Is the internet spatial?

Abstract

One of the internet's design principles is to make physical distance transparent, which has made it a motor of globalization and economic development. However, the physical world within which it operates is spatial, and reduction of actual spatial distance is a factor of high impact leveraged, e.g., in today's commercially successful content distribution networks as well as new technologies, such as edge computing. Predictable network behavior is key for reliable intelligent environments. It is therefore worthwhile to study whether and how spatial the internet actually is. We studied a small set of globally dispersed public universities on six continents, recording delays reported by the ping tool to a server in the Pacific Northwest area of the U.S. at four different times of the day (morning, noon, evening, midnight) and over both weekends and weekdays, thus retrieving more than 17,210 data points. We then determined the Pearson's correlations between distance and delay and between hop count and delay. The experiment yielded that there is an overall correlation of 0.57 (a border case strong correlation) between distance and delay and a weaker correlation of 0.4 (medium strength correlation) between hop count and delay. The results suggest that there is a physical distance effect and that this effect is stronger than the hop count effect.