Not all BOINC project support the combination ARM+Linux (sometimes they only support the use of ARM with Android, or no ARM at all). Also, I do not know what stressing out the computer so much for prolonged time does to it over time. I am not saying to not do this, but please be aware of this.
You cannot determine median based on range alone. You basically need to know how your datapoints (people in this case) are distributed to be able to actually calculate the median, because the 50th percentile does not have to lie smack in the middle between the extreme ends of your range.
That same article in your link already states that the etymology is rather unclear and could have alternative explanations. And i am quite certain that Alyaza was not meaning to use those terms in bad faith.