TESTS NEEDED BEFORE COMMITTING OVER 600 MILLION DOLLARS

A new report, "Review of Firearms Registration, TR1994-9e,"
commissioned by the Research Section of the Department of
Justice, has revealed that there are severe problems hidden
within Canada's firearms existing registration system.

Both the accuracy and the cost-effectiveness of the existing
registration system have been called into question by the report,
and now require further checking.  The report demonstrates that
the database is seriously defective--so changes to hardware or
software will not "fix" it.  The old rule of computers: "Garbage
in, garbage out!"--is still true.  A check is necessary to know
what changes are needed to make the system cost-effective.

The system was instituted in 1934, and no cost-effectiveness
check of it has ever been done.  If the system is working well,
Parliament needs to know that; if it is only a huge waste of
scarce and costly police resources, Parliament needs to know
that, and needs to know if or how it can be made cost-effective.

These suggested tests will be inadequate to give good statistical
data, but can be done quickly and cheaply.  Results will strongly
indicate whether or not more far-reaching tests should be done.
Test II is designed to check for the rate at which records "go
stale" [become errors in the system, TR-1994-9e, page x, para 2],
a variable dependent on the time records have been in the system.

TEST I

1.  Obtain from the RWRS computer data base a printout of the
first 100 registration certificates found dated 2 to 3 years ago,
the first 100 dated 4 to 5 years ago, plus 100 from 6 to 7, plus
100 from 8 to 9, and so on to a final 100 with 20 from the batch
dating from 20 to 24 years ago, 20 from 25 to 29 years ago, 20
from 30 to 34 years ago, and 20 from 35 to 60 years ago.  (Some
records in the system may date back 60 years, to 1934.)  The
search should be controlled to prevent the name of each chosen
possessor from being chosen again during the search.

2.  Send a police officer or other messenger to the address given
on each registration certificate, and have him learn if the
person still lives at that address, if the firearm is still in
that person's possession, and if the firearm identifying
information (found on each registration certificate in the 8 data
fields purporting to uniquely identify that firearm) is all
correct.

3.  Present a report to Parliament on the percentage of
certificate holders who no longer reside at the given address and
the percentage of registered firearms that are no longer in the
possession of the registered possessors who do still live at the
registered address.  Also, each of the 8 "firearm-identifying"
data fields on each registration certificate should be checked
against the actual firearm, and the percentage of errors found in
each data field should form part of the report.

Data from the "identifying" data fields will be difficult,
because there are many "right" entires for the same datum (see
the RCMP's National Firearms Manual, Appendix 4-5, for 118 pages
of "common" identification examples).  The test should show up
gross errors such as recording "Model" number as the serial
number; "Model" as the "Make," and vice versa; year of production
as "Model:"; rifles, submachine guns and machine guns registered
as "HG" (HandGuns), etc., etc.

The most severe problems in doing such a check will be preventing
the firearms control bureaucracy from burying the results in
secrecy or "fudging" the data--for example, by tracing a person
who has moved without notifying the system, and then recording
his registration as correct, although it wasn't.

It would be worthwhile (if possible) to delegate full control of
this task to a group with no vested interest in the registration
system, such as the Auditor-General's office or the Library of
Parliament.  That aspect should be researched.

Whether or not "clear standards" [TR1994-9e, page ix, para 1 and
2, plus page xv para 1] are possible also needs to be tested:

TEST II

1.  Obtain 20O photocopies of complete registration
certificates (RWRS records), selected randomly from the RWRS
files.  Make enough photocopies of the 200 RWRS records to give
one set to each available firearms expert, and give one set to
each of them.

2.  Tell the experts that in future, there will only be one
"correct" entry per data field for each "Make", "Model", "Type",
"Action", and "Calibre" entry.

3.  Explain that where the data field for "Calibre" shows ".380
ACP" in one registration certificate and ".38", ".380", ".380
CAP", "9MM", "9MM CORTO", "9MM KURZ" "9MM BROWNING", or "9MM
BROWNING SHORT" on others of the same "Calibre," 7 of the 8
entries are wrong--because all are describing the same cartridge,
and there can only be one "correct" entry which positively
identifies that exact cartridge--without specifying which entries
are "incorrect" or contain insufficient identifying data.

4.  Give each expert a highlighter pen, and have them highlight
each "Make", "Model", "Type", "Action", and "Calibre" entry in
the 200 copies which is, in their opinion, "incorrect."

5.  Gather each RWRS record for a given firearm, and compare
which entries have been highlighted as "incorrect."