The problem in this case is the identification of the role of Computer Service and its relationship between it and other departments. The three solutions all have advantages and disadvantages.
I think if we still keep the current system, the role of CS department remains unclear. If the staff are not clear about their role, they would behave nor too conservative nor too risky. So whey still lack can-do attitude on occasions, because they have no criteria in their mind to judge their action.
However, if we make CS a profit center, and free the researcher from their ties to CS, new problems arise. As CS could set any prices it liked and change them as often as it liked, and researcher also are allowed to go outside and purchase the software and hardware they want, hopefully it can work like total competition market. But the CS firm has no difference between other hardware store provider outside. That is to say there is no need to build such an independent department inside the company. It is not performing its function and create value for the company. Besides it is a waste of time which actually decrease the efficiency of the whole company.
For the third solution, running CS as an expense center says the total cost would be allocated to programs or groups in proportion to hours billed. I think this the cause for the potential of encouraging researcher to ignore the costs. Instead I think it is better for every individual of group assume the risk for purchasing new hardware and software.
So in my opinion, I think every solution have its advantages and disadvantages. And it actually also depends on the environment and style of strategy. If the company is in a world developing fast, and the company is trying to expanding itself, it is better to make CS expense department, it can encourage the researcher to apply the newest technology and get advantage of it. However, if the economy is not good and...