The premise in ballistics is that all these casings might be considered “unique.”
This “new look” at ballistics from the Aaron Hernandez conviction is similar to the back and forth decades-long argument against the now debunked “hair matching” in four areas:
- Identifying people and objects to a single source. either being a specific gun or person.
- Both have been forensic “workhorses” for the FBI and their progeny in US police crime labs.
- Police crime lab communities stands fast that all is “well.”
- Final decision-making usually falls on non-science trained judges and lawyers to decide what threshold of proof are sufficient for our justice systems.
excerpt from the Boston Globe:
The scrutiny now includes the field of firearms analysis, often called ballistics, after a report last month by the President’s Council of Advisors on Science and Technology questioned whether the science of analyzing firearms is sound enough to support the standard of proof that is constitutionally required for a criminal conviction.
The report has already sent ripples throughout legal and law enforcement communities, including in Massachusetts, as lawyers in one of the state’s most high-profile cases – the Suffolk County murder trial of former New England Patriots player Aaron Hernandez – asked a judge earlier this month to toss expert testimony related to a firearm police have linked to the case.
FORENSICS and LAW in FOCUS @ CSIDDS | News and Trends 
ATF’s response to the PCAST report rates a “mostly true.” Matching bullets to guns is a valid, legitimate discipline. There are some guns that leave very few identifying marks on a projectile (like a new S&W .38) and some that leave marks that are not reproducible (like a Clerk First .32), but for the most part, the striations imparted to projectiles are unique and reproducible. This kind of firearms identification (FAID) is the benchmark. Now, when it comes to matching a screwdriver to a tool mark, that is far more subjective. In between is identifying the marks left on cartridge cases. There is simply far less information to be had than can be found on projectiles. The fact that both projectiles and cartridge cases can be compared by a computer to a database indicates that there is science in there, but like much of forensic science, the discipline is largely based on pattern recognition and pattern matching. PCAST made the mistake of lumping all “feature comparison” disciplines together. Microscopic hair analysis is useful for exclusion, i.e., screening, but most forensic scientists agree that it is outdated and requires DNA analysis to be meaningful. Bitemarks need to be eradicated. Putting FAID in the same class as bitemarks is way wide of the mark. Generally, generalizations lead to error. The ATF’s response, though it is a tad emotional, is pretty close to accurate. It can be found here. https://www.theiai.org/president/20160921_ATF_PCAST_Response.pdf
Of course ditto on the bitemarks. No valid database existed until the U of Buffalo studied dentition match rates a few years ago. The PCAST winds are blowing that such rates will be expected for other matching genres. From what you said, it seems the computer projectile matching makes the exclusion phase less risky. I do think that the high ballistic examiner reliability testing results is misused as a substitute for getting to the final testing for Type I and II errors, etc.