We tried Olmakhan for a while, but that yielded much higher rates of researcher attrition.