On similar shops operating in close proximity - it makes sense to cluster in order to maximise the number of customers you have access to. Imagine two ice-crea

On similar shops operating in close proximity - it makes sense to cluster in order to maximise the number of customers you have access to. Imagine two ice-cream vendors on the beach, selling identical ice-creams for the same price. Since number of sales are a function of how far people have to walk to your stand (and price etc), you want to make sure you are as close to as many people as possible. If your competitor is closer to them, they will go there instead, so you either need to be very far away or right next to the competitor. If a person could visit either of your stands, the competitor can 'steal' them by simply moving their stand closer to yours, so that the customer previously in the middle is now closer to them. A discussion on the general case is here [0]. This principle obviously doesn't extend flawlessly to the real world, but is a significant contributor to, for example, the clustering of car dealerships today. [0] https://math.stackexchange.com/questions/583157/ice-cream-vendor-problem https://math.stackexchange.com/questions/583157/ice-cream-ve...