www.dhaugh.com

https://www.dhaugh.com

As a seasoned computer architect, I was curious about the engineering behind our voting systems.BY Dennis HaughAfter the negative reactions to the 2016 election results by the Democratic party and the 2020 election results by the Republican party, I decided to look into our voting systems in some depth. As a very senior computer engineer who has led and implemented major computer systems for companies including Intel, Digital Equipment Corporation (DEC), and Lockheed Martin, I thought I might be able to offer a useful perspective on the topic. Never having looked into voting systems, I approached the problem with no preconceived notions.Unlike the efforts to prove voting fraud, this report is focused on the engineering practices used to create our current system. To be clear, the audit and regression analyses that are looking for fraud are investigating the results produced by the product of the engineering practices. This report assesses the engineering that produced the voting systems. The use of Dominion artifacts should not be taken as the sole focus. The goal is to look for improvements to the process.There can be no doubt that our voting systems are a target for both internal and external (foreign) attack. Every target has a breaking point, and it is no stronger than its weakest link. That is true whether the system is a military defensive position (like a fort) or an Internet application. Even Masada eventually fell. The question is: How truly vulnerable are our voting systems. Any intelligent commander will review his/her defenses and strengthen the weak links before they are breached. Voting systems must be approached with the same mindset.The software component of our voting systems is only one link in the chain from a voter to its final tabulation. Although the scope of this report is primarily focused on the software engineering practices, these same practices apply to a wider range of engineering problems. The wider scope of a total voting system is addressed by project management.OverviewSince Dominion Voting Systems were the target of a lot of ire, the logical place for a software engineer to start this investigation was to look for all the available data on code reviews for the Dominion software. Code reviews are one of the first steps of software verification. My hope was that I could satisfy myself that the engineering was satisfactory and go on to something more productive. Unfortunately, that was not the case.From the initial code review audit, I could work both backward and forward to get as complete a picture of engineering truth as I could, without access to the actual hardware and software source code of the voting systems. As expected, the results of the code review audit were not totally satisfying; however, it did produce some nuggets of information. In particular, it provided links to look at the engineering steps prior to the code reviews as well as pinpoint some apparent weaknesses in the code itself.There is no question that the Dominion software could be stronger, but it is but one link in the chain and perhaps not the weakest link. The weakest link may well be the process by which ballots enter the system. This is the classic “garbage in; garbage out” (GIGO) with which all computer nerds are familiar. It is a project management issue that is caused by a lack of firm requirements for the voting system. Fortunately, the Project Management Body of Knowledge (PMBOK) is consistent with the Software Engineering Institute’s Capability Maturity Model Integration (CMMI), so we can view the whole system as a continuum.The current Voluntary Voting System Guidelines (VVSG) is not a robust set of requirements that can be rigorously verified. Under the current system, rogue states can actually subvert the will of their people and also disenfranchise the citizens of other states in the process. The Election Assistance Commission (EAC) was created by the Help America Vote Act of 2002 (HAVA). HAVA was passed in response to the contentious 2000 presidential election. By commissioning the VVSG as guidelines rather than requirements, the EAC is respecting the constitutional power of the states in the 12th amendment. There is a constitutional solution to the problem.What is needed is for Congress to call a convention of states like the “Colorado River Compact” (not an Article V convention). By so doing, experts from the several states could hammer out real requirements that could be verified. The VVSG could become the VSR (Voting System Requirements). This would elevate the certification process to a uniform level without disrespecting federalism. By recurring review, like the Colorado River Compact, exposure of weak links can be addressed, and the states can hold each other responsible. The first engineering step is always the most important one, and that’s what this would be.RequirementsIn laymen’s terms, good engineering practice starts with a set of requirements and ends with verifying that those requirements have been met. Every software development organization has its own methodology built upon development stages to satisfy themselves that they have built what was required. The degree of rigor required to satisfy customer(s) varies according to the product’s use and the nature of the customer base. Companies like Intel and Oracle compete for market share. Once a sale is made, the technical support for the products is equally important as the engineering that built them. On the other hand, the engineering for human flight or to send a satellite to another galactic body is unforgiving. National pride can suffer with a big waste of taxpayer money. Even worse, people can die – especially with human flight.It became immediately apparent that there were not enough artifacts to make a solid conclusion about the maturity of the Dominion software, but there were enough to realize that the engineering that had been done was not adequate for a system that has the repercussions of a voting system. When satellites are launched, there is danger to perhaps tens of people on the launchpad and surroundings. Those systems are required to meet CMMI level 3 management and use IEEE 1012 standards. Human flight can threaten the lives of hundreds and must meet CMMI level 5 engineering and meet DO-178B standards. It makes sense that voting systems that can adversely affect the lives of literally all Americans, present and future, should far surpass the standards for human flight. They don’t even come close.Software Audit of Dominion Democracy SuiteThe first document produced for this study was the “Audit of Code Reviews for Dominion Democracy Suite.” It was discovered in the reviews from SLI Compliance that the Dominion Democracy Suite is composed of over 1.5 million lines of C/C++ and over 1.8 million lines of C#. Having written about that many lines of C/C++ in my career as an operating systems kernel developer, I had two immediate reactions to this: What on earth are they doing in a voting system that requires that much code? I read the entire VVSG to get a handle on the “requirements,” and there is nothing there that suggests this magnitude. The main thrust of a voting system should be simple tabulation that should really be at most a couple hundred thousand lines of C/C++ and perhaps the same amount of C# - even with support for a Windows GUI or CGI browser interface. This means there is a lot of unnecessary complexity that is not driven by actual requirements. That increases risk. I give odds that the system has several “memory leaks.” Even the best developers can fall prey to a memory leak when writing C/C++. Memory leaks are one of the prime security exposures for a hacker to exploit. One coding convention used to combat this kind of error is to use a “lint” tool that does static code analysis. The newer “splint” does a more effective job of finding potential memory leaks, but it is prone to false positives. A developer can use “tags” to override the tool, so an insider threat can easily defeat the tool. A review of Dominion’s coding conventions showed no evidence of the use of such a too

Keywords:

Category:

City:

State:

Country: Unknown

Currency :

Avg product price:

Platform: WooCommerce

Technologies used: Facebook SDK, reCAPTCHA

Contact page: https://www.dhaugh.com/contact-us

Social links: