Early development of a next gen probably mostly consists of pinging potential hardware component suppliers for data on their future expected technologies. Lots of data helps when eventually the time for decision making comes. Actually specifying a console down to tiny details is likely fruitless exercise when done too early.
In the case of most consoles, there's a lot of pressure to maintain long life cycles. Consoles are usually sold at a loss initially, so time is needed to recoup investment and then bring in profit. Consoles are usually expensive now, so consumers want to have a long life to get good value out of a large initial investment on their part. Long life cycles also lead to larger install bases, which helps rope in developers for more game production.
However, if a console can be sold for an initial profit rather than loss, sold cheaply, and be cross-compatible with previous generations, then much of that pressure to extend life cycles evaporates. There's no hardware investment to recoup (other than development of the console). Consumers are less concerned about investing in a new console when it's cheap to start with. Developers can largely consider install base to reach across hardware generations.
At the moment, Nintendo is meeting all that criteria...sell for profit, sell relatively cheaply, and maintain compatibility. It's probably in their favor to fairly rapidly produce new hardware if they can keep it going. Lets say they produced 2 console generations for every 1 that Sony produced. The generation that coincided with Sony's would obviously be inferior in performance and features generally speaking, but would be cheaper and easier to swallow (quickly adopt.) The in between generation could easily catch up with Sony's current, and still be cheap. All the while, money keeps rolling in.
In PC terms, it's a lot like buying a mid-grade video card every couple years and getting good value instead of buying a top-end budget buster and trying to make it last a long time.