You need to determine what results they must achieve, clearly and completely unambiguously, so that they understand what they can control (in essence, how they work, the order in which they develop things, etc.) and what they cannot ( usually what is expected of them - both in terms of the actual product, and with supporting materials, such as progress reports, and when all this is intended for delivery). You also need to tell them what resources they have - can they order highly specialized machines or order new software, for example, or is this all decided?
I would also ensure that one of their early results was a schedule of completed steps against which you could measure progress and agree with them what would happen if they started to miss the milestones.
But I have little doubt that you are going to determine version control, bug tracking, and so on. Surely this is what you should let them decide? In the end, they are part of the process. Personally, I would say that they should have version control, centralized logging of defects, etc., But the mechanisms, tools and processes should correspond to them.
It looks like you are saying that you want to create results only in a production environment, but do not trust them. If you say what you are going to do is create a ROWE, then you need to make sure that otherwise you really only perform half the process, and these situations rarely deliver the benefits people hope for.
In the end, either you trust them or not, but if you cannot trust them to decide how to make version control, which is frankly secondary to developers, you probably shouldn't trust them with a schedule that is usually much less simple a question.
Jon hopkins
source share