In long established research organizations, it is necessary to safeguard good research ideas originating at lower organizational levels. The upward progress of unusually good ideas in an organizational hierarchy may be compared with that of ordinary ideas by means of a mathematical model, with the assumption that ideas follow a Poisson distribution and Markov random chain walk with an absorption level. Good ideas are novel, synthesizing, simple, aesthetically appealing, problem oriented, and synergistic. The problem of acceptance arises with the interaction of good ideas with the management structure, the good ideas being embedded in a matrix of ordinary ones. Good ideas are more extensively screened than others, having a much lower probability of acceptance at the first review level; they are vetoed so effectively that only one in thirty is ever adopted. Methods for improving the survival probabilities for such ideas include extraofficial documentation, exterior reinforcement, and the bypassing of supervisory cognizance. An organization can become more hospitable to such ideas by reducing the number of levels in the hierarchy and by promoting within the organization on an unselective basis rather than solely on the basis of merit or qualifications. (Mathematical data are included.) (DF)