This article studies the effects and implications of sybils (secondary accounts created by a person in an online platform) through the game World of Tanks from interdisciplinary, mixed-methods perspectives. Considering sybils allows us to access a "person based'' network, instead of an "account based" network, revealing formerly undetected patterns. We move on to behavioral differences between the "parent" (the initial account) and "child" (those created afterwards) accounts in a sybil relationship. We explore the behavioral patterns of sybils using network, chat, and gameplay data. We find that sybils represent players experimenting with new roles or features without damaging their play record. We find that there are significant behavioral differences between different sybil accounts, and we leverage them to build a machine learning classifier to differentiate sybils. This classifier is able to identify sybil accounts with over 95% accuracy for sybil/non-sybil, 61% for parent/child. This study demonstrates the underexplored but rich potential for sybils to improve research and industry practitioners' understandings of user practices and experiences.