As my work on the latest review comes to an end I find myself having been exposed to some interesting projects around government at varying stages of delivering agile solutions. There are some key findings worth musing over from the experiences of the various leads.
1) Development can be agile but be aware the organisation it is being carried out in is not. Be aware that government is not agile and nor are its ingrained processes and scrutiny and approvals. This will improve but not in a short enough time.
2) Costs will not be significantly lower due to agile. It will improve outcomes but don't expect huge discounts. Rework and acceptable failure will inevitably swallow cost as will 1) above and delays.
3) Agile purism is naive in government. If the purists take over then projects will fail. Keep it real.
4) You cannot use agile as an excuse for not doing normal things like design. Preparation is even more important in agile than in other methods. Drop good prep at your peril. Agile means flexibility on what you deliver for the business. NOT how you deliver it. By all means thin out on upfront requirement capture. But that is not an excuse for dropping best practice in technology terms.
5) Do not hire developers until you are ready. You will waste money that way.
The future of agile in government is a hybrid. It will not work in a purist way as it relies a lot on trust rather than business cases and approvals and the two concepts do not sit well together. Tax payer's money must be accounted for period. This imposes structures and processes that do not align well with the agile approach. A happy medium will be found.
