Ghosting has been a problem for me. I have 2 R2's and decided to build my own machine from EU3030 extrusions using the R2 mechanical design as a reference. It has 4 posts for the bed and 2 z steppers so it does not vibrate relative to the hotend. It uses all of the R2 code and thinks it is an R2. I changed the bearing blocks to not have springs and to use open loop belting. My hotend removed the sloppy LM6LUU bearings and put in bronze bushings. I now had zero slop and not surprisingly my ghosting nearly disappeared. I now put the revised hotend and bearing blocks on an R2. This reduced the ghosting quite a bit, not as much as my home brew machine but a big difference none the less. Now I wanted to get the same results on an R2 with the minimum amount of change. Having determined that the largest ghosting component was due mostly to a loose hotend (bushing block component is unknown because all my machines have been modified) I decided to put polymer LM6LUU bearings in the original R2 plastic. This was a disaster, the bearings are so tight that they were useless. I put back in the steel LM6LUU bearings as a sanity check. Everything moved freely, too freely and the hotend now had play. I decided the loosen the gears and move the rods such that the hotend had some resistance and no play. I did this for both axis and lo and behold the ghosting was much reduced. I have not measured how much out of square I have made the machine, but it felt so slight so as to not matter. So if ghosting is a major problem you can experiment with adjusting until the play in the head is removed.