Abstract
This paper describes a range of problems for discovering (mainly linguistically interesting) data via the CLARIN Virtual Language Observatory. It analyzes te causes of these problems and makes a range of suggestions on how the situation can be improved.
