I refined the patch a bit. If anyone's interested, here's details:

I attenuated and offset the A118 to obtain a random voltage X between
0 and T (the A150's threshold voltage). The bottom row of A155 knobs
was also scaled to obtain a probability voltage P, also between 0 and
T. This was added to the random voltage in a mixer.

For each sequencer step, the X+P voltage is sampled, and the output of
the S&H is patched to the A150's CV input. The A150 is used to gate
the A155 triggers.

The triggers are delayed slightly by the A162, in order to give the
A148 and A150 time to operate. The A162 output is attenuated to avoid
the A150 bleed-thru bug.

When a probability knob is set fully left, the X+P voltage will always
be below the A150 switching threshold, and the step will never produce
a trigger - probability 0%, the same as switching the step OFF.

When a probability knob is set fully right, the X+P voltage will
always exceed the A150 switching threshold, and the step will always
produce a trigger - probability 100%, the same as switching the step
ON.

When a probabiliy knob is set between these two extremes, the X+P
voltages will be randomly distributed above and below the A150
switching threshold. Turning the knob more to the right or to the
left increases or decreases the likelyhood that the switch will close
to produce a trigger for that step.

Joe