Essentially useless.
Let's look at what it might accomplish:
Surely, there isn't much that benefits John Q. Citizen to know Norman Neighbor owns a firearm; what would John do about it? No, the benefit seems restricted to LEOs being able to properly guage risk during the performance of their duties, and solving some crimes.
Suppose we implement
perfect registration. In that world, every firearm transfer must go through a process that establishes the owner and address of every firearm.
Such a database would be able to answer questions such as 'are there any guns at 123 Main Street?' or 'does the owner of car X, license plate AAA 111 own any guns?'. With such answers available, our LEOs ought to be prepared for possible armed resistence.
Or would it work that way?
Suppose John Q. Public owns no firearms; suppose Gary Gunowner, a legal possessor of a registered firearm, is visiting John, and the police get a 911 call from John's house. Will the police know about Gary? Nope. Would a sane LEO rely on static data for his safety? Again, nope. Each contact requires that an officer prepare for a wide range of eventualities; at best the gun database information
might suggest there may be a law-abiding citizen involved, but until every person is tracked 24/7 LEOs cannot eliminate the possibility of a firearm owner being in a place where there are no registered firearms.
A similar situation applies to traffic stops - what if John lends his vehicle to Gary?
LEOs must already act as if a firearm might be present; knowing that no legal firearm should be associated with a person or a place isn't enough, because firearms are mobile.
Even more simply: Gary registered his firearm while he lived at 123 Main Street. Later, Gary moved around the corner to 246 Pine Street, or he moved from Brownville to Greenville. There is
always some lag in updating databases; should LEOs assume 246 Pine Street is gun-free, even if they can confirm Gary is present? Which data should rule - address or person? Should they assume 123 Pine Street is gun-free, even knowing Gary isn't there any more?
Registration just doesn't change anything for LEOs.
Let's look at crime solving. What crimes might this include? It has to be a crime where
(1) the firearm is recovered, at the scene or elsewhere, and associated with the crime
(2) a perpetrator is not arrested immediately
(3) physical evidence on the firearm can connect a person with the crime
(4) the registered owner of the recovered firearm can be located
(5) the registered owner of the firearm can be identified as the person whose physical evidence -- DNA, fingerprints, fibers, whatever -- is associated with the recovered firearm.
If the perpetrator of the crime is arrested in possession of the firearm, the tracing to a registered owner is unnecessary. If the gun is not recovered, tracing is not possible. If the recovered firearm were lost or stolen -- and properly reported, of course! -- step (4) is irrelevant. If the evidence doesn't identify that owner, the crime is usually no nearer being solved.
For this level of benefit, just how does anyone propose to justify the data collection, the administration and security, the inconvenience to the owners, and the cost of the whole thing? Look for example, at Canada's registration program - billions of $Canadian for nearly no benefit.
Look also at the BATF records on who owns machineguns and 'other weapons'. There really are not all that many, and it has been required that all be registered since 1934, yet the database is widely reputed to be quite corrupt and unreliable.
California, where I endure, has the 'all transfers must be recorded' rule. But it's imperfect - not all current transfers really go through a dealer (that's illegal); not all guns which moved here with their owners are registered (legal until 2001, I think, and then only handguns have to be registered); not all guns purchased in California were purchased after the 'all transfers recorded' rule was in effect, or even after the Feds began recording sales (all legal). Smuggling? Theft? Perfection is impossible.
See also
Can Gun Control Work , James B. Jacobs, 2002, Oxford University Press, for a different take but similar conclusion.
He also refers to
http://www.police.govt.nz/resources/1997/review-of-firearms-control/ "The Thorp Report" where New Zealand found registration to be unworkable and dropped it in 1983.
