COURT FILE NO.: 514/05
DATE: 20070621
ONTARIO
SUPERIOR COURT OF JUSTICE
DIVISIONAL COURT
carnwath, greer & swinton JJ.
IN THE MATTER OF the Municipal Freedom of Information and Protection of Privacy Act,
R.S.O. 1990, c. M.56, as amended
B E T W E E N:
TORONTO POLICE SERVICES BOARD
Applicant
- and -
INFORMATION AND PRIVACY COMMISSIONER/ONTARIO and JAMES RANKIN
Respondents
Darrel A. Smith, for the Applicant
William S. Challis, for the Respondent, Information and Privacy Commissioner/Ontario
Tony Wong, for the Respondent, James Rankin
HEARD: April 27, 2007
CARNWATH J.:
[1] James Rankin is a journalist with the Toronto Star. He made two requests to the Toronto Police Services Board (“the Board”) for certain electronic data. The data is contained in the Board’s Criminal Information Processing System (“CIPS”) and in the Board’s Master Name Index (“MANIX”). The Board refused access to the information saying that to fulfill the request would necessitate the creation of records by the Board.
[2] Mr. Rankin appealed the refusals to the Information and Privacy Commissioner/Ontario (“IPC”). The Tribunal member, Frank DeVries (“the Assistant Commissioner”), ordered the Board to respond to the requests by issuing access decisions in accordance with the Municipal Freedom of Information and Protection of Privacy Act, R.S.O. 1990, c. M.56, as amended (“MFIPPA”).
[3] The Board seeks judicial review of the IPC decision. The issue to be decided is whether it was reasonable for the Assistant Commissioner to issue the access decisions without considering whether the records could be produced by means that were “normally used by the institution”.
BACKGROUND
[4] Mr. Rankin requested access to an electronic copy of data contained in the CIPS system. He wanted the names of individuals referred to on that computer record “replaced with randomly-generated, unique numbers, and that only one unique number be used for each individual entered in the database”. In his request, he indicated he did not seek access to personal information or information that could potentially be used to identify individuals.
[5] Mr. Rankin also requested an electronic copy of data contained in the MANIX database, again asking that names be replaced with randomly-generated unique numbers and that only one unique number be used for each individual entered in the database. In addition, he asked that home address information be limited to the full postal code of each individual. Mr. Rankin did not seek access to personal information or information that could potentially be used to identify individuals. He also asked for a list of all fields in the MANIX database, along with descriptions of the data contained in those fields. Mr. Rankin stated that he was not interested in obtaining MANIX software and asked that it be released “in some other, off-the-shelf database format”.
[6] The Board denied both requests for access to information, stating among other things, that the requests “would necessitate creation of records by the institution” and that access could not be provided to the data as specified “as such records do not exist”.
[7] In its initial response to the request, the Board noted that fulfilling a request would necessitate the creation of records by the institution, as the records did not exist. The Board stated that it was not required under MFIPPA to create records.
[8] In response to a Notice of Inquiry from the IPC, the Board made further submissions. It reiterated that the material sought did not exist as a record within the Board’s database. The Board submitted that what the requester sought did not exist as a machine-readable record. Each entry on CIPS was specific to an occurrence and did not provide a unique identifier for all individuals whose information is contained within the database. To replace names with unique numbers would constitute creating a record, according to the Board. It would have to create a new record by introducing an entirely new field to the system, in order to add a unique number in place of each individual’s name. To do so, said the Board, would require the creation of a new algorithm, since it could not be created using software normally used by the Board.
[9] The Board also said the record could not be created with any degree of accuracy and that its preparation would be an enormous hardship requiring time and expense.
[10] I pause to note that the IPC filed three volumes of material divided by fifty-six tabs and totalling six hundred and eighty-nine pages. Of that material, I find only four items that require close scrutiny:
The affidavit of Richard Faulkner, sworn January 27, 2004, found in Vol. 2, p. 222 of the Record;
The affidavit of David Angus, sworn June 7, 2004, found in Vol. 3, p. 477 of the Record;
The “reply” affidavit of Richard Faulkner, sworn September 15, 2004, found in Vol. 3, p. 573 of the Record; and,
Order MO-1989, the decision of the Assistant Commissioner, dated November 7, 2005.
The Affidavit of Richard Faulkner
[11] Mr. Faulkner holds a B.A. in Economics and a Masters in Library Sciences. Since 1981, he has worked as a database analyst, a programmer and a database administrator. At the time of his deposition, he was an IT consultant for the solicitors of the requester. In that capacity, he provided management of the firm’s litigation support, corporate services, intellectual property, library and intranet databases.
[12] For the purposes of his affidavit, he reviewed a “Screen snapshot” of the fields in the modified CIPS database that was released by the Toronto Police to the Toronto Star in June, 2002. He reviewed a printout of the “person investigated card” stored in the MANIX database. He also examined a list of the fields contained in the complete CIPS database.
[13] Mr. Faulkner concluded the CIPS database was “a relational database” because it consisted of multiple tables linked by key fields. His conclusion was derived from his review of the Screen snapshot. He concluded that one table could describe an accused person and store information such as names, sex, date of birth, height, weight, etc. Such a table would have a key field containing a unique identifier (usually a number) which serves to uniquely identify records within the table. A separate table could describe arrests and records and record such information as date, time, circumstances, etc. An accused person’s record could be linked to one or more arrest records by including, as a field in the arrest record, an “accused person ID field” which would provide a link to a record in the accused person table. In the CIPS database, Mr. Faulkner concluded, it “appears that the ‘arrests’ table is linked to the ‘charges’ table through the ‘STAR ARREST ID field’ ”.
[14] Mr. Faulkner concluded, in paragraph six of his affidavit, “if, as I believe, the CIPS database is a relational database, it should be possible for the Toronto Police Services Board to extract from it and produce a record containing the precise information sought by Mr. Jim Rankin”.
[15] Mr. Faulkner then stated that he was aware that the Board had stated that “each entry on CIPS is specific to an occurrence and does not provide a unique identifier for all individuals whose information is contained within the database”. He acknowledged he did not have the opportunity to review the entire CIPS database. He did believe that it would be anomalous for a database of CIPS not to have unique identifier information for all individuals. He added that some of the field names appeared to be person specific unique identifiers. Their existence reinforced his belief that CIPS’ database did have a unique identifier or identifiers for each person in the database.
[16] In Mr. Faulkner’s experience, creating unique randomly-generated numbers to correspond to unique identifiers was not a difficult process. A unique identifier, he said, can be converted into a randomly-generated number through an algorithm which is a relatively simple computer program. He believed that a professional computer programmer would be able to create an algorithm to replace unique identifiers with random numbers with little difficulty.
[17] Mr. Faulkner then turned to the nature of the MANIX database and deposed that he did not have the opportunity to view it. However, he reviewed excerpts from the Direct Data Entry Procedure Manual. From that review, it appeared that MANIX is a database that stores “person investigated cards”. Each “person investigated card” appeared to contain a social insurance number field.
[18] Mr. Faulkner then deposed that he “would expect that the TPS could replace individual social insurance numbers with randomly-generated unique identifiers without difficult. As with CIPS, this could be done through creating a simple algorithm”.
The Affidavit of David Angus
[19] Mr. Angus has been a member of the TPS since October 14, 1980, and is assigned to the Analysis Support Unit. Before 1997, he developed a small stand-alone database which became the basis for CIPS. He was involved in the design and implementation of the underlying tables and database structure, as well as the design, business rules and logic for the entire application. In 1977, he developed a Microsoft Access database that has been used by TPS crime analysts. This database transforms the data extracted from a mainframe database into a relational database. In 1998, he was part of a team that created a database that is the source for both crime statistics and a repository for occurrences/contact information. At the time of his deposition, his duties involved the design and implementation of TPS crime analysis/investigative databases, and extracting and creating reports from those databases using industry standard tools. He was responsible for developing and maintaining the programs that extract weekly, bi-weekly and annual crime statistics for the TPS. He read the affidavit of Richard Faulkner.
[20] On the question of unique identifiers, Mr. Angus explained that an MTP number is a Metropolitan Toronto Police number issued by the TPS when a person is photographed and fingerprinted. On the other hand, an FPS number is issued by the R.C.M.P. and is a national fingerprint identifier used to support criminal records and unique to an individual. Police Services submit fingerprints to the R.C.M.P., who then check to see if the print is on file. If it is, the R.C.M.P notifies the submitting agency of the existing number. If the print is not on file, a new number is created. Moreover, only individuals charged with indictable offences are fingerprinted in the normal course.
[21] CIPS does not generate or validate FPS or MTP numbers. MTP numbers have been issued since the early ’70’s, but there is no standardized format for entering them within the CIPS application. They are recorded in varied fashion (e.g., 88/2310, 2310/88, 88/002310).
[22] Similarly, there is no standardized format in the CIPS application for entering FPS numbers. These, as well, are recorded in varied fashion (e.g., 12345C, 0012345C, C12345, 12345 C).
[23] Since CIPS does not mask or validate the format of the data inputted into the FPS and MTP fields, any of the above-mentioned formats will be accepted. It was Mr. Angus’ conclusion it is not possible to simply match FPS and MTP numbers in CIPS’ records.
[24] Mr. Angus deposed that, at the time of his deposition, the TPS did not have an algorithm capable of replacing a person specific unique identifier with a randomly-generated number. He stated an algorithm would have to be developed specifically for the MANIX database and for use with CIPS. Mr. Angus pointed out that Mr. Faulkner had assumed that the CIPS application is used to track a person’s arrest history. He said this assumption was incorrect and, therefore, his conclusion that CIPS is an “anomaly” because it does not uniquely identify individuals is incorrect. He said if Mr. Faulkner were to review the entire CIPS database layout of tables, he would see there is no connection between individual person records. There is no built-in mechanism to link an individual person to different arrests. This was a decision the TPS made in 1996, that CIPS would not be designed to link persons to multiple arrests.
[25] Mr. Angus further deposed that the social insurance number field (“SIN”) is not mandatory when completing the “persons investigated card”. From his personal experience, those cards rarely entered a SIN. He deposed that the SIN was recorded in so few instances, it would be useless as a unique identifier.
[26] Mr. Angus concluded by stating his information and belief that the MANIX database which captures some of the information contained in the Persons Investigated card does not capture an individual’s SIN.
The Reply Affidavit of Richard Faulkner
[27] The reply affidavit speaks to the assertion by Mr. Faulkner that “unique identifiers” do exist. He submits the Angus affidavit confirms that the CIPS database does contain unique identifier ID’s. Mr. Faulkner states that, “it is unusual that there is no standard format for entering FPS numbers”. He takes the position that the information provided in the Angus affidavit appears to confirm the statement in his affidavit of January 27, 2004 that a random unique identifier number could be created easily.
THE ASSISTANT COMMISSIONER’S DECISION (Order MO-1989)
[28] The Assistant Commissioner began his analysis by asking himself if the requested information was a “record” within the definition section of the Act. Section 2 of the Act specifically defines “a record” as follows:
‘record’ means any record of information however recorded, whether in printed form, on film, by electronic means or otherwise, and includes,
(a) correspondence, a memorandum, a book, a plan, a map, a drawing, a diagram, a pictorial or graphic work, a photograph, a film, a microfilm, a sound recording, a videotape, a machine readable record, any other documentary material, regardless of physical form or characteristics, and any copy thereof, and
(b) subject to the regulations, any record that is capable of being produced from a machine readable record under the control of an institution by means of computer hardware and software or any other information storage equipment and technical expertise normally used by the institution;
[29] Section 1 of Regulation 823 under the Act states:
A record capable of being produced from machine readable records is not included in the definition of ‘record’ for the purposes of the Act if the process of producing it would unreasonably interfere with the operations of an institution.
[30] In his analysis of this question, reported at p. 590 of the Record, the Assistant Commissioner identified three issues:
whether the basic information – that is, unique identifiers – exist in a “recorded” form in the identified database and are capable of being produced from machine readable records by means of a computer hardware and software or any other information storage equipment and technical expertise normally used by the institution.
if the answer to 1) is yes, the second issue to be decided is whether the process of producing the information would unreasonably interfere with the operations of the police as the definition of ‘record’ is limited by s. 1 of Regulation 823;
if the process of producing the record would not unreasonably interfere with the operations of the institution, then the unique identifiers constitute a ‘record’ for the purpose of the Act.
[31] Therefore, the final issue to be decided, according to the Assistant Commissioner, is whether the police were required to provide Mr. Rankin with a record which replaces the unique identifiers with randomly-generated, unique numbers.
[32] On the first question, the Assistant Commissioner considered the submissions of the parties in considerable detail. He concluded that a unique identifier existed in a form accessible through the police’s CIPS system. He accepted that the extraction of the requested information would be time-consuming, and would result in certain inaccuracies in the information. He found that unique identifiers exist in CIPS and that they are capable of being produced from a machine-readable record. He came to the same conclusion with respect to the police’s MANIX database. I note, at this point, that nowhere in the Assistant Commissioner’s discussions is there any reference to the words “normally used by the institution” found in s. 2(b) of the Act.
[33] The Assistant Commissioner then proceeded to ask himself if the process of producing the record would interfere with the operations of the institution. The following discussion in his reasons centres on s. 1 of Regulation 823 under the Act, referred to above. As will become apparent in these reasons, directly passing to a consideration of the Regulation overlooks the necessity of considering whether the “record” can be so produced “by means of computer hardware or software or any other information storage equipment and technical expertise normally used by the institution”. [Emphasis added] The Assistant Commissioner concluded that the process of extracting and manipulating data to create the record Mr. Rankin sought would not unreasonably interfere with the operations of an institution as provided in s. 1 of Regulation 823.
[34] Finally, the Assistant Commissioner asked himself if replacing the unique identifier which he found to exist with a unique number would constitute creating a record. Following five pages of analysis, he concluded that replacing the unique identifiers with randomly-generated numbers did not change the nature of the information; rather, in those circumstances, it served the purpose of “anonymizing” the information. He concluded doing so did not result in the creation of new information or the creation of a new record.
ANALYSIS
[35] For ease of reference, I reproduce s. 2(b) of the Act:
‘record’ means any record of information however recorded, whether in printed form, on film, by electronic means or otherwise, and includes:
(b) subject to the regulations, any record that is capable of being produced from a machine readable record under the control of an institution by means of computer hardware and software or any other information storage equipment and technical expertise normally used by the institution;
[36] In construing s. 2(b), I conclude that the words “normally used by the institution” qualify both “by means of computer hardware and software” and “any other information storage equipment and technical expertise”.
[37] I do so for two reasons. First, the use of the words “any other” convey the sense that the two phrases connected by “or” speak of equipment, similar in nature, capable of producing a record from a machine-readable record.
[38] Second, it would make no sense to have different requirements of the institution to “produce a record” depending on whether it could be done “by means of computer hardware and software” or whether “by any other information storage equipment and technical expertise”.
[39] Thus, an analysis of s. 2(b) requires:
a finding there is a “record” capable of being produced from a machine-readable record;
a finding that such a “record” is under the control of the institution; and,
a finding that the “record” can be produced “by means of computer hardware and software or any other information storage equipment and technical expertise normally used by the institution”. [Emphasis added]
[40] If requirement three is not satisfied, that is the end of the matter. If it is satisfied, there remains the requirement established by s. 1 of O. Reg. 823 that the “producing” must not unreasonably interfere with the operation of the institution.
[41] The Assistant Commissioner, in answering the first question, found that a unique identifier existed in a form accessible through the CIPS system and the MANIX system. It matters not, for the purposes of this review, that I might not have come to the same conclusion. What is fatal to the Assistant Commissioner’s decision is his failure to consider whether the means required to produce the record were means “normally used by the institution”.
[42] I have reviewed the Assistant Commissioner’s decision, paragraph by paragraph, and nowhere do I find where his mind turned to the question he was required to ask of himself. Having failed to do so, his decision cannot stand.
[43] Mr. Wong submits there is no requirement that the software that would be required to extract the data from the CIPS or MANIX databases must exist at the time of the request. I reject this submission. The submission ignores the plain language of s. 2(b) of the Act, which requires the “software” to be “normally used by the institution”.
[44] Moreover, had the Assistant Commission so turned his mind, the evidence from both Mr. Faulkner and Mr. Angus confirm that the “means” to produce the record were not “normally used by the institution” and indeed, did not exist.
[45] Mr. Faulkner deposed:
• ‘I believe that a professional computer programmer would be able to create an algorithm to replace unique identifiers with random numbers with little difficulty.’ (Affidavit, para. 11, p. 222 of the Record) [Emphasis added]
• ‘As with CIPS, this could be done through creating a simple algorithm.’ (referencing MANIX, Affidavit, para. 14, p. 225 of the Record) [Emphasis added]
• ‘…the Angus affidavit appears to confirm that a random unique identifier number could be created easily.’ (Reply affidavit, para. 9, p. 575 of the Record) [Emphasis added]
[46] Mr. Angus deposed:
Currently, the TPS does not have an algorithm capable of replacing a person specific unique identifier with a randomly generated number. One would have to be developed specifically for the Master Name Index Database (“MANIX”) and for use with CIPS. (Affidavit, para. 19, p. 471 of the Record)
[47] The application is granted. Order No. MO-1989, dated November 7, 2005, is quashed.
[48] If costs cannot be agreed upon, the parties may make brief written submissions within fifteen days of the date of this order.
CARNWATH J.
GREER J.
SWINTON J.
Released: 200506
COURT FILE NO.: 514/05
DATE: 20070621
ONTARIO
SUPERIOR COURT OF JUSTICE
DIVISIONAL COURT
CARNWATH, GREER & SWINTON JJ.
B E T W E E N:
TORONTO POLICE SERVICES BOARD
Applicant
- and -
INFORMATION AND PRIVACY COMMISSIONER/ONTARIO and JAMES RANKIN
Respondents
JUDGMENT
CARNWATH J.
Released: 20070621