I'm not sure your objection to the overbuild argument is coherent. If we take it as given that building an NBN is expensive, then it surely means we do it right the first time or we'll be stuck with a network thats going to become a chain around our neck when things like 4K TV become a reality, or more fanciful but still possible things like VR telepresence and the like. And of course cloud computing, etc.
25meg is awful for the task of cloud computing. We're a big-ass country and particularly in the mining sector we're often quite distributed in the way we do business. In some respects we've been doing cloud before it was even a word, because we had to. But it also has never been an entirely satisfactory technology because it has required ridiculously expensive fibre roll outs to businesses that have kept it out of the reach for smaller consultancies and the like, especially for the mining sector where large datasets need to be transfered around by small , often one man spare-room companies doing metalurgy, and the like.It just doesn't work well.
So we either do it right the first time, or face the very real possibilty we'll need to start pulling it out before we've even finished putting it in. The turn around here is not in decades , but in years as far as capacity is concerned. ADSL2 is only maybe 7-8 years old (for widespread use) but its already painfully slow for most people and in desparate need of upgrading. So adding 30-40% extra speed seems like its not going to last that long. Assuming the curve of things, its probably fair to say 25mbit is going to be redundant in 5-6 years. We won't even have finished this damn network and we'll be needing to pull it up and.... go fibre.
So why not just cut the nonsense and go fibre, like, right now.