But they would be making the sensor bigger and vastly increasing the pixel count to maintain a similar pixel density so the advantage of full frame may be lost. I don't know, just speculating.
It does beg the question though, if the D800 sensor is a full frame 36mp derivative of the D7000 and performs better than the D700 in terms of iso, why is the top of the range D4, hobbled with a measly 16mp?
The advantage of full frame is in the area of the
whole sensor. Don't believe me? The D3x has comparable pixels than the D90 in almost every way, and they're only just barely bigger. Yet the D3x smacks the D90 silly IQ wise, given that it's near indistinguishable from a D3 up to ISO1600 (it's arguably better from 100-800) and keeps up very well up to 6400 (the highest it goes). As long as the component structures making up the pixel can be fabricated at the right size, pixel size on its own means nearly nothing.
A bigger sensor captures more light. So for a smaller sensor to be equivalent to bigger, it needs to be more efficient. Putting it in FX/DX terms, for a DX sensor to equal an FX one, it needs to have more than twice the QE while maintaining the same DR, or the same QE and well capacity while more than halving read noise at every ISO. And if you do that to a DX sensor, well, there's nothing but price stopping you from doing the same things to an FX sensor.
It doesn't really beg the question. A DSLR is so much more than just the sensor. Camera designers have a lot of constraints to consider.
As for the D4 being hobbled? Mixture of practical and marketing constraints. On the marketing side, most photographers don't understand advanced maths, optics, signal processing or electronics. It's not especially relevant to a photographer beyond "does this camera do what I need it to?" which means that sometimes misinformation propagates.
Kind of like the misunderstanding of ABS and CDS putting a bit of a crimp in economic performance. You don't sell to your customers by telling them they're blatantly wrong, or by telling them that they have to put twice as much work in to get the same IQ (at high ISO, you don't get anything like the nominal resolution)
On the practical side, to make smaller pixels properly you need more advanced equipment. Silicon lithography equipment is hilariously expensive, so most camera stuff is done on castoffs from microprocessor companies. For reference, the newest CPU fabs use equipment that can make a transistor 22nm across. DSLR sensor fabs tend to use equipment that makes transistors 130-180nm across. Making smaller pixels can be expensive.
In addition to that, more pixels means more data. If you have a 36MP sensor running at 12 fps as the D4 does, you're generating...700+MB a second. To get that off the sensor in any appreciable amount of time means driving the circuitry at a very high frequency (the distance the data has to travel in a given amount of time and the amount of data tell you the needed frequency). To do that generates heat and crosstalk and other things that impact the integrity of the data, so it's not really an option. Not only that, if you're running at the moderate or high ISO range,
you're not getting the nominal resolution anyway. So you'd have made all these changes, charged money for them, and they wouldn't actually provide a benefit to your target market (high performance, high speed, low light shooting).
Those in the market for a D4 would rather have better FPS and improved rolling shutter than higher nominal resolution, and really, 16MP isn't exactly poor resolution. And even if the data could be read sufficiently quickly off the sensor, buffer performance wouldn't be the greatest, given the fastest cards we have at the moment write at 125MB/s and to go faster means attaching pricey SSDs to an already space constrained camera.
Marketing departments do perform a useful function as much as many engineers and customers think otherwise.
