We analyze the microlensing event OGLE-2011-BLG-0173 , which shows a small perturbation at the end of the microlensing event caused by the primary lens .
We consider both binary lens and binary source models and we explore their degeneracies , some of which have not previously been recognized .
There are two families of binary lens solutions , one with a mass ratio q \approx 4 \times 10 ^ { -4 } and a separation s \approx 4.6 and the other with q \approx 0.015 and s \approx 0.22 , i.e , both have companions in the planetary regime .
We search for solutions by using Bayesian analysis that includes planet frequency as a prior and find that the s \approx 4.6 family is the preferred one with \approx 4 ~ { } \mathrm { M _ { Uranus } } mass planet on an orbit of \approx 10 ~ { } \mathrm { AU } .
The degeneracies arise from a paucity of information on the anomaly , demonstrating that high-cadence observations are essential for characterizing wide-orbit microlensing planets .
Hence , we predict that the planned WFIRST microlensing survey will be less prone to these degeneracies than the ongoing ground-based surveys .
We discuss the known low-mass , wide-orbit companions and we notice that for the largest projected separations the mass ratios are either high ( consistent with brown dwarf companions ) or low ( consistent with Uranus analogs ) , but intermediate mass ratios ( Jupiter analogs on wide orbits ) have not been detected to date , despite the fact that the sensitivity to such planets should be higher than that of Uranus analogs .
This is therefore tentative evidence of the existence of a massive ice giant desert at wide separations .
On the other hand , given their low intrinsic detection sensitivity , Uranus analogs may be ubiquitous .