Summary: We propose a new measure for eternal inflation, based on search optimization and first-passage statistics. This work builds on the dynamical selection mechanism for vacua based on search optimization proposed recently by the author and Parrikar. The approach is motivated by the possibility that eternal inflation has unfolded for a finite time much shorter than the exponentially long mixing time for the landscape. The proposed accessibility measure assigns greater weight to vacua that are accessed efficiently under time evolution. It is the analogue of the closeness centrality index widely used in network science. The proposed measure enjoys a number of desirable properties. It is independent of initial conditions and oblivious to physical vs comoving weighing of pocket universes. Importantly, the proposed measure makes concrete and testable predictions that are largely independent of anthropic reasoning. Firstly, it favors vacua residing in regions of the landscape with funnel-like topography, akin to the energy landscape of naturally-occurring proteins. Secondly, it favors regions of the landscape that are tuned at dynamical criticality, with vacua having an average lifetime of order the de Sitter Page time. Thus the predicted lifetime of our universe is of order its Page time, \(\sim 10^{130}\) years, which is compatible with Standard Model estimates for electroweak metastability. Relatedly, the supersymmetry breaking scale should be high, at least \(10^{10}\) GeV. The discovery of beyond-the-Standard Model particles at the Large Hadron Collider or future accelerators, including low-scale supersymmetry, would rule out the possibility that our vacuum lies in an optimal region of the landscape. The present framework suggests a correspondence between the near-criticality of our universe and dynamical critical phenomena on the string landscape.