Almost 2 years after I originally made the request, the ICO has found partly in my favour: DfID must publish 64 fields relating to the 8,000 projects in the database.
However, 6 of the fields - those which contain comments relating to project performance - have been withheld under the argument that their disclosure would be likely to inhibit civil servants' ability to freely and frankly exchange views and advice.
According to the letter (which will be published on the ICO's website), DfID must publish the information within '35 calendar days' of 23 June 2011.
What's going to be released? The Decision Notice only shares the titles of the fields but together with this user guide to the ARIES database (PDF), we can guess at what is likely to be most interesting:
- 'Total impact score' - a measure of how effectively the project achieved each of its outputs
- 'Output risk' - a measure of the risk of project failure
- 'Disbursement suspended' - information relating to the suspension of disbursement
- 'Method of scoring' - whether reviews were conducted by DfID staff, consultants, etc
It is really exciting that so much information will be released and I look forward to examining it and mashing it up with other datasets. Hopefully the data publication won't be a one-off: it would be great if the fields were regularly published on DfID's website.
What's been refused? The fields which won't be released contain comments made by DFID staff and partners in relation to the projects under review. The ICO argues (para 88) that because some of these comments "could potentially embarrass foreign governments", their disclosure could make civil servants "more circumspect in expressing their views about the performance of a project" - and that this would therefore make them less able to provide free and frank advice.
This makes sense. But here's a question: wouldn't it be possible for each of the records to be assessed to determine which records are sensitive and which are not? Not according to the Decision Notice: DfID explained to the ICO that doing so would place "an impossible burden on the department", particularly since the expertise required to do such a review is spread right across the organisation. It is obviously not right to place such a burden on DfID and I accept the Commissioner's decision on this.
Stepping back from this particular FOI request, we are still left with the problem that rich information about what works (and doesn't work) in development will still be withheld from the public domain. Perhaps the problem could be approached in a different way? Rather than attempting to review the entire dataset at one fell swoop, a team from DfID or ICAI could annually review a subset of the comments (say, for projects completed 1 year previously) and determine which information could be released and what should be redacted. At least then there would be a mechanism for releasing the raw data - and ensuring it isn't lost to the sands of time.
It is clear that a lot of people worked hard to respond to this request. My thanks go to everyone at DfID and the ICO who were involved.