In the previous post, I compared randori to a physics experiment.
In a science lab, experiments are often performed over and over and over again, carefully controlling the experimental variables and changing one variable at a time to see how the varaibles affect the system.
Randori can be very profitable when approached in a similar vein.
Often it is not the end result (someone falls down or gets in an armlock) that is the most interesting thing in an engagement. Often the really interesting part comes 2-3 steps before the fall or the terminal lock. In such instances, it can be very educational to set up a given condition in randori and then run it over and over again so that you and your partner can experience how that certain situation unfolds.
One example might be repeating release #1 into oshitaoshi (katatetori ikkyo) over and over again seeing how uke falls each time. Pretty soon, uke (without trying too hard) will start to walk out of your technique, either forcing you to modify your technique or perhaps reversing the technique on you.
Or, you might start in kesagatame and try repeatedly to transition into tateshihogatame with a cross-collar choke. Even if you try to be really precise every time, pretty soon you'll start to see variation slip in, and both partners will be getting a ton ov experience inside and around and related to those particular positions and transitions and techniques.
So, try approaching your randori as a scientific experiment, controlling certain variables and letting others change and see the results.--