In my previous posting I promised to go into more detail on elements I think are useful when writing a business case for opening up PSI.

Writing such a case is useful for two things. First to help make decisions about the allocation of resources in your government institution, and second to clarify for yourself what should be the things that get most of your attention.

This because a generic business case for open PSI cannot be made in a way that is relevant for your government institution. The only somewhat generic case to be made for open PSI is the current cost of 'adding openness' at the end of your process, because of the FOIA requests you are receiving, in comparison to the cost of designing your processes in a way that creates open PSI by default. In all other instances it needs to be more tangible than that to be able to both convince colleagues, and help you plan the actions to get it done.

Always start any business case with the statement that opening up PSI is the law. It's not a question of whether you should do it, you already MUST do it. So a business case is needed not to provide the rationale for doing it, but to provide the rationale for how to do it, why it is useful to your institution beyond the law, and when to do it. A business case should address why your institution WANTS to open up PSI and do so according to the open government data principles, and what is needed so you CAN do it. To make clear what you mean with open PSI and open government data, make sure to include the open government data principles.

If you look at the list of underlying assumptions hidden in generic 'business case' requests, you can now juxtapose each of them with a more constructive approach. I would make these juxtapositions explicit in your business case. I hope the following paragraphs help you in building your case this way.

1 Connect your open PSI business case to your institution's policy goals (in place of generic case)

What policy goals do you have that match different possible areas where open PSI is an opportunity? What can open PSI mean to delivering on those policy goals?

What tasks are you already performing for stakeholders connected to those areas of opportunity where open PSI might help to improve delivery of those tasks? How are you currently addressing your stakeholders existing information needs? E.g. the Dutch Ministry of Education finds that by providing standard information products as open PSI, the demand for specific information products declines, while the remaining specific questions are easier to answer with the standard information products as starting point and building blocks. Thinking in terms of your organization's goals also helps you focus on the PSI you actually need to open. This focus on both goals and the PSI needed for it makes it possible to create a tangible case.

2 Describe the results you're after in terms of those policy goals (in place of trying to know and calculate all returns)

If your policy goal is about increasing participation, what is the impact of your open PSI plan on participation? And which PSI needs to be open to create that impact? Or if it is about internal efficiency, where will it help save costs (like reducing the amount of work spent on responding to FOIA requests, by making PSI public in reusable format from the start)? Describing your results in terms of your policy goals, avoids the pitfall of trying to describe your results in terms of money only. Also try to word results as the minimum result likely to be realized, in place of trying to quantify the results absolutely which is largely impossible in most situations.

3 Make visible that there will be other results as well (in place of trying to know all returns)

Any opening up of PSI will create effects in multiple places. What those results will be may or may not be easily visible at the start. But it is very well possible to use illustrations of effects of opening up PSI elsewhere to get an indication of what is likely to happen. Including the non-obvious or surprising things. Find examples of effects elsewhere, and compare them to your own context to see how likely it is you will see similar results. The connection to your organisation's policy goals in the point above should be leading though. This is more of a bonus which can help tip the scales. Some good starting points to find examples are this EPSI platform, the US Sunlight Foundation, Open Knowledge Foundation discussion list on open government data, Our Data, and the data.gov.uk or Spanish government data catalogues. It may also be useful to ask your relevant stakeholders what they think they could and would do with the specific open PSI you're aiming at.

4 Make your plan a journey, not an event (in place of a one-off decision, and in place of trying to 'prove' causality between investment and result up front)

As opening up your organization's PSI in reusable ways is a question of when and how, not if (as you must do it anyway), it makes sense to start where that makes most sense, for instance with easily available, non-controversial data sets. Which of those 'easy' datasets do you have whose release you can connect to your organization's goals? Also if those first steps are relatively light-weight, there may not be a big need to justify the effort in the first place. You can perhaps 'just do it'. This makes for great experiments that are not disastrous if anything goes wrong.

Use those first steps to gain experience in dealing with organizational and technological issues involved (things like learning to work with the relevant open standards inside your organization, or how to deal with feedback received on published material). Use those first steps to also measure results both in terms of actual results (downloads, forms of reuse, feedback you got), as well as in terms of what did not happen (no 'mis-use' of data, no negative feedback in the press etc.), and use those measurements to justify the when and how of your next steps.

5 Count costs only if they can be realistically allocated to you (in place of trying to separate all costs involved)

There are a lot of costs that you could count towards your open PSI project, but can just as easily be argued to not be part of your project. The link to your organizational goals is the linking pin here. If your goal is higher participation for instance there will already be people allocated to activities in this field. An open PSI project can be part of their portfolio, without having to count all employee costs towards your project. Also there will be costs that become apparent because of an open PSI project, but are not realistically part of the costs of the project. Think for instance of switching to using open standards for all the PSI in your organization. That is required by law in the Netherlands since 2008, regardless of the fact whether you are going to release that PSI publicly or not. If your organization hasn't switched to open standards yet, and needs to do so to release open PSI, those costs really should not be allocated to your project. Also it may be very well possible to add certain steps of your project to existing efforts, without causing those existing efforts to run up more costs. Where there is less costs, making your case is easier as well as less necessary to convince others (but still very useful to gain insight yourself).

6 Compare costs against previous and other projects (in place of having the investment completely create its own return, outside of any other context)

When you have connected your open PSI project to your organization's policy goals, you can make a comparison of how your projected results and costs, compare to those of other programs and projects aimed at the same organizational goals. It may create a completely different perspective on the relation between investment and result, by showing your case in the context of other work of your organization.

7 Use experience from elsewhere to show which risks are and aren't realistic (in place of trying to list all risks), involve your stakeholders in this (in place of preferring to pass up opportunity)

A number of fears surface every time around open PSI, and with the experience gained by already open PSI there are already good examples out there to help you visualize realistic and unrealistic fears. List those fears and show with examples how they play out, and how they can be dealt with. Such as the notion 'we can't publish this data, because there are errors in it'. In practice (the UK e.g.) we see that the quality of datasets that are published actually increases, both due to feedback as well as because of more attention being given to generating the data in the first place. Such as the notion that 'non-experts will interpret the data wrong', where in practice we see that all kinds of wrong interpretations are out there already, and having the data widely available means more people can point that out, as well as verify for themselves what's right and not. Reduce fears about e.g. privacy violations by choosing your datasets, as well as the aggregation levels involved wisely.

It is I think also important to involve all relevant stakeholders in conversation around this. Some perceived risks are based on assumptions being made about certain stakeholders, without ever having been verified by asking those stakeholders.

When you can't really meet the list of points mentioned, it still doesn't make sense to say 'no', as opening up PSI is still required by law. It means you probably need to look to different organizational goals and different parts of the PSI your organization holds to get started and to make your case. That will help you start in a place that actually will yield results as well as momentum to do more.

Opening up PSI is not merely a technology question, despite it being the law and even if everybody would be prepared to act accordingly already.

There's a range of questions involved, and each of them can be a reason for people to not act. They may be less relevant, or less problematic if you choose your goals and PSI to start with wisely. But keep in mind that all of the following issues will surface at some point in the process, around technology, organizational issues, markets and legal aspects:

Technology: level of knowledge about open standards, what 'reusable' and 'machine-readable' formats actually are and how they are connected to open standards, meta data, consistent ways of publishing PSI.

Organization: where to design publishing of open PSI in your processes, how are you going to handle feedback from outside, how will you monitor reuse so you can learn from those examples, how to stay in dialog with all relevant stakeholders, how to measure the results and impact of your open PSI for your own organization

Markets: ensuring equal access and a level playing field for everyone, licensing (specifying how you deal with government copyright), fees for providing information (soon to be restricted to incremental costs of distribution in the Netherlands) and its impact on a level playing field, defining where you just publish PSI 'as is' and where you want to enrich the info based on one of your own tasks as government institution, monitoring new economic activity generated by your open PSI.

Legal: knowing who is making the decisions about licenses/copyright and withholding PSI from publication based on national interests (security, economy), and communicating those decisions, as well as making sure the reasoning behind those decisions is verifiable.

I hope some of my notions described here help you move forward open PSI and open government data in your government organization!