Fixing FOI

In this world of web 2.0 the delivery of the Freedom of Information act is still very much government 1.0, i.e. one step above stone tablets. Whilst the information that comes from requests gets much attention, whether it is the Independent pursuing its latest lead on whatever happened to Humphrey or the Telegraph lamenting bonus payments to civil servants it is also sometimes the lack of information that gets focus, as with the proposals to exempt members of parliament from the act (to avoid driving “a coach and horses through the relationship that we have with constituents” says Jack Straw) or through the claim of a necessary exemption. Some departments appear to have more exceptions than successes, as with the Foreign Office which has claimed exemptions for 70 per cent of all requests it has received. In total, of the 62,852 requests made to central government since 1 January 2005, 26,083 have not been granted. Those figures, from the Independent, don’t quite seem to add up (26k is not 70% of 62k) but the general point stands.

Since the act was put in place, attempts to use it have fallen:

Figures released by the Department for Constitutional Affairs reveal requests to central government fell to a low of 7,641 between July and September, compared with 13,603 in the first three months after the law came into force. Freedom of information campaigners warn that this might be evidence that the public have become frustrated with their failure to get answers. Overall, the success rate for requests across all departments has fallen by 2 per cent to 60 per cent in the past six months.

The very thought of exposing information to the wider population makes some public servants quiver with fear – after all, developing legislation is a process, like making sausages, that even the most ardent fans of the output probably don’t want to be exposed to.

Perhaps the barriers to entry – i.e. how you ask for information to be disclosed, what you can ask for and what the limits on cost of discovery are – are deliberately set high so as to make it difficult to get some information so that only the most determined or those with the most time on their hands see it through. That may mean that some relatively simple requests get caught in the net because it’s too hard to get the information or perhaps even because the question, as phrased, doesn’t get quite to the nub of what the person asking is really looking for – after all, a 2 word google search turns up millions or hundreds of millions of answers and adding words, whilst reducing the number of items returned, doesn’t necessarily get you nearer what you actually want especially where the words you choose don’t match the exact syntax in the original document.

These are all signs that the process is broken – usage is falling, exemptions are high, frustration with those asking is higher still – and therefore, if the desire was there – the desire to make it a true “freedom of information” act – it could be fixed.

With the government 1.0 process that we have, there may be lots of ways to fix it. After all, a process that goes “compose request”, “send”, “wait for time to expire”, “wait for extended time to expire”, “get answer but maybe not the one you were looking for” hardly engenders customer loyalty. If I were technical I’d name this an asynchronous and error-filled process.

If the desire really is to create an open interface to government, a way where anyone and everyone can see what is happening inside the bowels of the machine then a properly thought through migration to a new model is needed.

First some assumptions on scope

1. Only final versions of documents, be they legislation, policy, white papers, green papers etc should be made available to the public. I wouldn’t like people to see the drafts of these blog postings as I edit and shift words around and nor, I believe, would those working inside government. Proposals are often put up, shot down, revised and put up again whilst issues on capability, capacity and do-ability are reviewed. There will be many who want to see every last version so that they can try and see who rejected and I understand that, but it doesn’t help create a model to work on – it will increase the resistance to opening things up and result in additional controls and checkpoints. It might even encourage more document generation “off the air” so that only final versions are saved on the network.

2. Minutes of meetings, notes of actions, risk and issues registers for programmes, gate reviews, independent reviews and so on should all be available, again in their final form. Generally there is a tendency in the civil service to record the flow of conversation in many meetings and, unlike in corporations, not just the actions. Allowing access to these minutes will, I’m sure, over time result in briefer and briefer minutes and a greater focus on just the actions.

3. Financial information is fair game. Costs per line item – salaries, software, hardware – etc are all allowed; sub-line items are also allowed, e.g. cost for HP hardware or Oracle software. Likwise travel costs – aggregate hotel bills, transport bills etc are all allowed. What isn’t fair game is costs that would allow identification of any individual at any level of the civil service organisation. How much any one person is paid in the government is, in truth, neither here nor there as it can’t skew the figures more than a tiny amount. I think this applies throughout the civil service but there is perhaps a different rule for our own MPs, where we seem to have a particular interest in the level of expenses they claim.

Now, if those scope assumptions are correct, all we need is a mechanism to get the information that isn’t time consuming, isn’t costly and, as far as possible, is automated.

(1) and (2) are, I think, far easier than (3). Encouraging the use of keyword flagging, whether in MS-Word or one of the open equivalents, ought to be easy; mandating it would be harder but is not a huge step. Keywords required would relate to the policy area, the stakeholders, the minister involved and so on. One last flag would say “final” or “public” or “FOI” or something similar so that it was known that it could be made available – any document that didn’t have that last flag would be exempt. Every document would be saved to a network area and, once a day or once a week or whatever, those documents would be trawled, indexed and appropriately stored – in a parallel file store to the one that the author put in place. The new store would have a web facing front end with a search engine – let’s call it govgle – that could speedily look through all the documents, picking up both the flags and the content. I’d honestly expect far more of the documents that made if this far to be pdf files simply because it would reduce the chance of seeing historic edits laid bare.

(3) is far more problematic. Whilst I’m pretty sure that every major and the vast majority of minor departments use Word, Excel, Powerpoint or their nearest equivalents, I’m absolutely sure that standards in the financial area are a long way behind that. It’s not simply a case of differences in the systems but differences in accounting lines, cost codes, classification of funds, dates for booking, month end closure periods, accrual processes and so on. Disentangling those is a huge job – and, very possibly, an impossible one. Perhaps the easiest way forward in the short term would be to create a standard “summary account” that ran monthly that posted cost and revenue lines into a short, say 100 or 200 line, report. The job then would only be for each department to group their costs and revenues (mostly costs I’m sure) into those lines and create the report. This report would then be made available in the same way as the earlier items.

Is this better than what we have now? Certainly – what we have now is either whatever is published on the web across 100s or 1000s of websites or whatever is turned up via individual requests to individual departments. Neither is e-government 1.0 let alone anything more advanced than that. This process does a few things:

– It opens up the process of FOI allowing for more trial and error. Different questions can be asked quickly and easily with close to instant results. This will allow “skimming” for potentially interesting topics and may even expose a few of the inevitable absurdities that come with multiple large organisations acting out of alignment, but that shouldn’t be a bad thing, the inevitable Daily Mail headlines notwithstanding

– It shifts the burden of consolidating the information from the government to the inquisitor – i.e. the search engine will return lots of results and pulling those together is the job of the person asking, just as it is now whenever we use a search engine. If the engine is good – and Google or Autonomy or similar would be a sensible choice – it will quickly see what the most important items are based on document criteria as well as usage (link volume will, I think, not matter so much here)

– It allows whoever is asking the question to aggregate responses from multiple sources within government quickly; that aggregation may well expose a lack of joined up thinking in government, but that would hardly be news

– It will allow the creation of some commercial models where information providers or indexers, lawyers, accountants, lobbyists etc, will be incented to pull together packages of information that might be useful to their clients

– It will make follow up questions more incisive. The day to day information that is available will sate the appetite of most but those pursuing more specific or comprehensive information can go through the offline process. This process can be set up knowing that the bulk of questions are going to be difficult – and appropriate charges can be put in place for it; after all, there’s no reason why it can’t be a profit centre. You want to ask a difficult and time-consuming question, well, take a number and pay the fee.

There are only then two questions to ask:

1) Does anyone actually want to make it easier to get such information?

and

2) Is FOI important enough for there to be a business case to do this?