The Internet, as a general purpose technology, increasingly surrounds us everyday and has become an integral part of our lives. It has shaped how our world operates in daily activities ranging from our working day to how our leisure time is allocated. By participating in the Internet, every user both extends its reach and reveals information about themselves through direct and indirect channels. In a way, every participant becomes a real-time survey respondent who self-reveals time allocation preferences in contrast to the stated preference responses of traditional surveys. In parallel, recent decades have seen the urbanisation of humankind such that now cities, and their surrounding urbanized areas, have become the loci of interest in social economics research. In this context, we provide a highly granular daily dataset of internet activity which traces online/offline behaviour in 15 minute intervals for 900 major cities worldwide for an average of 800 days. This dataset provides a fundamentally new perspective on human behavior at a global scale yet with high temporal granularity. We demonstrate how humanity's interaction with the physical internet can reveal new insights, such as how time-use preferences change due to global city transport dynamics, religious festivals and our everyday night time rest.