The gap in question is the one between genome and phenotype – a knowledge gap – and one that we will struggle with for some time, despite, or perhaps because of, the rapidly advancing field of epigenetics. One of the “messages” that systems biology has communicated up to now is that (counter to previous thinking) biological systems tend to be highly robust – a necessity for embryonic development and adult homeostasis, e.g., in the sense of gene networks. And one of the “messages” emerging from RNA epigenetics is that these small molecules are probably the fine-tuners of network properties, rather than the grand orchestrators. However, in crucial respects, those messages might well be wrong, or at least have very important exceptions. A strong message on the power of epigenetics (in its broad meaning) was sent recently by the twin study of Baranzini et al. in Nature,1 which demonstrated no detectable difference at the level of genome sequence, epigenetic marks or gene expression levels (mRNA) in CD4+ lymphocytes in monozygotic twins, one of which suffers from multiple sclerosis, the other of which is free of the disease. But what about non-coding RNAs? A somewhat contrary picture to that of non-coding microRNAs being merely fine-tuners is suggested by findings that they are responsible for some pivotal and large changes in development and function of immune cells (for a review see Ref.2). Those gene expression networks are certainly not robust to several-fold increases in microRNA concentration. As presented in two articles in this issue, small non-coding RNAs are revealing themselves as ubiquitous players in regulation, constituting a hitherto unimagined proportion of the non-coding genome (Costa3) and possibly embodying the very substrate of epigenome-environment interactions itself (Mattick4). One of the major gaps in our knowledge of epigenetic mechanisms is how an environmental stimulus is translated into the epigenetic modulators of gene expression networks: what do the signal transduction pathways from cell membrane to network look like? I will hazard a speculation that epigenetic stimuli acting through such pathways can, in principle, have very large effects because of the cascade nature of the pathway, and the amplification of the signal at various points. That is what one would expect if the stimulus is a normal part of development or response of the immune system, and must make a pivotal change in response to a short-lived “spike” of external ligand. Should a transient environmental stimulus activate such a powerful modulator of gene expression in an aberrant way (wrong cell, wrong time), it might well not leave a signature in the sense of epigenetic marks or mRNA levels. It might cause a permanent change at the level of non-coding RNA levels, but to all intents and purposes, it would be impossible to trace its cause. Still, looking for increased microRNA levels as surrogate markers for early diagnosis would appear to be a wise strategy, as studies from the field of cancer research demonstrate (for a review see Ref.5). We might envisage an environmental stimulus acting on a protein network by tilting the epigenetic landscape – à la Waddington – or distorting it in a certain region, hence causing the marble to roll drastically to a different stable equilibrium. The usefulness of Waddington's concepts of epigenetics is noted in two articles in this issue in relation to explaining evolutionary genetics via systems biology (Bard6), and the generation and modulation of phenotypic variation (Jamniczky et al.7); by extension, it might be good to apply Waddington's epigenetic landscape systematically to investigating the kinds of environmental stimuli that cause disease: imagine distorting the landscape of a particular network in a computer simulation, and seeing where the new equilibrium comes to rest. If that new equilibrium mimics the disease state, the point of distortion would be a good place to start looking for the more distal point of action of the environmental stimulus. In this way, “environmentally vulnerable” components in networks could be identified. Moreover, it might tell us which kinds of environmental, or medically administered, stimuli might reverse the epigenetic disease state. Where does that leave the predictability that science and biomedicine hanker after? When we are born, are we dealt a “poker hand,” the strength of which unfolds as other (environmental) players in the great game of life lay down their cards – partly, and forever, beyond our control? The analogy is tenable in certain respects (e.g., clear mono-genetic diseases excluded), but a greater appreciation of the strong modulatory effect of epigenetic factors coupled with a development of Waddington's principles in the light of systems biology might enable us to glimpse our opponents' hands, or even replay a move in the game.