That's the point, if you don't do the conversion to low voltage at the source you lose a lot in losses from the resistance in the lines.
Let's say you're using the normal 14 gauge solid core wire commonly found in houses right now for a 5V supply. A 200ft run of that in the house (both ways) isn't out of the question if there's a central supply in a 2 story building. That's about 1 ohm [1], which doesn't sound like much at first until we look at what it ends up meaning.
A phone might take an 1 Amp to charge at it's fast rate (tablets 2 or more), so we've got 1 Amp, with a 1 Ohm resistance. That means we're going to have a 1 Volt drop to that outlet, which means the phone is likely not going to charge at the correct current since it'll only see 4 Volts at the outlet. It also means you've just had a 20% loss in the power line in order to have a larger central power supply.
Combine that with the fact that most Switch-Mode Power Supplies aren't actually that efficient when not under load (part of the problem) then it might not be unreasonable to have it running at 80% efficiency to begin with. At this point you're down to 64% total efficiency instead of the 80-90% that the small wall warts end up getting you.
You could use a higher voltage, 48V isn't uncommon because of the telephone industry, but then you need conversion at the outlet again and end up with the same problem you've got with wall warts.
The other option is to use more copper to carry the current, say 10 gauge, which brings the losses to 0.4V and about 10% losses, but you need 150% more copper than 14 gauge.
Trying to regulate 5V across 14Ga wiring isn't anything anyone would do. If 48V became some sort of DC standard then devices would work off 48V. That might not make sense of course and how much sense it might make could depend on line voltage (120V vs 240V).
There is a cultural thing here too. If you live in a place where all heating is done with combustion then there is no particular reason so not do everything with low voltage DC. Now that lighting is LED based there wouldn't be anything that would need the higher voltage and power.
No it's absolutely not what someone would do, but it is what the grandparent proposed, that's why I addressed it. I'd love to see it used for more efficient lighting, as if it was done correctly I think you could get some pretty nicely efficient and long lasting lights that way with a bigger better current driver for the leds. This might be a good idea to move Track lighting to since they'll have more freedom on the sockets and interfaces.
As far as other devices needing the AC power, you've got hair dryers, kettles, vacuum cleaners, air conditioners, water pumps, etc that all need more power than would be reasonable with low voltage DC (some might need 25-50 amps to run at the same power they currently use). All of that just means you'll still have the AC running through the house for a very long time anyway, so you'll end up with nothing changing.
Yes, I imagine if you have a large 2 story house with 30 metre cable runs you would need to use a backbone of solid bus bars, with shorter runs of cable to the outlets.
As I said, I have my doubts it would be a win anyway - for much the same reasons: I^2R losses and because I'm not sure that large switchmodes are significantly more efficient than smaller ones anyway.
Let's say you're using the normal 14 gauge solid core wire commonly found in houses right now for a 5V supply. A 200ft run of that in the house (both ways) isn't out of the question if there's a central supply in a 2 story building. That's about 1 ohm [1], which doesn't sound like much at first until we look at what it ends up meaning.
A phone might take an 1 Amp to charge at it's fast rate (tablets 2 or more), so we've got 1 Amp, with a 1 Ohm resistance. That means we're going to have a 1 Volt drop to that outlet, which means the phone is likely not going to charge at the correct current since it'll only see 4 Volts at the outlet. It also means you've just had a 20% loss in the power line in order to have a larger central power supply.
Combine that with the fact that most Switch-Mode Power Supplies aren't actually that efficient when not under load (part of the problem) then it might not be unreasonable to have it running at 80% efficiency to begin with. At this point you're down to 64% total efficiency instead of the 80-90% that the small wall warts end up getting you.
You could use a higher voltage, 48V isn't uncommon because of the telephone industry, but then you need conversion at the outlet again and end up with the same problem you've got with wall warts.
The other option is to use more copper to carry the current, say 10 gauge, which brings the losses to 0.4V and about 10% losses, but you need 150% more copper than 14 gauge.
[1] http://www.cirris.com/learning-center/calculators/133-wire-r...