I understand what you're saying, but based on there being no expectation that current tablet prices are about to fly through the roof, I would contest that high DPI screens would be particularly expensive to make in desktop sizes. Besides, if they were (say) 50% more expensive, that simply creates more choice in the market, since manufacturers would continue to offer budget panels alongside their high DPI offerings.
Equally with a desktop OS, high PPI tends to mean smaller text. It shouldn't have to mean that - it would be relatively easy to have a phone/tablet-style feature that scaled text and start menu (and everything else) to a default size (it's not like your computer doesn't know what size monitor you're running), and that default size could then be customised so you can have everything as big or small as you choose. This could be done on traditional screens but on high PPI screens has the significant advantage that things won't look blurry when scaled up or down.
As for gaming, we get back to the point that just because you have an uber-resolution display doesn't mean you have to use all the pixels. I actually think I undersold myself here before as I was working on the (current) principle that using non-native resolution = decreased IQ. One of the points with a high PPI display is that you can set the 'virtual' resolution to be whatever you want, yet because the pixels are too tiny to see individually, you end up with roughly the same image quality as an 'old-style' panel (e.g. 1680x1050) with a native resolution of whatever you set it to.
Obviously if you were going to set your new super-high-def panel to a mediocre resolution all the time there wouldn't be any point buying it over a cheaper one with a mediocre resolution as native, but (hopefully) you see my point.
So cost-effective gaming shouldn't be a reason to dismiss high PPI panels - they can simulate low-PPI panels without loss of IQ as compared to a low PPI running at native res.