Question about public service projects, may make it an Ask HN or a blog question as well:
The Agile or iterative model, coupled with the sec/devops CI/CD model that includes "own what you build," means that developers and highly skilled devops teams are available to manage a CI/CD pipeline throughout the lifetime of the product.
There is an underlying assumption that you are building to support the growth and revenue model that SaaS companies require to survive and in turn pay for continuous developers.
In public services, there is no such revenue growth. You have a budget for developing it, then hand it over for production, and then you manage it over the long term with more cost effective operators who are largely unionized employees, engaging developers only if needed to reconcile the code with an infrastructure change.
Keeping developers engaged and maintaining CI/CD on a product that does not have revenue or growth means that the cost curve diminishes much more slowly than in the waterfall engineering model. It means we have to ask whether the additional cost yields commensurate benefits, and what budgeting for a service supported by CI/CD truly costs without hockey stick revenue growth attached to it.
I contract to an agency that would benefit from an 18F.gov or gov.uk like digital service, but I have not seen this particular cost issue addressed. Has anyone in gov.uk, 18F or another digital service run up against this, and found solutions?
From what I have seen, this is one of the major problems that the GOV.UK / GDS service model has thus far failed to address.
The advice is "keep the team around for each service indefinitely". Unfortunately this doesn't mesh well with either government technology funding models or the demand for increasing numbers of services.
The Agile or iterative model, coupled with the sec/devops CI/CD model that includes "own what you build," means that developers and highly skilled devops teams are available to manage a CI/CD pipeline throughout the lifetime of the product.
There is an underlying assumption that you are building to support the growth and revenue model that SaaS companies require to survive and in turn pay for continuous developers.
In public services, there is no such revenue growth. You have a budget for developing it, then hand it over for production, and then you manage it over the long term with more cost effective operators who are largely unionized employees, engaging developers only if needed to reconcile the code with an infrastructure change.
Keeping developers engaged and maintaining CI/CD on a product that does not have revenue or growth means that the cost curve diminishes much more slowly than in the waterfall engineering model. It means we have to ask whether the additional cost yields commensurate benefits, and what budgeting for a service supported by CI/CD truly costs without hockey stick revenue growth attached to it.
I contract to an agency that would benefit from an 18F.gov or gov.uk like digital service, but I have not seen this particular cost issue addressed. Has anyone in gov.uk, 18F or another digital service run up against this, and found solutions?