The paper has two main tasks: to trace the systematic connections between two recent pieces of epistemological work by David Henderson and Terry Horgan, and to criticize as unintelligible the concept of epistemically relevant possible worlds, which is central to one of them. Iceberg Epistemology sketches a general account of the structure of our cognitive organisation, which can, by and large, be classified as an externalist, reliabilist account. I argue that Henderson & Horgan's new objective epistemic value (labelled robustness of reliability and loosely defined as reliability across a wide range of epistemically relevant possible worlds) is introduced to clarify the connection between reliability and the ideal of truth (and by that to remedy a standard shortcoming of externalist accounts). However, the concept of an epistemically relevant possible world seems to be unintelligible (at least on the basis of the introductory examples offered so far). Some concluding remarks address the status and the achievability of robustness of reliability as an epistemic ideal.