There must be at least 50% loss in this. AC was designed for transmission lines, which run for miles.
While true that AC is very efficient for use in transmission (mostly due to the simplicity of stepping up the voltage as well as smaller line losses) the losses he would be experiencing are probably closer to ~10%. Modern computer power supplies and inverters run somewhere around 90-98% efficient. Also a run of DC in "meters" could cause significant voltage drop if the conductor was not seriously over-sized for the current. As long as there is any sort of real distance to be covered and bus bar systems are not being considered, inversion from DC and then later conversion does not cause the kinds of losses you are alluding too.
The simple example of the system employed here is the battery backed UPS next to your tower. The enterprise grade UPS here shows a system efficiency at full load of ~93%. This gives us the efficiency of a system that is AC->DC -> DC->AC and by simply rearranging the order of the modules we will get the DC->AC -> AC->DC. While this does not include the actual modules used in the system it does show the current order of efficiency that this system can be operating at.
In the article it was also stated that the panels were only producing at 90.16% of rated which is overall damn good considering a panel rating of +/- 5% and the variability of nature. Additional considerations should also be taken in the non-standard voltage produced (41 VDC) in the panels which must be either boosted to 48 VDC or bucked to 12 or 24 and regulated for use as a stable DC supply.
Overall the losses in conversion can pale against line losses through a DC system the size of a house including the regulation that would either need to occur on the entire system or more practically at every device.
If it has syntax, it isn't user friendly.