From a dev perspective... two specs is not too bad. Tweak a few quality parameters here and there that you have anyway, enable a couple of the "special" shaders you didn't think would ever ship, and you're good to go.
Most currently in development (and shipped but still patching) games will be doing this at the moment, but we'll have to see how the market / critics respond going forward.
As soon as one game ships with base output of 1080p@60 and Neo output of 2160p@60, then that's going to become the benchmark that Neo versions are measured against.
"They went to full 4k rendering (even though that's 4x as many pixels and the Neo only has 2x the horsepower) without sacrificing framerate, so why doesn't your game do it to?".
Neo titles that are doing 900p->"1080p plus better shadows and post fx" will probably be judged harshly in comparison.
Optimizing properly takes a lot of time, so if you do it quickly by tweaking a few knows, your non-lead SKU (currently Neo, but I guess later, base ps4) will fall short of its potential. There's a lot of performance cliffs in modern HW that you want to push up as close to as possible while not stepping over. Things are often non intuitive too - e.g. I've got a "generate mip maps" compute shader that calculates 3 mip levels at a time to save bandwidth. It was heavily ALU bound, so I tried changing it to do just two levels at a time... The result was that it became 3x *slower* because the new LDS access pattern contained a bank conflict. This kind of stuff needs profile guided hand tuning on each console SKU :(
There's also some interesting TCR/TRC rules regarding performance, resolutions, framerates, etc, which would probably force a certain level of care onto these "half new console" ports.
The problem is that much past two or three hardware specs starts to become a headache real quick, so I'm wondering if Sony and MS are planning to make a habit of this.
God I hope not. I actually hoped the Neo/Scorpio would be a bit stronger as to actually match gaming PC's for a while.