Crime Gun Intelligence Bogging Down?
It’s all about descriptive meta data - the best cranberries are selected by their bounce and ability to float not by their image.
What is Crime Gun Intelligence (CGI)?
The National Public Safety Partnership, which is supported by the Bureau of Justice Assistance, reports that: “Crime Gun Intelligence combines all available information on crime guns and shootings to identify and disrupt the shooting cycle. CGI reveals leads not otherwise available to assist in identifying offenders who are illegally purchasing or transferring firearms and the offenders who use them to commit violent crimes.” [1]
The definition of Crime Gun Intelligence (CGI) is clear, concise, and on point. My wholehearted support leads me to the question raised in the title above.
According to the definition: Crime Gun Intelligence combines ALL available information on crime guns and shootings, and it reveals leads not otherwise available about offenders who illegally acquire or transfer firearms and those who use them to commit violent crimes.
There’s certainly a lot of information out there to be had. Much of it is being captured in various data systems, by many agencies at different levels of government.
The steps to CGI are progressive, moving from data - to information - to intelligence.
The primary CGI building block
The local law enforcement agency information about criminal shooting events - the who, what, where, when, how, etc. - is the primary building block of CGI and should be accessible to the authorized parties who need it.
Over the last two years, I assisted a research team in studying a 15-year Statewide Program in New Jersey to Generate Enhanced Investigative Leads on Crime Gun Violence.[2]
One of the many innovative programs examined was the “NJ POP Collective”, a project named by a former analyst referring to “pins on paper”. The NJ POP effort created a statewide database of shooting incident data that included reporting information, shooting location, weapon, victim, and suspect data.
Over 500 state and local law enforcement agencies in New Jersey fall under the umbrella of the NJ POP collective project. What if every state had one? For example, Texas has the most state and local law enforcement agencies – over 1,900. Rhode Island has the least - just under 50. [3]
Two additional basic building blocks of CGI
As stated, the data and information collected locally about criminal shooting events represent the 1st building block of CGI. Following right behind in the number 2 and 3 positions would be data from the crime gun itself, the internal ballistics data left on fired ammunition components, and the external manufacturer’s identification data (e.g. make, model, serial number) for tracing the crime gun’s history of acquisition and disposition.
Additionally, other forensic data may be present and collected from a crime gun as the next level of CGI (e.g. DNA, latent fingerprints, and trace evidence).
The CGI sharing and management of ballistic data on local, state/provincial, and national levels
Globally, networks exist to manage and share CGI on local, state/provincial, and national levels. For example, in the United States, ATF’s National Integrated Ballistic Information Network (NIBIN) was intended to produce and share firearm information on the local, state, and national levels. It has two essential applications, one as an investigative tool and the other as a forensic tool.
NIBIN’s primary focus today is as an investigative tool, to produce timely investigative NIBIN leads.
“A NIBIN lead is an unconfirmed (presumptive) potential association of firearm evidence based on a correlation review of digital images in the NIBIN database by either a firearm examiner or a trained NIBIN technician. There is a high probability that a subsequent microscopic comparison, by a firearm examiner, will confirm the association between the firearm ballistic evidence.”[4]
NIBIN’s use as a forensic tool remains important for searching and confirming potential matches as "hits" for presentation as evidence in court.
“NIBIN hits are confirmed matches by a certified firearm examiner conducting an in-laboratory microscopic examination of actual physical evidence to confirm those items of ballistic evidence were fired from the same firearm. This type of actual physical examination was necessary to establish a foundation for the admissibility of courtroom testimony as NIBIN hits are often crucial evidence in the prosecution of murders, assaults, and other firearm offenses.”[5]
“The process for confirming a NIBIN hit is labor-intensive, requiring the transportation of physical evidence to the lab and the services of certified firearm examiners; expensive requiring access to laboratory facilities and “bench time” in the lab with specialized equipment; and most significantly, time-consuming.”[6]
Conversely, NIBIN leads can be processed more quickly as timely and actionable crime gun intelligence to help advance investigations.
ATF began pioneering ballistic data-sharing networks near the end of 1992. At that time, no other ballistic technology systems were commercially produced and marketed in the United States that processed fired bullet evidence.
Twenty-some-odd years later several new ballistic technologies have entered the global market. Some of these systems are currently in use in the United States. They do not match the same technical specifications as the standard technology approved by ATF to operate on the NIBIN network.
Many of the off-NIBIN-net systems are used to perform “triage”. When multiple pieces of fired evidence are collected from crime scenes triage sorting helps operators identify if one or more guns were involved. They select the best exhibit representing each gun to send to a NIBIN site for NIBIN entry. Some of these systems perform off-NIBIN-net ballistics analysis to generate CGI.
One off-NIBIN-net system, for example, has been in operation in a busy county in a large western state for almost 10 years servicing over 30 incorporated city police agencies. At the same time, over 25 other cities and counties in that state are NIBIN sites operating NIBIN-net-compatible systems.
CGI sharing concerns with the data collecting in off-NIBIN-net systems
From time to time, I’ve been asked by interested parties about my thoughts on the off-NIBIN-net systems especially recently as their number of deployments has increased.
I know many of the people involved in the business of producing crime-solving CGI technology tools. They are serious and passionate about the important criminal justice work that their tools contribute to. They all tell me that they want to help and do more. So do I, and I hope this article is evaluated in the context of helping to do more good works.
That said, speaking as a former criminal investigator, I never gave thought to the brand of technology that was used to provide me with a potential crime-solving investigative lead. I needed and appreciated the lead. All I could hope for was that it be timely and have a high probability of being correct so that I wouldn’t be chasing shadows into dead ends. I would follow up on that lead to exhaust it to its positive or negative conclusion.
The off-NIBIN-net systems are not collecting and processing CGI according to the expectations raised in the definition: “Crime Gun Intelligence combines all available information on crime guns and shootings to identify and disrupt the shooting cycle.”. Neither are the NIBIN sites operating in the same areas where off-NIBIN-net sites are operating.
The off-NIBIN-net sites are not getting “all available” information as they don’t exchange data with NIBIN sites, and vice versa. All to say that one day, when a key investigative lead falls through the cracks and is overlooked, a victim may be denied justice, their loved one’s resolution, and their neighbors a restored sense of peace.
As the number of NIBIN and off-NIBIN-net sites increase within a given area so does the potential for more CGI to fall through the cracks when it shouldn’t be the case.
I’ll try and explain more as I go on.
Figuratively speaking
Living on Cape Cod for seven years, I learned that the best and freshest cranberries bounce when dropped and float to the top when the bogs are flooded during harvesting. The older and damaged berries don’t do either. Moreover, the best berries can be of different colors, and again they are selected by their bounce and buoyancy, not the image they reflect.
Like any other database, you must put something in before you can get something out. It’s reasonable to assume that every ballistic system developer has built their system to help users input the metadata required to track and reference the system’s outputs. This input data includes case numbers, dates, locations, agency names, gun makes and models, ammunition makes and calibers, serial numbers, etc.
The input and output data become the information that can expand into valuable intelligence for investigative purposes. This non-image data has metadata descriptors that can easily be exported in standard data formats to an Application Programming Interface (API), commercial intelligence software, a proprietary database, or a simple spreadsheet for further management and CGI generation.
I propose that we share the ballistic data that can easily bounce from one system into another and float to the top as information and intelligence of investigative value, rather than trying to reproduce or “shoehorn” ballistic images from one system to another. It’s all about the descriptive metadata – not the scratch and dent markings depicted in the ballistic images.
What can typically go right and go wrong with certain CGI data-sharing efforts?
It depends upon the perspective from which the automated ballistics technology is viewed. Is it a tool for investigative or forensic purposes?
When NIBIN is focused as an investigative tool for developing actional CGI in the form of investigative leads as discussed above, the forecast for sharing CGI is good. The NIBIN “input data and output data” needed for this purpose does not involve ballistic images it is simple textual data of many case and evidence-related descriptors that can be easily shared in a standard data format.
When NIBIN is focused as a forensic tool, the forecast for sharing CGI is not good. The discussions typically lead toward standardizing the image capture, processing, and automated comparison capabilities for universal interoperability. In other words, all ballistic images can be processed and searched with the same degree of accuracy on any system that meets the standard.
Digging down a little deeper, the automated search and matching process is called the data correlation process. It is the heart of all technologies that claim to be automated ballistics comparison systems. It involves the “secret sauce” - the methods of data extraction from the images and the matching algorithms that are applied which can set one brand apart from another in terms of speed and accuracy of correlation. It can also set brands apart in terms of cost of ownership. The more it costs the maker to produce, the more it costs the consumer to buy.
These standardization/interoperability discussions tend to call for studies and testing and forays into the inner workings of proprietary systems which in turn causes the technology developers deep concerns over losing their intellectual property. This alone often ends up killing what seemed like a good idea. At a minimum, it will cause all stakeholders undue stress, waste money, and delay justice.
Going down this rabbit hole of standardization/interoperability can delay this good idea of sharing CGI data across disparate ballistics data programs. Delay runs the risk that at some point, some investigator, somewhere, won’t get the investigative lead that he or she so desperately needs to solve their case and remove a violent criminal from the community. The result could mean no justice for the victim, no resolution for their loved ones, and no sense of restored peace for their neighbors.
Conclusion
As I have mentioned, I know many people who make and use these crime-solving CGI technology tools. They are all very passionate about the criminal justice work that their tools contribute to. They all tell me that they want to help and do more. So do I, and I hope this article is evaluated in the context of helping to do more.
The steps to CGI are progressive moving from data - to information - to intelligence.
The sharing of CGI generated by ballistic technology does not require the cross-correlation of images nor the re-design of proprietary ballistic technology systems. Let system A process A’s correlations and system B process B’s correlations. Then focus on sharing the descriptive input and output data resulting from those correlations - data that can easily bounce between systems and float to the top as information leading to intelligence of investigative value.
In other words, actionable CGI stems from the descriptive metadata revealing information about the relationships of people, places, and things to crimes involving the use of firearms – providing intelligence as to connections between cases, guns, and suspects.
Importantly the stakeholders must begin to think and act together to facilitate their sharing of the relevant input and output data from their ballistic data systems – through the one national network established for that purpose - NIBIN.
One more thing
There are standard file exchange formats that are used for sharing basic ballistic images. They do not contain the standardized specifications needed for correlation purposes, but they can be used for visualization. Should the operator see something during an on-screen system viewing that could indicate a potential match to another exhibit, there are steps the examiner can take to obtain the physical exhibits in question for comparison or use double-cast clones.
[1] https://www.nationalpublicsafetypartnership.org/Clearinghouse/Resource/794/PSP-Research-Brief-Crime-Gun-Intelligence
[2] https://www.ojp.gov/library/publications/research-15-year-statewide-program-generate-enhanced-investigative-leads-crime
[3] https://bjs.ojp.gov/library/publications/census-state-and-local-law-enforcement-agencies-2018-statistical-tables
[4] https://www.atf.gov/firearms/docs/report/nfcta-volume-ii-part-i/download
[5] https://www.atf.gov/firearms/docs/report/nfcta-volume-ii-part-i/download
[6] https://www.atf.gov/firearms/docs/report/nfcta-volume-ii-part-i/download