It's quantity per capita. There's something like 12 x more people in NYC than in Detroit. That means if there's a 1,000 pizzerias in NYC, it would take only 90 in Detroit for Detroit to outrank NYC. Their methodology is doubtlessly flawed but it's a little more clever than just counting absolute totals (as New York would on account of its size win every category by default which wouldn't really be representative either - after all 600,000 people need fewer restaurants than 8 million people to have a similar perception of choice).