A large apparent violation of the OZI rule has recently been found in many channels inpp annihilation at LEAR. An interpretation of these data in terms of the "shakeout" and "rearrangement" of an intrinsicss component of the nucleon wave function is proposed. This gives a channel-dependent, non-universal modification of the naïve OZI prediction. Within this approach, we interpret the strong excess of φ production in S-wavepp annihilations in terms of the polarization of the nucleon'sss component indicated by deep inelastic lepton-nucleon scattering experiments. This interpretation could be tested by measurements of the f ′ 2 (1525)/f 2 (1270) production ratio in P-wave annihilations and by experiments with polarized beams and polarized targets. We also propose a test of the intrinsic strangeness hypothesis in φ production in high-momentum transfer processes, via a difference in constituent counting rules from gluon-mediated production.