Clearly getting more pixels on target at the same light sensitivity levels is better. What confuses people is likly the squared / square root difference in pixel to pixel density and the fact that they do not read the light level specifications (assuming they are all done to a standard measurement technique). to double the pixel density of a 2MP camera, you would need an 8 MP camera. If you can buy an 8MP camera with light sensitivity of 0.006 lux or better (lower) for less than the money in your pocket, then you would likely have a better camera for 99% of your viewing pleasure. The 4 MP Hikvision claims to be 0.01lux, which it is not. it is pretty good but not that good. Even with this spec it needs 2x the same about of light (think shutter speed if you will), and there is the DSP clearing things up.
One simple way to look at it is for a reasonably well lit but dim area (say 1 lux), the total light that enters the lens in some fraction of time (say 1/25 sec), either gets divided up to 2M pixels or to 4M pixels. Which one do you think will be clearly presented?
Now if we have a very bright scene, like over 100 lm it is likely that the more pixels on target the better. There are indeed some hours of the year that fall into this category. Getting WDR and shutter speed correct is still needed, but the quality higher MP camera will likely prevail. Here in sunny southern cal i would guess that this accounts for 20% or so of the time in the year and more like 2% of the time that i care about. If i wanted to watch cars go by during the day, i would probably opt for a 4K camera over a 2MP starlight. Good to have more than one camera on target and have the money to get there
.