and to show that the costs are bigger than most people probably think and that the documentation is unclear and inconsistent and may contain warning signs of potential unexpected costs.  

Initial costs for VoteHere's MiBT system will be significant.  REALS' Ballot Tracking and Accountability report (Business Case) cites $2.7 million available in grant resources from the Help America Vote Act (HAVA) that will "support procurement and significantly offset initial costs" for a range of voting technologies now under consideration, including the MiBT.   Two observations are in line here.  First, that these HAVA funds only "offset" -- and do not fully cover -- the initial set up.  Second, these initial costs would be followed by years of ongoing costs that would not be offset by this HAVA grant.

Ongoing costs are likely to exceed initial costs within the first few years if not sooner -- and are not anticipated to be covered by federal grants.  Of Washington's 39 counties, most use some kind of ballot tracking system, including 25 counties that currently use the VoteHere MiBT.  Most of these counties use the tracking systems for either outbound ballots (blank ballots sent to voters) or inbound ballots (completed ballots mailed back by the voters) -- not for both. (3)   The VoteHere MiBT plan currently before the King County Council would do both.  

According to an article in the Olympian, Ballot Tracker in Fiscal Pinch,  Washington's legislature decided this year against allocating $3.5 million to maintain the ballot tracking systems in the counties that use them.  Given King County's status as the state's largest county (MiBT costs are based on the number of registered voters) -- and the fact that it would adopt a comprehensive MiBT plan tracking both outbound and inbound ballots, it is reasonable to assume that ongoing costs for the county would equal a significant percentage of the combined costs for all other counties as they currently use these systems.

This general information can give us a ball-park idea of what we'll be on the hook for in King County in coming years.  I outline in the tables below a summary of the specific
information that I have found.  It doesn't include the kind of costs -- related to staff costs for training, learning curve, and support -- and all the other contingencies that accompany complex systems, including, equipment maintenance and replacement, insurance coverage, etc.  I have not seen any indication that these costs have been considered and compared with current system costs.

In fact, REALS' Business Case glosses over this kind of analysis.  A section of the report entitled: "Cost Benefit Analysis," offers reassurance that automating procedures with the VoteHere MiBT and other products will "mitigate against rising costs associated with the current labor-intensive manual process".  Despite the section heading of "Cost Analysis", there is no analysis -- and no data -- to back up these reassurances.   I lingered over this section of the report for awhile, surprised at this lack.  Later, I was glad to see that I was not alone in my questioning.  A July 9 staff report prepared by Nick Wagner, Principle Legislative Analyst for the council, notes that this cost benefit section contains "...no quantitative analysis to support its conclusions."  

There has been such a push by REALS from the beginning to champion this purchase that it has not undergone an adequate cost-benefit review.  I would also have appreciated seeing, within this section of the Business Case, mention of a factor
recognized in the January, 2007 Main-in Ballot Tracker Report to the Legislature prepared by Washington's Secretary of State.  The report notes that "The automated ballot tracking systems that are currently available use different approaches and have varying strengths and weaknesses.  For example, because MiBT is relatively labor intensive, it is of greater benefit to small and mid-sized counties."

INITIAL COSTS
The initial capital and operating costs for the VoteHere MiBT, are given on page 7 of the July 9, 2007 Staff Report on KC Council's Motion # 2007-0328 as $345,000. Other documents I've looked at list costs that do not seem to align with this quote -- and include additional costs that this report does not appear to consider. Maybe this staff report cites only those costs for the MiBT that would be covered by HAVA?

DIMS TO GEMS Data Download	Software License	Equipment	Standard/ Professional Support Both an initial and an ongoing cost
Diebold explains in a 4/17/07 report to REALS (starting on page 50 of the Business Case) that there will need to be a “DIMS to GEMS election data download for ballot layout” to ensure data consistency on precincts, districts and other information between the two systems. DIMS, King County’s Data Information Management System, contains voter registration data. GEMS, the county’s Global Election Management Software, contains voter information necessary to tabulate ballots. The proposed system requires a new consistency between these two data formats because tracks ballots from when the county first mails blank ballots to its registered voters (DIMS) – and concludes when it opens and verifies the envelopes containing the completed ballots (GEMS). Jason Osgood cited a $750,000 cost for this data download in a 9/18/07 story on Washblog. Last week he alerted me in a followup email that he believes he was in error and that this cost would be closer to $650,000. I have not been able to find a cost estimate for performing this download. Given my experience in pricing services like this in the library sector, $650,000 does not sound off base. I assume that, although the use of the VoteHere MiBT would require this data download, it would not be performed by VoteHere.	Page 17 of VoteHere’s MiBT Proposal (Approximately page 168 on the pdf of REALS' Business Case) gives a one-time licensing fee of $675,000 (900,000 registered voters at $.75 per voter). This is an undercount. King County had 999,134 qualified registered voters in the August, 2007 primary election. At $.75 per voter, that's approximately $750,000. I am told, however, that VoteHere’s charge also includes voters whose registrations are inactive, and that this involves a total of 1.2 million King County voters, which would bring the initial licensing fee to $900,000. A table entitled “Evaluation of Vendor Proposed Solutions” that appears at the very end of REALS’ Business Case (page 189 in my pdf) gives a different estimate. This table cites $.30 per registered voter, at 1 million voters, for an initial licensing fee of $300,000. Assuming 1.2 million registered voters, that fee would be $360,000. The discrepancy between the cost given by the county, and the estimate given by VoteHere does not appear to be explained in the Business Case, so it is unclear whether it represents an error, a misunderstanding, a change in plan, or some other difference in calculating costs. Based on a reading of a King County Council Staff Report, the Council is operating under the assumption that the lower cost of $300,000, cited by REALS, is the correct one. I wonder if it is possible that there are two separate charges during the first year, the $.30 per voter fee that will be charged every year -- and an initial licensing fee of $.75 per voter, as listed in the VoteHere document.	Pages 28-29 of King County’s Business Case outlines additional anticipated equipment needs related to the VoteHere MiBT, including an unspecified number of desktop scanners – at $200 each, and an unspecified number of personal computers – at $1,500 each. Alternately, or in addition or some combination thereof, an unspecified number of high speed scanners at a cost of $15,000 to $70,000 may be required.	A standard support fee of 15% of the $.30 per-voter cost will be charged in the first year and all subsequent years. If the number of registered voters in King County is 1 million in 2008 (it is likely to be higher) this fee will be $45,000 for the first year. I don’t see, in REALS’ business case, any explanation of what services are offered in exchange for this standard support fee. Based on San Juan County’s contract with VoteHere, it is likely to cover only problems with the software, described in that document as: “Diagnosis (and resolution) of reported problems or performance deficiencies of the Software or errors in the Documentation.” Professional support, which is charged at $160 per hour, is available for other needs.

 

ONGOING COSTS

Fee charged per registered voter each election

Professional/ Standard Support Both an initial and an ongoing cost

At the very end of REALS' business case, in a table presumably prepared by REALS' staff, entitled "Evaluation of Vendor Proposed Solutions", an ongoing cost of  $.30 per registered voter is given.  This would be $300,000 per election based on an estimate of 1 million voters in the county.   The cost would rise to $360,000 per election under the assumption that there are 1.2 million registered voters. As the number of registered voters in the county increases, so will costs. It is puzzling that VoteHere does not present this cost in its  MiBT proposal (see page 17 of REALS' business case).   However, it is reasonable to assume that a per-voter charge on this order is likely to be assessed, given the fact that that San Juan County's contract with VoteHere contained a similar charge, of $.35 per voter.

VoteHere charges a standard 15% fee for support.   I don't see, in REALS' business case, any explanation of what professional support entails.  Based on San Juan County's contract with VoteHere, it is likely to cover only problems with the software, described in that document as: "Diagnosis (and resolution) of reported problems or performance deficiencies of the Software or errors in the Documentation." In addition to this standard support, there is a $160 per hour fee for professional support that I assume covers help needed for any other need not related to deficiencies in the software.

 

---

FUNCTIONALITY AND SECURITY

---

According to the expert review commissioned by the King County Council, King County can achieve the goals that the VoteHere MiBT is designed for with a simple web form that can be created in-house (4).  Is it possible that this technology, which would cost millions of dollars over the years, is simply unnecessary? I don't see indication that there has been serious consideration by REALS of this statement by the independent reviewers that the MiBT is not needed.

The product also appears to be a risky proposition.   According to an article in the Olympian, Ballot Tracker in Fiscal Pinch, Washington's legislature had decided against allocating $3.5 million for ballot tracking last session, due, in part, to concerns that some legislators had about that technology.  (5)  Among other considerations, San Juan County in Washington State is currently being sued in relation to its use of that technology.  The plaintiffs provide evidence that the technology performed very poorly.  (6)

The charge of poor performance of VoteHere software is not new.  In 2003, a VoteHere engineer named Dan Spillane, sued the company charging that the software had numerous flaws, that VoteHere's management refused to correct them, and that when he tried to alert the US Government Accounting Office about these flaws, his employment was terminated by VoteHere. (7)

There are also risks involved in dealing with private companies, particularly when they are in transition.  Ms. Huff, Director of REALS, reassured me in our 9/19 conversation that there are no significant implications for King County related to the recent sale of the MiBT from Dategrity to The Election Trust. Botwithstanding Jason's testimony, she said, Dategrity had not dissolved, but was simply moving to another location. All the programmers and other support staff for the MiBT have moved with the product to The Election Trust. The only thing that's changing with the transfer of VoteHere to The Election Trust, Ms. Huff said, is that the county will deal with a different sales force. This kind of change is very common in the industry and has no substantive significance.  Furthermore, Ms. Huff said, she wished to correct a misconception that she said I held about VoteHere. VoteHere is not a company she said but, rather, a product. It was developed by Dategrity.

I believe that Ms. Huff is mistaken on several counts here.  Dategrity has not simply moved to a new office, but has indeed vacated its office, phone, and website.  According to John Cook of the Seattle PI, its founder and president, Jim Adler, has started up a new company, Demoxi, and has moved the cash resources of Dategrity to that new company.  This is not a name change.  The two entities are registered as separate companies in Washington State: Demoxi and Dategrity. Dategrity remains a legal entity, which, according to Dale Miller of The Election, who I spoke with on 9/28, continues to hold the intellectual property rights for the VoteHere MiBT.  But all indications are that it has been "emptied out" and is no longer conducting regular business.  

As for VoteHere, it is not, as Ms. Huff told me, a product.  The name is sometimes used as a verbal shortcut to refer to the MiBT.  But VoteHere is the company that developed the MiBT.  John Cooke of the Seattle Post Intelligencer characterized VoteHere in a 7/20/07 entry on his Venture Blog as "a Bellevue online voting startup that -- despite raising more than $20 million -- never really took off."  He goes on to quote Tom Simpson, "an original investor in Dategrity/VoteHere and a board member at Demoxi" as explaining that the original concept at VoteHere (presumably what King County is now considering purchasing) "didn't develop as quickly as the company thought. But he expressed confidence that the cash and brain power that went into the idea can be reused at Demoxi."

These distinctions on the state of Dategrity and whether VoteHere is a company or a product may seem like splitting hairs to some.  But they are significant for several reasons.  

First, the Director of Elections should understand and clearly represent the structure, ownership, and status of the companies and products in which she is proposing to make such key investments.  In addition, there are important uncertainties related to the proposed purchase of this equipment that have yet to be addressed.

At the very least, the County will be signing a contract with one company and not the other.  This difference is particularly important when dealing with proprietary knowledge-based products like the MiBT because the user is usually limited to seeking technical support only from the company granting the license.   When businesses are in transition, uncertainties deepen.  The fact that the licensing and servicing for the MiBT is held by one company -- and the intellectual property rights by another -- indicates to me that there may be even more potential for King County to be limited in its options should any unexpected situation arise with this product or the associated companies.  

I also found it interesting that my call to the Washington Secretary of State's office caused an official there to notice during our conversation that The Election Trust, LLC, has a legally impermissible name.  As per RCW 25.15.010, she said, limited liability companies may not include the word "trust" in their names.  My apologies to Election Trust if this precipitates a name change.  It was probably bound to happen eventually, even without my call.  

This is a small detail, having only symbolic significance.  But the sybolism is ironic.  What happened with The Election Trust's due diligence when it incorporated in 2003?  Didn't the founders check the name for standard legal conflicts?  It's the small detail that's famous for crashing complex systems like elections.  All this hints at the importance of performing adequate research and evaluation on all potential major vendors.  We are placing matters of critical public trust in their hands. It's not responsible to do that in such a rush and with such wilfulness that evaluation of vendors is given short shrift.

The vendor selection process section of REALS' Business Case starts on page 21. It considers the technical capabilities offered by the vendors, but contains no substantive evaluation of the companies themselves. The report states on page 26 that, in contrast to the other vendors considered, VoteHere was evaluated on only two of twelve criteria because its system involves no hardware. These two criteria are technical: "the ability to capture voter identifying data after the envelope was opened, and system integration and process management." Later in the vendor evaluation section, VoteHere's dependability as a vendor is briefly characterized: "VoteHere's MiBT ballot tracking software is in use and working effectively in several jurisdictions in Washington." The report does not mention that this positive evaluation is disputed in at least two lawsuits involving this company (6, 7).  Due diligence checks on companies prior to signing major business deals with them seems like a logical step.

 

---

PROCESS

---

I believe the record shows that REALS' process for evaluating costs and benefits and selecting and evaluating vendors for this major purchase was inadequate.

The King County Council required three independent reviews of REALS' proposed Business Case, conducting its own review, and inviting citizen testimony.  But these recommendations were not heeded. The investment of citizen and expert time as well as taxpayer funds to conduct these reviews did not translate into policy.  Nor did they appear to spark further review by REALS.  It appears to me that Sherril Huff and the rest of REALS continued with their plan as if there had been no review.  

There must be some reasonable way to require that vendors be adequately evaluated for financial stability and other factors, that cost benefit analyses contain analysis, and that peer, expert, legislative, and citizen recommendations be considered.  I also found myself wondering if there is a county-level equivalent to the fiscal analysis that Washington's Office of Financial Management performs for fiscally-significant state bills.

The VoteHere MiBT is proposed to increase citizen confidence in elections after our experience in 2004 with a close gubernatorial election and a recount that uncovered several instances where ballots had been misplaced. The inadequate process used in this case is not likely increase citizen confidence -- and carries costs and risks that may do substantial harm.

We have also learned in recent years that faith in technology does not cure human error.  The recent California Top-to-Bottom Review, which decertified the equipment of all 4 major electronic voting equipment manufacturers in the United States has been a wake-up call to elections officials across the country.  What reason do we have to believe that The Election Trust would perform any better?

Automation, outsourcing, and digital technology all have their place.  But sometimes we get better results with the old-fashioned approach of listening to input from the people around us, doing tasks in-house, and relying on human judgment and labor - as well as electronic judgment and machine automation.  

NOTES

1. Jason Osgood's testimony at the council meeting starts at 1h 22m of the King County Television recording of the 9/17/07 King County Council meeting.
2. This exchange occurs at 2h 02m of the King County Television recording of the 9/17/07 King County Council meeting.
3. Main-in Ballot Tracker Report to the Legislature, Washington Secretary of State, 1/07.
4. Overall Ballot Tracking Recommendations: Recommendation One: In 2008, consider keeping the current ballot tracking system in place with only modest changes being made.  The current voter registation and election management system (DIMS) can produce reports which could be made available to the public via a web interface, to allow individual voters to check that their ballots were received and accepted for tabulation (verified)."
   Expert and Peer Security Reviews of Elections Ballot Tracking and Accountability Case, July, 2007, page 4.
5. The Olympian, in a May 2007 article entitled Ballot Tracker in Fiscal Pinch reports that, statewide, $3.5 million was requested of the state legislature by the Secretary of State to continue the ballot tracking program.   But the legislature, in part due to concerns over the way the technology was used in San Juan County, did not authorize the funding.
6. Timothy White and Allan Rosato vs. San Juan County Auditor's Office
7. See Vote of Trust: VoteHere Touts Safeguards to Allay Fears, Puget Sound Business Journal, 9/5/03.  Also see Voting Software Firm get Sued, Wired, 2/27/03.