I think that someone posted some examples of what would happen if processes were 99.9% accurate. This had examples of how many people would die in surgery, how many planes would never make it to their destination, etc. I am having problems searching for it, probably my own ineptitude....

Can someone point me to this, my boss saw this same information somewhere and now wants me to include it in my annual refresher training that is coming up (tomorrow:mg: ) I am glad that I work well under pressure, and thanks in advance, I just know that you guys will pull my :ca: out of the :blowup:, again. What would I do without you?:thanks:

http://www.dtrdenver.com/gpage6.html

If 99.9% Were Good Enough, then...

Two million documents will be lost by the IRS this year.
22,000 checks will be deducted from the wrong bank accounts in the next 60 minutes

1,314 phone calls will be misplaced by telecommunication services every minute.
12 babies will be given to the wrong parents each day.
103,260 income tax returns will be processed incorrectly this year.
2,488,200 books will be shipped in the next 12 months with the wrong cover.
5,517,200 cases of soft drinks produced in the next 12 months will be flatter than a bad tire.
Two plane landings daily at O'Hare International Airport in Chicago will be unsafe.
3,065 copies of tomorrow's Wall Street Journal will be missing one of the three sections.
18,322 pieces of mail will be mishandled in the next hour.
291 pacemaker operations will be performed incorrectly this year.
880,000 credit cards in circulation will turn out to have incorrect card holder information on their magnetic strips.
$9,690 will be spent today, tomorrow, next Thursday, and every day in the future on defective, often unsafe sporting equipment.
20,000 incorrect drug prescriptions will be written in the next 12 months.
114,500 mismatched pairs of shoes will be shipped this year.
$761,900 will be spent in the next 12 months on tapes and CDs that won't play.
107 incorrect medical procedures will be performed by the end of the day today.
315 entries in Webster's Third New International Dictionary of the English Language will turn out to be misspelled.

And here: http://elsmar.com/Error_Proofing/sld022.htm

Thanks guys! Jim, I knew that there was something here, I just couldn't find it. Thanks Sydney for all the additional ones. I did find some of those on the internet too. Ah, you just gotta love a Cover for coming to the rescue:D

just an update, I put these together with some others I found. this made a big impression. I did not post the one about 50 babies getting dropped each day by doctors on the slide, but left it to close the session with. There was an audible gasp from the group when I closed with "just remember that if you feel that it is ok to accept a tenth of one percent error; you are saying it is ok that everyday, 50 babies will be dropped by doctors." Thanks, what a closer!

btw, one down nine to go! you gotta love the end of year training crush!

just an update, I put these together with some others I found. this made a big impression. I did not post the one about 50 babies getting dropped each day by doctors on the slide, but left it to close the session with. There was an audible gasp from the group when I closed with "just remember that if you feel that it is ok to accept a tenth of one percent error; you are saying it is ok that everyday, 50 babies will be dropped by doctors." Thanks, what a closer!

Unfortunately, it's very misleading, though. If we're going to use statistics to make predictions, there is some level of baby-dropping that we can confidently predict will happen. We can predict the number of people who will be killed in automobile accidents, or break their necks falling from ladders, or drunken cheeseheads who will drown after driving their pickup trucks onto inadequately frozen lakes. The fact that we can predict those types of unfortunate events has nothing to do, however, with whether or not we consider them "acceptable."

In point of fact, there are instances where some level of defective output is acceptable, and even prudent. We shouldn't be comparing falling babies with responsible use of statistics in manufacturing.

Actually the point was complacency....we all know that these kind of statistics are not really going to change our production processes. We have just begun to see some indication that some folks have begun to slide into that area of "we are doing great, we have been great, therefore, we will continue to do great." (in a softening market) So, it was an excercise in complacency does not lead to continual improvement theme.

But your point is valid, and thanks for sharing.

your point is well taken - although I don't think she was trying to go there...just making a point that 99.9% is not as good as most people believe it to be.

of course we must recognize that the severity of an event contributes to how acceptable any given occurence rate is.

A 99.9% safe landing rate at O'Hare is less acceptable than a 99.9% rate of dropping babies which is less acceptable than a 99.9% rate of 99.9% rate of checks being deducted from teh wrong account...

I just had a similar discussion last week when training FMEA. teh topic came up when we discussed what type of event would warrant a severity rating of 10. In our industry, the worst thing that can happen is someone's pet is misdiagnosed and dies before proper treatment can be administered. Somone wanted to argue that since the death of a pet wasn't as bad as the death of a person/baby then we shoudl use a 10 - we should cap our ratings at say a 7. It took a long time to get this person to accept that by using a 10 for the death of a pet was NOT the same as us saying that the death of a pet was just as bad as the death of a person and that the 1-10 scale was only relative within our business not across businesses and the scale was only used to rank which concerns we addressed first...in our business the death of a pet is bad. not only is a beloved companion animal goen before it's time, a problem that results in misdiagnosis would cast doubt on our custoemr's abilities (Vets) and our equipment and could cause loss of market share which would then result in layoffs... The whole discussion was a nightmare. I had visons of going to HR for being perceived as saying that a dog ws more valuable than a baby!

Sorry, but the whole premise makes no sense. Here's what it's saying: I ask a coworker what percentage of defects he would consider "acceptable" for some process x. He says, 1/10 of 1%. I counter by saying, "Oh, really. If .1% of newborn babies were dropped on their heads by physicians, that would mean 50 tiny babies dropped every day! Do you consider baby-dropping ACCEPTABLE???"

It's needless and gratuitous hyperbole, and has nothing to do with the point, which should be making people aware of process capability, and ways of improving it.

There is a philosophy, that is attributed to INPO (Institute of Nuclear Power Operators) called Human Performance. This philosophy does recognize and accept that humans do make errors. Babies are dropped by doctors. But, we work to minimize the impact of the errors. For example, I assume that doctors limit the height above the surface they hold a baby (to limit the fall) and also the surfaces are likely padded.

I believe what the HPI types tout is that a typical person makes 5 errors an hour.

Oops this must be one of mine . ..:agree1:

I'm not saying that. I agree with your point and making a further point...sorry I thought that was clear.

I've used these statistics before to merely point out that 99.9% is not as small as we think it is especially when you consider the severity of the defect/event. OF COURSE I always use, teach, comment and SAY that the relative acceptability of any defect rate is related to it's severity.

But perhaps steel maiden should weigh in on what she intended by her parting use of the dropped baby example?

your point is well taken - although I don't think she was trying to go there...just making a point that 99.9% is not as good as most people believe it to be.

of course we must recognize that the severity of an event contributes to how acceptable any given occurence rate is.

exactly, and when your product is used in applications that could cause a death if failure occurs...........

It's needless and gratuitous hyperbole, and has nothing to do with the point, which should be making people aware of process capability, and ways of improving it.

If my training topic was process capability, you would be 100% correct. Since it was customer focus and meeting requirements, a figurative example can make people sit up and think about things they take for granted.

Babies are dropped by doctors. But, we work to minimize the impact of the errors.

Exactly, no babies were harmed in the making of this training program. And, we'd like to keep it that way. 99.9% is not acceptable in our industry.

There is a philosophy, that is attributed to INPO (Institute of Nuclear Power Operators) called Human Performance.
Steve makes a good point. There have been dozens of studies looking at human error rates. For example, for single actions it has been suggested "A good general figure for mechanical error rates appears to be about 0.5%.". For sequences of actions:
Procedural error in reading list of instructions. Per list. Dhillon [1986] 6.5%
Using a checklist incorrectly. Swain & Guttman [1983] 50%

So much for dropping 3.4 babies per million ;)

People make things tough for quality in service industries.

There is a philosophy, that is attributed to INPO (Institute of Nuclear Power Operators) called Human Performance. This philosophy does recognize and accept that humans do make errors.
I can't tell whether you're kidding or not. We need a "philosophy" from INPO to tell us that people make misteakes mistakes?

Babies are dropped by doctors. But, we work to minimize the impact of the errors.
Yes, but we do work to prevent the errors as well. I wasn't suggesting that prevention efforts should automatically stop at some point. Sometimes they get put on hold, waiting for technology to catch up, but there's still a load of money waiting for the guy with the better mousetrap.

For example, I assume that doctors limit the height above the surface they hold a baby (to limit the fall) and also the surfaces are likely padded.
I think most physicians, especially pediatricians, are aware that it's not a good idea to climb ladders while holding babies, if only for the reason that it represents two opportunities for misbegotten statistical citations--baby dropping and ladder accidents. Research also indicates that it's really dangerous to climb a ladder while holding an infant if there's an open fire below the ladder. So you'd better keep an eye on that punch press.

:topic: The issue of human performance is one of optimization. Where do you get a bigger bang for the buck - working to reduce the error or working to reduce the impact of the error.

Example - how many of you have put a 3 1/2 inch diskette into your computer upside down? Or plugged in your thumbdrive upside down? Both systems have been engineered to limit the impact of that error. Unless you really try to jam it in, there is no damage, you simply say to yourself oops and put it in the other way.

By the way, there is some relation of this discussion to poka-yoke or whatever you like to call error-proofing.

The issue of human performance is one of optimization.

This is exactly correct. But we need to remember that "optimization" is not synonymous with "perfection." When something has been optimized, it's working as well as possible, given the known constraints. When the baby-holding process has been optimized, some babies might still be dropped.

This is exactly correct. But we need to remember that "optimization" is not synonymous with "perfection." When something has been optimized, it's working as well as possible, given the known constraints. When the baby-holding process has been optimized, some babies might still be dropped.

Interesting discussion. I just don't think perfection is an achievable goal.

And - further developing the "dropped baby" scenario - how many times to you suppose mothers drop their babies? How many times to babies themselves impact things with their heads? What is the real impact to the baby? Do we have reports of deaths/injuries due to dropping the baby? Let's check our cause and effect.

I think this has applicability to the workplace - 1. We cannot achieve perfection. 2. We can reduce error rate 3. We can reduce the impact of the error. 4. And we do have to ask, what was the impact of the error in the first place?

Even the aircraft landing number can be discussed this way. Do pilots routinely make some sort of mistake during a landing? You betcha. But, they have safeguards to minimize the impact of the mistake. Co pilot, checklists, alarms, computers, ground control. Of course, then there was still the case of a British Airways 737 that lost an engine - and the crew turned off the good engine . . .

Interesting discussion. I just don't think perfection is an achievable goal.

Nor do I. :agree1: I fully agree with the idea of optimization of processes, and knowing with a reasonable degree of certainty what's going to happen in the future, given specified controls. Once that's known, it is what it is, unless the process is improved somehow. That was Deming's whole point, no?

Nor do I. :agree1: I fully agree with the idea of optimization of processes, and knowing with a reasonable degree of certainty what's going to happen in the future, given specified controls. Once that's known, it is what it is, unless the process is improved somehow. That was Deming's whole point, no?

True. Sounds good. The challenge (as Deming put it) was engineers can be uncanny with their predictions of what will happen, it is just that exactly when, where, and who it will happen to cannot be predicted.

True. Sounds good. The challenge (as Deming put it) was engineers can be uncanny with their predictions of what will happen, it is just that exactly when, where, and who it will happen to cannot be predicted.

You will find out, though. If you make some improvement to aircraft controls such that some number x fatal crashes will be prevented, you don't need to know that one of them will be United flight 1207 out of Chicago next March. You know when a plane does crash, that it wasn't one of the ones where the prevention thing worked.