The quality of volunteered and crowd-sourced spatial data is not in most cases audited prior to being made accessible to end-users. Studies have shown that this spatial data varies significantly in terms of its geometric quality, its semantic consistency, in terms of its comprehensiveness of coverage and in terms of its currency. Subsequently it often compares poorly with the authoritative data capture and mapping undertaken by national mapping agencies and commercial companies. In this paper we highlight a specific type of problem encountered with volunteered geographic information (VGI) – the naming of real-world features. Many Location-based Services (LBS) applications are using VGI as a spatial data source. The volatility in VGI, as shown by the results in this paper, will require LBS developers to carefully consider how they manage and use the spatial data sources generated by VGI and crowd-source paradigms.