Phase separation is familiar and useful, yet opportunities to manipulate it are surprisingly subtle and complex. Two just-published papers 1,2 remind us that in approaching any new scientific problem, there are generally three stages of reaction. Our first impression is of difference and strangeness. But once superficial dissimilarities are pierced, our feeling becomes the opposite extreme. Everything now appears fundamentally the same. All that we can now see are the underlying commonalities, with different forms according to their settings. Only in the third stage does real knowledge begin. We look anew for differences, not this time for those obvious elements that hit the eyes of newcomers, but rather for subtleties whose importance structures the problem. Phase separation is an evergreen subject that reinvents itself continually. Starting early on, scientists and engineers sought to control items such as steel microstructure and to understand workaday items such as why oil floats on soup, leading in the 19th and early 20th century to focus on nucleation and growth 3. Then came a surprise-the identification of spinodal decomposition 4 ; it was followed by an era of exploring the spinodal decomposition formalism in many manifestations, a pattern that to some extent continues today. Independently, explosive advances follow from progress in understanding critical points, an accomplishment recognized by a Nobel Prize in the latter part of the 20th century 5. Surveying the ensuing scientific literature one finds that the phrase "phase separation" now is used progressively less often. During the 21st century, this loss is balanced by increasing the use of the phrase "self-assembly". For experimentalists, much of the distinction between these concepts can be semantic. Both concepts, phase separation and self-assembly, share the concern with understanding how order appears from disorder. A new coarsening mechanism in phase separation 1 and bicontinuous nanoparticle gels 2 are reported from the groups of Hajime Tanaka at the University of Tokyo and Yun Liu at the University of Delaware/National Institute of Standards and Technology, respectively. On first reading, the subject seems to differ from what professors teach students about phase separation-where are the phase diagrams, tie lines, and other standard thermodynamic quantities? These studies, dwelling instead on how processes evolve in time, seem alien to the textbook 3 view. On the second reading, the phase-separated structures are familiar-coarsening 1 , a standard feature of nucleation and growth, and bicontinuous structures 2 , standard for spinodal decomposition, so from this perspective, readers can feel comfortable. It requires third reading to see where these interesting studies go beyond the standard view. Both studies exemplify how a mere difference in mobility can influence even the long-time structural appearance of two separating components-one of them relatively fast, the other relatively slow. This problem doesn't present itself in the usual textbo...