The Plan: Saturday and Sunday was skiing in Vermont with the family, back Sunday night, everyone off to school Monday morning, then to the airport to get to Miami, Tuesday and Wed meetings down there, then fly out Wed night so I’m back to teach classes Thur and Fri here in Boston. Easy Peesie!
Actual: Finish ski trip feel a little “strange” on flight down to Miami, slight chills. Tuesday do meetings just barely, Tuesday night it hits like a freight train, The Flu, or Malaria, Aliens. On Wednesday the freight train had fully passed over me so it stopped and spent Wed backing up over me again. Thursday it went forward again over a pile of human jello. I’m not flying out for the clases on Thur and Fri, or anytime soon for that matter. Just getting home is my mission for the week.
MTBF of a human: The Thursday Friday class had been a long time in the works. There were a lot of people involved and other activities based on it being completed as scheduled. So if it hinged on one key factor, me, shouldn’t we have evaluated the likelihood of this singular critical component failing? So we can do the whole MTBF of a human is like 1,200 years for non-fatal random failure due to accident or injury. Mean time to pair (MTTR) for an illness is like a week and a half, so human availability is , bla bla bla. Who cares the class didn’t happen!
The bottom line is that the classic strategy of dealing with the unreliability of humans should have been addressed. Have an understudy. Methods like redundancy are a better strategy for defending against critical single points of failure in serial systems. The reliability of a paired redundant (parallel) system is exponentially higher than a single highly reliable component. It would have been worth the extra cost to have a second person participate in the coursework development. The ROI compared to the reschedule expensive would justify it.
How many components in your products do you invest tremendous engineering and refinement to create to be highly reliable? Getting a critical singular failure point component from a 99.2% reliability to a 99.96% reliability can be 1 year and $1.2 million dollar endeavour. How much would it cost to simply create a redundancy in the singular point with a duplicate of the 99.2% component. The redundancy would make that singular point with two old 99.2% components be 99.994%. So with almost zero development time and project budget that singular part of the system is now the strongest by orders of magnitude.
Look for chances for your understudy.