Ballistic Imaging: Fantasy, Spin and Simple Solutions
Introduction
One of the newest controversies in the gun control debate is the issue of "ballistic imaging" (sometimes incorrectly referred to as "ballistic fingerprinting"). Ballistic imaging technology has been around for awhile. The U.S. Federal Government has been working on an imaging network (The National Integrated Ballistic Information Network - NIBIN) for about ten years. This system matches crime scene bullets and brass casings against a database of bullet/casing records. Administered by the BATF (Bureau of Alcohol, Tobacco & Firearms) this system allows law enforcement to "trace" ballistic evidence linking guns to crimes, and in some cases, individuals. Law enforcement routinely adds to the system when they do ballistic testing of guns and bullets found at crime scenes. As currently established, the system focuses exclusively on data generated by criminal investigations. Congress had barred its use to build a gun registration database.
There is now a campaign being waged to change both the nature and the scope of the Ballistic Imaging Database. Lead by Sen. Charles Schumer, D-N.Y., and the Brady Campaign to Prevent Gun Violence, proposals are being put forward to make the Ballistic Imaging Database a universal database of all bullets and casings from all guns in the US. This process would start by extending a Maryland law mandating that a fired bullet and shell casing be submitted with each new handgun sold. Carried to its logical conclusion the program would eventually be expanded to all gun sales and ater be made a requirement of gun ownership. These proposals would change NIBIN from a "crime only" database to an "all gun, all gun-owner" registration database. Proponents of this multi-billion dollar expansion of NIBIN have made extensive claims about the ease and utility of the project:
"The [sniper] shootings are a perfect example of how valuable complete ballistic fingerprinting would be", Coalition to Stop Gun Violence.
(Ballistic Imaging would have) "solved this crime after the first shooting.": Eric Gorovitz, policy director for the Washington-based Coalition to Stop Gun Violence
"Doesn't it make sense for us to give law enforcement the tools they need in order to solve such crimes?": Sarah Brady of the Brady Campaign.
"It's just a smart, safe, reasonable thing to do": Lesley Heilman of the Brady Campaign. "It's not gun control; it's just a law enforcement tool."
"It's not perfect. It's not DNA. But it's close": Eric Gorovitz, policy director for the Washington-based Coalition to Stop Gun Violence.
"There’s nothing complicated about constructing a national "ballistic fingerprint" bank. Every new gun would be test-fired and its unique fingerprint would be indexed into a computer databank. If the gun were used in a crime, any recovered bullet could be quickly linked to the gun, perhaps to the gun owner or at least to a trail of possession.": Dick Meyer, a veteran political and investigative producer for CBS News.
This article examines Ballistic Imaging and addresses these claims.
Ballistic Imaging Explained
Ballistic Imaging is a relatively new computer facilitated technology used to record information on bullets and their brass casings. The name itself is somewhat misleading since the technology focuses on the spent bullet and empty casing rather than the actual "ballistics" of the bullet. The physical basis of Ballistic Imaging is the marks left on the bullet or casing when a gun is fired. These marks, left on the bullet (projectile) by the rifling of the barrel, and on the brass casing by the chamber, ejector and firing pin, are believed to be unique to each gun. In its crude form, this is as simple as differentiating a 45 cal bullet from a 22 cal bullet or a pistol casing from a rifle casing. This crude level is where the Police get their information about what "type" of firearm has been used in a crime. This sort of crude "profiling" is now routine Police work and is well established as a valuable crime-fighting tool.
At a more sophisticated level, the marks on the bullet and brass casing can be examined microscopically to determine if they match a test bullet and casing fired by the gun in question. It is extremely important to note that this individual matching takes place at the microscopic level. Modern firearm manufacturers routine produce arms that are standardized to 1/1,000 and 1/10,000 of an inch tolerances. It is a standard manufacturing practice to make each gun as nearly identical to all other guns of the same model as possible.
Much of the confusion in the current Ballistic Imaging debate stems from the difference between the crude and sophisticated applications of Ballistic Imaging. Any knowledgeable person can identify common casings and bullets at a glance. Matching a particular bullet or casing to a particular gun requires sophisticated equipment and a high level of training. The normal process of bullet/casing identification starts with a computer search to isolate "likely matches" (cold hits). A highly trained ballistic expert then makes a manual examination using a special microscope.
We can note a few complexities in this task. Some guns, for example revolvers do not "eject" a casing and consequently do not leave a casing at a crime scene unless a criminal manually ejects them. Other guns, such as shotguns and smoothbore muskets, have no rifling and do not leave distinctive marks on the bullet. Unlike fingerprints and DNA the microscopic markings left on bullets and casings can change. Criminals rarely leave their gun at a crime scene.
The National Database
A national ballistic imaging database already exists. In 1997 the Bureau of Alcohol Tobacco and Firearms (BATF) and the Federal Bureau of Investigation (FBI) entered into an agreement creating the National Integrated Ballistics Information Network (NIBIN) Board. The NIBIN Board is comprised of a BATF, a FBI and a State/local representative. The Board's goal is to unify Federal efforts to deploy Ballistic Imaging Technology. Currently the Board is working to integrate IBIS and DRUGFIRE, the two agencies separate software systems. The two systems both began in 1993 and contain some 800,000 images. Together these two systems have produced 8,000 "cold hits" (likely matches) in over 16,000 cases. Director Magaw of BATF has stated that "Computer ballistic imaging technology is the most important forensic advancement since the development of the comparison microscope over 70 years ago." Director Freeh of the FBI stated, "This joint effort is a key component in ongoing programs aimed at permanently reducing gun crime."
There is no question that the current "crime only" database has been a successful law enforcement tool. This database, in its current form, is widely supported by both anti- and pro-gun forces. The current controversy is not about the existence or development of this "crime only" database. The controversy is whether this database should be extended and modified to become a "registration" database containing computerized records on all US guns (current proposals focus on new handguns, but the intent of expansion to all guns is clear). Such an expansion would involve huge costs to the government, firearm manufacturers and customers. Such an expansion would also have to address the legal "chain of evidence" issues involved in storage, recording and matching hundreds of millions of records at a level suitable for evidence in a court.
The creation of a ballistic imaging database requires the recording of microscopic information from a bullet or casing as a computer record and the storage of that record on a computer. This is not a simple task and there are a number of serious technical challenges that need to be addressed if evidence quality information is desired. This article will not address these highly technical issues but it should be noted that they exist. There are also significant database indexing and reliability problems that increase as the size of the database increases. The problems involved in creating and maintaining accuracy in a large database are also significant issues but will not be addressed in this article.
The California Report
On October 5, 2001, the Bureau of Forensic Services, California Department of Justice, issued a report examining the feasibility of converting a "crime only" database to a "registration" database. (Technical Evaluation: Feasibility of a Ballistics Imaging Database for All New Handgun Sales, currently available at http://www.nssf.org/PDF/CA_study.pdf) Since this is a recent report prepared by firearms experts intimately familiar with ballistic imaging’s use in crime control we will examine its findings in some detail.
The overall conclusion of this report was that "automated computer matching systems do not provide conclusive results" requiring "potential candidates be manually reviewed". The experts further concluded that a database including all new handguns would grow by about 108,000 entries a year (California only), resulting in a database where "the number of candidate cases would be so large as to be impractical and will likely create logistic complications so great that that they can not be effectively addressed". An enlarged "registration" database (they are only considering including new handguns) would expand to the point were the computers would generate so many "likely matches" that it would swamp the examiners, generating a statistical nightmare.
The report also addressed a number of technical problems associated with ballistic
imaging. Briefly summarized here, these problems demonstrate just how far this
technology is from simplistic got-the-evidence,solved-the-crime rhetoric.
- The system requires highly trained experts for entry, searching and verification "use of technicians typically results in higher numbers of false positives",
- It is "unknown at this time whether or not the algorithm (software) can successfully ID a cartridge case fired after typical break-in and wear have occurred" (note: this point completely negates the usefulness of a "sample at time of sale" database, the current proposal),
- It acknowledges that "firearms generated markings on cartridge casings can change with use and can also be readily altered by the user. They are not permanently defined". A 600 round test (no details about the type of ammunition) saw "signature degradation", but was inconclusive. They also noted that it was "very common to fire 2 boxes or 100 rounds of ammunition for (a single) practice".
- "Changing ballistic fingerprint on casing, enough to defeat recognition took less than 5 minutes",
- "Cartridge casings from different manufacturers of ammunition may be marked differently by a single firearm such that they may not correlate",
- "as progressively large numbers of similarly produced firearms (firearms of the same make and model) are entered, images with similar signatures should be expected to make it more difficult to find a link",
- "fired cartridge casings are much easier to correlate than fired bullets"
(Note: crime scene bullets that hit something normally deform making ballistic
matching difficult or impossible. Bullets that don’t "hit something" disappear in
the distance making testing impossible), - "not all firearms generate markings on cartridge casings that can be identified back to the firearm",
- "rim fires not considered…utility of such imaging for unique individual characteristics has not yet been independently verified",
- "Regulations restricting firearms or firearms ammunition types lead to large numbers of similar casings and bullets (same make, model, ammo) this causes real matches to be lost in a host of "close matches"",
Since the problem of separating "true matches" from "close matches" is one of the basic flaws in a "registration" database, we will examine the results of the "similar weapons" test in detail. The California Highway Patrol lent 792 identical pistols to testers. The pistols were then test-fired under laboratory conditions (note: lab conditions tend to produce results that are clearer than evidence retrieved from the field). Four different cartridge types (all S&W .40 calibre) were used in the test. Over 2000 fired casings were used in the test (note: apparently, there was no bullet testing. Bullets (projectiles) are a far more common form of evidence.)
The results of this test were:
- When identical ammunition was used the system missed 38% of the ballistic matches,
- When ammunition from different manufacturers was used, the system missed 62.5% of the ballistic matches.
From this, and other evidence, the examiners concluded that when examining casings from the same gun but with different ammunition "78% of the evidence cases that should have a counterpart in the proposed database, may not be detected". They also noted that some primers on new cartridges had similar manufacturing marks that could be construed as coming from the breech face of the firearm and generate false matches.
In a footnote, the examiners raised another issue stating that, "many agencies are concerned with good numbers, rather than accuracy. Are agencies funded by the so-called success rate of the automated system going to accurately represent their true cold hit (likely match) statistics?".
California's pro gun-control Attorney General Bill Lockyer responded to these findings by silencing the experts. One panel member said he was gagged by the AG's Office and told not to speak of the study or about ballistic imaging.
An example of how difficult ballistic identification can be is the experience of Shelby County Prosecutor John Campbell. In 1997 he had the Martin Luther King assassination rifle re-examined to determine if James Earl Ray deserved a new trial. Despite eighteen test firings they could not match the rifle to the bullet that killed King. Each of the bullets had different markings. "Every test bullet was different because it was going over plating created by the previous bullet," he said. Campbell informed the court that "the history of this rifle testing and the barrel in this case has shown that it doesn't mark the bullet the same way every time it's fired."
The BATF Report
Some commentators in the controversy have referred to a July 2001 report by the BATF that supports ballistic imaging. The report said the technology could track guns and had provided valuable leads for investigators helping to solve many violent firearm crimes, "which would not have been solved without it". The report credited the current database with producing 8,800 ballistics matches (Note: they are talking about "likely matches" or "cold hits" not actual matches), from 17,600 crime scenes during 2000-2001. A May report by the BATF also praised the system for providing "significant investigative leads" in thousands of crimes over the past decade.
These reports clearly state the value of the current "crime only" database. However, there is no controversy regarding the usefulness of the current database. The controversy is about the expansion of the "crime only" database into a "registration" database. Accordingly, these BATF Reports have no bearing on the current debate and are something of a red herring.
The BATF’s past history with databases is also problematic. On October 18, 1995, Thomas A. Busey, then Chief of the National Firearms Act Branch and the official custodian of the National Firearms Registration and Transfer Record stated that there were problems with the gun registrations database. He stated that although the Bureau had been testifying that the database was 100% accurate the actual accuracy was between 49% and 50%. Every prosecution and forfeiture action brought by the United States and involving an allegedly unregistered NFA firearm had required testimony under oath that no record of the registration of the firearm in question was found.
When this story leaked in 1996 a U.S. Department of Justice investigation learned that the NFT&TR inaccuracy problems had been the subject of internal BATF discussion since at least 1979. BATF's files contained numerous minutes of meetings, statistical studies, memoranda and correspondence admitting the problem. Consequently, the investigation concluded that the BATF had been committing "institutional perjury and obstruction of justice" since 1979. These seventeen years of "institutional perjury" affected 60,000 cases nationwide.
The Effectiveness of Current "Registration" Databases
Maryland introduced its own ballistic imaging system in 2000. Under its new Registration Law every new handgun sold in the state must be accompanied by a spent shell casings for input into the state ballistic imaging database. Maryland budget figures indicate that approximately $5 million has been spent on the system (note: this is the cost of the central system; it does not include the larger distributed costs incurred by police and the courts.) The database contains over 11,000 imaged cartridges and has been queried 155 times. It has not been responsible for solving any crimes. This works out to $454 per record and $32,000 per query. We can explain part of these figures by attributing much of the cost to start-up costs. However, we can also note that the current system only records very easy to track new handgun sales. Applying the system to guns already in public hands would be much harder and more expensive.
This lack of success has also been repeated in New York where thousands of cartridges have been entered into their database with no criminal prosecutions. These "registration" database results are very similar to the results achieved by gun and gun owner databases. Hawaii, which registers both guns and gun owners, has spent tens of thousands of police man-hours administering these laws without positively identifying a criminal.
This lack of success is typical of "registration" databases and is tied to a "structural" problem. Simply put, it’s a numbers game. The FBI Uniform Crime Statistics note that less than 1% of guns are ever used in a crime. This fact "structures" the problem by ensuring that any "registration" database (guns or bullet/casings) will waste 99% of its resources tracking non-criminal guns. Even this dismal number makes the incorrect assumption that guns used by criminals will be registered. A 1997 U.S. Department of Justice study found that 40% of criminals in state prisons got their gun from an untraceable illegal source.
The California State Sheriff's Association, has come out strongly against ballistic imaging being used to make a "registration" database. They raised concerns about cost effectiveness and warned that the proposed California legislation will spread already undermanned police agencies even thinner.
Comments
The first observation we can make is how dysfunctional it is to have one simplistic answer to complex issues. For example, if we ask the question "how effective and efficient is ballistic imaging as a crime fighting tool?" We would get three very different answers:
If we are talking about its use on a crime scene were police are tracking a particular bullet/casing to a few "suspect" guns, then the answer is very effective and very efficient. Such tests provide timely, accurate results and do so with a very efficient use of resources.
If we are talking about ballistic imaging's use for a "crime only" database to trace guns to other crimes or criminals, the answer is,"somewhat effective, and fairly expensive", per positive link established. However, this process is probably still cost-effective, especially if one considers how often gangs "pass a gun around" (note: data on cost-per-positive ID is not available).
If we are talking about ballistic imaging’s use as a "registration" database, the answer becomes "extremely ineffective and extremely expensive". The system is a failure because of the technical problems discussed above, and because of the structural problem inherent in any "register everybody to catch the criminals" database.
For gun databases, more than 99% of the effort lands on law-abiding citizens, without affecting criminals. This is a problem of "focus". The closer to the crime and the more narrowly focused the effort, the more effective and efficient will be the result. As we disperse the effort to less related activities and to more and more associated activities, the effectiveness and efficiency of our efforts drops dramatically.
This raises the question of why this proposal is getting so much support. Much of this can be answered by looking at incentives (we can note that some of these incentives relate to fighting crime, while others are merely "gaming the issue" in pursuit of other agendas):
1. Technological Utopianism: This is another attempt to solve complex people problems with technology. It appeals to those who want simple, let science do it, solutions. It also appeals to those who don’t know much about science, or how complex the relationship is between science and people.
2. Sound Bite Politics: It wraps rhetoric in the mantle of science. Promotion is simple and simplistic; opposition requires an examination of the details and some thought. Very appealing to those more interested in political grandstanding then in actual results.
3. Funding and Authority: Putting a "registration" database in place will bring billions of dollars in funding and thousands of new employees. The new system will require a whole new set of laws and new criminal offenses, then more laws and criminal offenses to "close the loopholes".
To give some rough idea of how big this motivation is, we might consider Canada’s brand-new (1995) gun owner/gun registration system (far easier than recording microscopic data). The Registration Centre currently contains less than 3 million registrations. The Centre has cost a billion dollars to establish and employs 1900 people. There is a 90% error rate in the 27 page applications and a 42% error rate in firearm description(RCMP data). Somewhere around 20% of all Canadians are now felons for paperwork violations. The registration system has never solved a single crime. If we note that the vast majority of costs are enforcement costs distributed onto police and the courts, we can get some idea of the costs and bureaucracy involved in registering 260 million American guns.
4. Gun Registration: The system would require the registration of every gun and gun owner to work. This makes it very attractive to those who desire gun registration. The fact that it will not work is – strangely enough – a plus for such people. The vast expense and the fact it doesn’t work can be used to "prove" that the only "sensible" policy is gun elimination. The registration database can then be used to track down law abiding gun owners and disarm them. As Britain, Australia, Canada and numerous US jurisdictions have found, this massively expensive project will actually encourage crime. However, for those who want to social engineer an end to this part of the American heritage, it is extremely attractive.
5. Crime Shifting: This is now a well-established trend in the American Criminal Justice System. As agencies are caught between a public demanding an immediate end to crime and the difficulties involved in stopping real criminals committing real crimes they tend to seek bureaucratic solutions. The easiest way to do this is to "shift the crime", by creating an ever wider net of paperwork and regulatory crimes that are associated with the original problem. This disperses resources and effort onto people with no criminal intent or effect, a move that actually reduces the efforts against real criminals.
However, it is far, far easier to convict law-abiding people committing "technicality crimes". The agency's numbers immediately go up with the success rate triumphed as "working" and the increase in "crime numbers" justifying funding. The shift from hard-to-catch violent criminals to easy-to-catch technical/paper violations creates "beautiful statistics". This process is very, very popular with "tough on crime" politicians and agency bureaucrats.
----------------------------------------------------------------------------------
Bruce Gold is a professional researcher and analyst with a Master's degree in Intellectual History and a Master's degree in Public Administration. He is the author of InfoWar in Cyberspace: Researcher on the Net. The book is a primer for students, researchers and Internet activists. It can be found at: http://www.booklocker.com/bookpages/brucegold01.html.
Bruce can be reached at: bgold@alumni.uvic.ca