My experienceMy comment on this (from experience) is that it is more dangerous for a government organisation to venture into the world of free, open source products than it is to stick to the tried and tested model of writing very complicated specification documents, three-way tender processes, months of shortlisting, more scrutinisation, fine tuning, financial checks and finally a procurement contract. My rather cynical view of this process is that we've over-engineered the process to the point where the more a project costs to deliver, the more interest it generates internally and along with that, a label of importance. A lone internal developer using open source software could easily create a killer app which would be largely ignored internally because no actual money appeared to have been spent on it. My argument is that this should be reversed; if someone creates a killer app for nothing, it is a double win and should be regarded as such.
Hack DayFirst comes a set of requirements for what WE want to do, not what the company is trying to sell us. My experience with Microsoft Word is that 95% of users tend to only use 5% of the product's potential (glorified typewriter). This suggests that we'd be better off in the main with a free equivalent for 95% of users and only pay the expensive software licenses for the 5% or power users who actually needs those advanced facilities.
So we determine what we want. Then we share it with others to see which features are common to all. At this stage we should not expect 100% of what we feel we need. We should be aiming to be happy if the proposed solution does at least 70% - 80% of what we want. Once this is off the ground, the rest will follow very quickly.
Next, we bring in the techies/geeks/coders/hackers who are usually less interested in ideas and more interested in making stuff. We let them soak for a while and then the fun starts. The ideas people and the techies start working together to develop the protypes. We need people to comment on the functionality, others to keep usability in mind, others who can design lovely buttons, graphics and banners and the coders to pull it all together.
I'd say this could easily be a 2 or 3 day event. It could be a Thurs/Fri worksday followed by an optional Saturday or it could start at the weekend and continue into a Mon/Tues. If it is kept to a single day event, I'd say it would need to be preceded by a few weeks of online conversation with proposals and votes to determine up-front what people want to 'hack'. This could be done in two stages with a public policecamp/copcamp to tease out the ideas and crystalise the requirements followed by a period of online discussion and culminating in the hack day where something is actually built.