In both commercial and defense sectors a compelling need is emerging for rapid, yet secure, dissemination of information to the concerned actors. Traditional approaches to information sharing that rely on security labels (e.g., Multi-Level Security (MLS)) suffer from at least two major drawbacks. First, static security labels do not account for tactical information whose value decays over time. Second, MLS-like approaches have often ignored information transform semantics when deducing security labels (e.g., output security label = max over all input security labels). While MLS-like label deduction appears to be conservative, we argue that this approach can result in both underestimation and overestimation of security labels. We contend that overestimation may adversely throttle information flows, while underestimation incites information misuse and leakage.In this paper we present a novel calculus approach to securely share tactical information. We model security metadata as a vector half-space (as against a lattice in a MLS-like approach) that supports three operators: Γ, + and ·. The value operator Γ maps a metadata vector into a time sensitive scalar value. The operators + and · support arithmetic on the metadata vector space that are homomorphic with the semantics of information transforms. We show that it is unfortunately impossible to achieve strong homomorphism without incurring exponential metadata expansion. We use B-splines (a class of compact parametric curves) to develop concrete realizations of our metadata calculus that satisfy weak homomorphism without suffering from metadata expansion and quantify the tightness of values estimates in the proposed approach.