Microlensing planets in M22: free-floating or bound?

We use detailed numerical simulations and theoretical estimates to show that, if confirmed, the unusually brief microlensing events observed by Sahu et al. (2001) in the field of the globular cluster M22 might be explained as a result of microlensing by a population of clustered MACHOs, a dark cluster or RAMBO, not associated with the globular cluster. If real, this dark cluster would be located between M22 and the Galactic bulge and could include at least $10^6$ substellar members with a typical size of 1-3 pc. Bound planets in wide or/and eccentric orbits are also able to reproduce the observed microlensing behaviour, but only if multiplanet systems (including large Kuiper-belt-like objects) are abundant, although, our calculations argue against the latter scenario as the ionization rate in M22 is very high. Dynamically ejected or lone planets are, in principle, incompatible with the observational findings as they either escape their parent cluster in a relatively short time-scale after ejection or segregate toward the outskirts of the cluster. We discuss additional implications of the dark cluster scenario, including the existence of a population of RAMBOs toward the Galactic bulge.