Although there is consensus that datasets should be treated like “first class” research objects in how they are discovered, cited, and recognized, this is still far from a reality. Datasets are poorly indexed by search engines, and they are rarely cited in formal reference lists. A solution that a number of journals are implementing is to publish discovery and citation proxy objects in the form of peer-reviewed “data papers.” A strength of this approach is that it requires dataset creators to write up rich and useful metadata for the paper, but an accompanying weakness is that busy creators are not always willing to invest the necessary time and energy. To enhance dataset discoverability without burdening creators, EZID (easy-eye-dee) will begin using dataset metadata to automatically generate lightweight, non-peer reviewed publications that will increase the exposure of the metadata to search engines. EZID (ezid.cdlib.org) maintains public DataCite metadata records for over 167,000 datasets, any of which could be viewed as HTML or as a dynamically generated PDF. In cases where the creator has submitted only the required DataCite metadata, the document will function as a cover-sheet or landing page. If the creator chooses to submit optional Abstract and Methods metadata (over 2,000 records already contain Abstracts), the document expands to more closely resemble a traditional journal article, while retaining the linking functionality of a landing page. A potential bonus is that providing an incrementally improved document in exchange for the effort of submitting incrementally improved metadata may encourage authors to submit more than the minimum required metadata.