It turns out that Greg Mankiw has blogged about Oregon's mileage tax proposal and, lo and behold, frequent guest blogger and friend of the blog, Fred Thompson has chimed in with some clarification. The essence of which is that congestion pricing is indeed a major motivator of the whole GPS idea and (as Fred rightly notes) without such motivation, the GPS doesn't make much sense.
But I still wonder how you deal with the problem of non-residents. Since the bulk of the congestion problems in Oregon are in Portland, particularly the I-5 and I-205 crossings of the Columbia river, it is not clear how something tied to Oregon registered vehicles will work. A lot of the congestion in these areas is apparently coming from Washington residents that work in Oregon, which means that the GPS in Oregon cars won't be effective in dissuading these drivers.
London has famously instituted congestion pricing, but there the tax works based on photographs of license plates. So if you enter London you pay, regardless of where your vehicle is registered.
Given the cost, complexity and incompleteness of this system, I still cannot see why it trumps the simple and effective gas tax. It strikes me as a wonderful idea of you are an engineer (especially a traffic engineer) because you get to play with new toys and tools, but I remain unconvinced in my mental cost-benefit analysis of the idea.