I was once told that the time it takes for the first 10% of the population to embrace a technology is roughly equivalent to the time it takes for the next 80% to embrace it (90% total). The final 10% embrace at a much slower pace. Therefore, instead of arbitrarily setting a speed that an arbitrary number of people should have available, the analysis should look at what the top 10% of speeds are and try to get those to 90% of the people in the same time frame that it took those top 10% to migrate from the national average to their current speed. Does any of that make sense?
Yes it does. But we already know what is needed to get symmetrical 100Mbps connections to the other 90%. Fiber optic (Fiber To The Home, FTTH) networks. Since that same fiber may be capable of symmetrical 1Gbps service, the installation of FTTH covers future growth needs in the area of data transfer rates. Now if those deploying DOCSIS 3.0 do the proper channel bonding they might be able to get symmetrical 100Mbps according to the theory of DOCSIS 3.0, so I could see that being used also.