In modern database systems, the representation of data is often degraded or obscured either artificially (for security and privacy) or inherently (because we do not have perfect knowledge, e.g., sensor and scientific data). Indeed, there is an intrinsic connection between privacy and uncertainty as studied recently [A08]. We say that any such data is obscured if it is not in its clear or definitive form. Accordingly, obscured data falls into two categories: one that is fully recoverable (e.g., by obtaining the decryption key) and one that cannot be fully recovered. Obscured data poses a great challenge for query processing. For example, to address the case in which the server is not trusted to decrypt the data, we have devised efficient algorithms that use a homomorphic encryption scheme which allows us to answer aggregation queries on the ciphertext [GZ07b]. The goal of this work is to understand the fundamental techniques that would be needed by a query processor for obscured data and to have a workable architecture for query processing in the two kinds of obscured data.