This article presumes acres for $420,000 in San Francisco. Not sure where the author is finding land values like that in the Bay - certainly not in San Francisco, where land is closer to $100/sqft at the low end ($4.5M).
The next proposal, "low rise" apartment block:
> Let's assume a 6000sf building footprint. This would contain twenty 250sf units per floor, with some shared space. Four stories means eighty units. With 80 units, we have 80 x $10,000=$800,000 budget for land. Our 6000sf building footprint sits on a quarter-acre (10,500sf) of land, so the land costs could be as much as $800,000 X 4=$3.2 million per acre.
And next, the "medium-rise" plan:
> Next, we have a six-story structure, which is about the upper end of the Traditional City or "low-rise" format. We will step up here to a better standard of construction, such as brick or steel-frame, with a cost of $200/sf. This reduces our unit size to a fairly tiny 120 square feet, plus some common areas. Our footprint is still 6000sf, but since the unit size is smaller, we can get 40 units per floor, or 240 units for a six-story building. At $10,000/unit for land, that gives us a budget of $2.4 million for land. We will use more of an "attached" format here, with a little space in the back, or a 7000sf (sixth of an acre) plot. This translates into $2.4m x 6=$14.4 million per acre land and other costs.
The author is saying something about the density required to support affordable housing in San Francisco.
I imagine the low-end is substantially less than the average, or could be if SF zoning and permitting wasn't so screwed up and we were actually building more low-end housing. The point of the article is to show what's financially viable if the system weren't so screwed up; that low-end housing is viable even in the face of absurd land-value prices.