Tentative reviewers: We had a number of reviewers who did not reply / were tentative. With the aim of getting 3 reviews per submission and helping the meta-reviewers, we send out more invitations, and once we achieved the quota, we deleted the tentative reviewers. It turned that some of the reviewers were having an independent conversation with the meta-reviewer, and perhaps agreed using another medium (e.g. email). We recommend that it is OK to leave tentative reviewers hanging in the system even after inviting new reviewers. This may have a double advantage: on the one hand we can't really know in advance if tentative reviewers will eventually respond nor if accepted reviewers will in fact comply with their promises, so the worst-case scenario for having 5 or 6 external reviewers with some "pending" is to end up with a few extra reviews for a few papers; on the other hand, by leaving them hanging in the system, by the end of the reviewing period we would have a clear picture of the "unreliable reviewers", maybe to serve as a heads-up for the next chairs (sort of a blacklist).