Ultimately, the issue is whether 18F is more effective than other means and accurate accounting is important to that end.
Let's say that the VA gets bids to create a new system for something. 18F says they can do it for $900k and ContractorX says they can do it for $1M. 18F gets the contract and produces a system that cost the VA $900k. Let's say the system cost taxpayers an additional $1M that isn't billed to the VA. That means that taxpayers are out $1.9M rather than $1M going with ContractorX. The issue can be that 18F isn't accurately pricing their services and hiding the actual cost to the taxpayer.
That isn't to say that I like contractors. I don't. But the point of 18F should be to prove that the government can create its own digital services at a lower cost than contractors. If they're "cheaper" in that they're charging agencies less, but then just taking money from a slush fund to cover the fact that they're not efficient, that's not a good thing.
Also, price being "value created - effort to get paid" isn't a good metric. You need to look at alternatives. If a system saves $1M, but you could pay one source $100k for it and another source $900k, you want to go with the $100k source. If we had to pay the value created for things, we'd stop a lot of economic activity. In fact, why would the VA want to do anything at all given that they're going to be charged the full value of any savings? 18F shouldn't be credited with $1M in savings if they spent $900k to achieve it when a contractor could have done it for $100k.
The value of 18F is if they can produce digital services better than other methods. Other methods are generally paying some non-governmental company for the services or hiring direct employees for the agency. In order to determine if 18F is a good model, they need to charge agencies enough to cover their costs. Otherwise, it's ridiculous. Someone in Health and Human Services would say, "we could hire people to do this project directly, but then we'd be paying 100% of their cost, but with 18F they'll charge us a fraction of the employees cost. In fact, we'll tell them to hire the dozen people we were going to hire and we'll only pay X% of the salary rather than 100%".
Plus, if agencies don't have to pay the full cost, they'll do projects that don't make sense. Let's say that a project for the VA creates $1M in savings. 18F says they'll do it for $900k. The VA green-lights it to save $100k. Great, right? $100k for spending on other things. But 18F actually spends $2M on the project. The VA gets another $100k in their budget, but the government is actually down $1.1M. When people see situations like this, they want to take advantage of it since they're able to basically spend other people's budgets. "If you use 18F, a bunch of the money for your project doesn't come from your budget," is the kind of thing higher-ups dream of.
I think 18F will get better at this. Figuring out how much one needs to charge is hard since it's hard to know exactly how much effort a project will take. I'm guessing that 18F is looking to generate good-will with agencies rather than going back to them with cost over-runs. Still, 18F needs to prove that it's a more effective way than alternatives and it can't do that by just under-charging and subsidizing whatever projects they get given. If 18F just offered free services, then every agency would want infinite 18F work (since it isn't coming out of their budget). Then which projects should be worked on? The ones 18F finds interesting? 18F wasn't created to be the arbiter of what the government should do. As such, they need to price accurately based on their costs. They will get better at this. They will likely have to start passing along cost over-runs to agencies. But if they get great, motivated people, they could provide digital services at a lower cost - it just has to actually be a lower cost.
Even if you hate contractors, a lot of agencies have tens of thousands of employees. Health and Human Services has ~80,000 employees. The VA has ~313,000 employees. Why shouldn't they just hire an engineering team? Why should they outsource a project to 18F? Agencies that large certainly have enough engineering projects to keep an internal team busy. They're not some restaurant that outsources their website because they don't have enough work to keep a dev busy. Well, maybe 18F can gain efficiencies through hiring better people at higher pay who can create better code and re-use stuff across agencies. Or maybe the overhead of outsourcing outweighs those efficiencies. That's something important to figure out and it's an answer that we can get closer to if 18F accurately charges agencies.
