Because people disproportionally follow defaults, both libertarian paternalists and marketers try to present options they want to promote as the default. However, setting certain defaults and thereby influencing current decisions, may also affect choices in later, similar decisions. In this paper we explore experimentally whether the default bias can be reinforced by providing good defaults. We show that people who faced better defaults in the past are more likely to follow defaults than people who faced random defaults, hurting their later performance. This malleability of the default bias explains certain marketing practices and serves as an insight for libertarian paternalists.