Take a look at: https://en.wikipedia.org/wiki/HVDC_converter It mentions IGBT based commutators which do not require AC machines, but which are less efficient.
I think all the thoughts about the switching harmonics being horrendous are correct.
If you are working on an electronics project which is sensitive to noise, you currently have the option of ditching your cheap, efficient DC-DC bench power supply and instead using a linear power supply, which is far less efficient (instead of producing the desired voltage by converting power, it does so by turning the extra power into heat), but is a very low-noise design.
If we switch to a DC grid, there'd be no point in using a linear bench supply to develop a low-noise project, because those nasty switching spikes you were trying to avoid aren't coming from your supply, they are coming from the grid!
Not to mention Biot Savart Law that says a long wire generates B ~= I/r
So where the electric current on average nullifies the Magnetic field in AC (other a period), on DC you would have to build with a constant F = qv vect I/r which means a constant magnetic field appearing and inducing resilient constant forces for every other charges in movement. 1/r decrease way less than 1/r² // 1/r3. That is what CE norms on electro pollution wants to avoid. A strong electromagnetical residual signal that can interfere with other electromagnetic devices.
I understand the concerns of persons that never had to learn electricity about the complexity of using Z (impedance) other R (resistance). But Fresnel law makes AC quite easy to grok.
In some countries it is accepted since a long time that the bug in AC is the side effect of loosing the phase (hence introducing peaks up/down phase). That is the reason why companies have to pay not only for the whattage but also for the disruption of phase. Some companies even get discount to regulate the phases (like train companies that uses a lot of electricity in peak fashion at trains start/stop and can accumulate energy selectively to charge huge capacitors and use them to correct a shift of phase).
I sometimes look at the electric cable in my cities. The way they are plugged anarchically and the transformers looking like not maintained, the electric boards with expensive gross protection, the corrosion at the connection points (yes surface effect with rust and water sure it does do weired loss) the fuse used in electricity/electronic devices .... And I feel like I was in Cuba in the 1986 in the middle of embargo.
I don't live in Cuba, this I see in Montréal, SF, miami ...
In supposed developed countries and I wonder if my eyes telling me it is broche à foin are right or my brain listening to all the assertive geeks and citizens telling me the opposite.
You know what, I do not know. I still have a strong feeling technology on the new continent is not as good as it is publicized.
If the grid were DC, what would do all of that stepping down? Is he proposing replacing all of those transformers with an army of industrial strength DC-DC converters? Wouldn't that eat into the efficiency gains he is proposing (I would also imagine that that gigantic active switching solutions are more expensive and less reliable than gigantic transformers).
My first thought at a suggested alternative was "why not try increasing the frequency of the distribution AC?". This redditor explains why that's not ideal: https://www.reddit.com/r/askscience/comments/1u45rz/why_dont...