I think the article puts forward one point - if the number of devices using DC is large enough, the load will become much less variable. Less variable loads mean one power supply can operate at peak efficiency rather than having many sitting idle most of the time.

I'm not completely convinced that this makes sense though. Any low voltage DC distribution network will need fairly thick wires to distribute anything but trivial power levels. The power lost in the cables over tens or hundreds of feet would probably make any efficiency gains fairly moot. The efficiency of modern switched mode power supplies is fairly high, and my understanding is that the no-load power use is almost trivial (milliamps).

One issue that seems to have surfaced recently is that there's loads of very, very cheap AC-DC power supplies which are awfully made. It would definitely make sense for a single high quality supply over a bunch of cheap, noisy and dangerous ones, but I think the best solution is investing slightly in wall-socket based power supplies of a known quality.

As an aside, power transmission with DC is actually quite a lot more efficient than AC. The reason AC won the power race was the ease of voltage change with transformers. As solid state power electronics improves, we're going to see more and more DC based transmission lines. We're probably centuries away from a DC grid (if ever).

The problem with "buy high quality wall socket supplies" is that it is impossible these days to tell where the line is between "high quality, but off-brand" and "low quality, being sold at a huge mark up".

I learned a few years back that higher cost doesn't necessarily equal high quality.

"power transmission with DC is actually quite a lot more efficient than AC."

And the subject to google for is insulation breakdown and compare the equations relating peak voltage and RMS voltage. Intuitively you need to insulate a high power transmission line to handle the highest voltage it'll ever see at any instant, which and unfortunately the peak voltage is a microscope amount of time at the tiny peak of the sine wave, but the DC equivalent power you can get out of a sine wave is only the much lower RMS voltage, whereas a DC line can carry peak power continuously, so for a given line $100M budgeted toward cable insulation can insulate more power if its sent as DC rather than AC.

This is a simplification, but take a bell curve and flip it upside down and that graph applies to efficiency, economic cost, and ecological damage per watt for DC power supplies along the y-axis and wattage along the x-axis.

At 2013 technology levels its a real bad idea on many different levels to provide 100 total watts of DC power via 50 separate power supplies to 50 devices when you could have just one DC supply.

As an example of sillyness, imagine if every chip inside a desktop PC required a completely separate wall wart power supply instead of sharing one common DC supply. Thats basically what an elaborate laptop setup replicates if you have a lot of external devices, especially if you're proposing running your room lights off the same system, etc.

The same curve at the higher end rather than lower explains why you'll likely not be running a kitchen stove or a HVAC system off low voltage DC anytime soon. Although that would be intellectually interesting to see. A massive array of mosfets to switch, and power leads thicker than water pipes (maybe actual copper water pipes for cooling?). It would be a sight...

For large datacenters, DC can be a big improvement. http://gigaom.com/2012/01/13/the-next-big-thing-for-data-cen... But this is mainly where UPSes are involved, or solar power. That might not be relevant to a house, but it could have other advantages besides efficiency. Imagine all your computers had tiny DC-DC converters instead of heavy, hot bricks of inverters. Laptops would only need a cable to charge instead of a power brick.

If you move the conversion one step further up, then you can get much better conversion efficiency and reduce wasted power by 'idling' wall warts.