WEBVTT

00:00:00.000 --> 00:00:33.341
I've talked about the need to find the root cause of a problem and not just to solve the symptoms.
It's important, by the way, to solve the symptom; you don't want that to stay around.
But we also need to get rid of the underlying cause,
or else it'll come back again – the symptoms will repeat.
So, how do you do that?
Well, the Japanese developed a technique on, actually, the Toyota assembly lines.

00:00:33.341 --> 00:01:03.864
They call it the 5 whys.
So, basically, when something goes wrong on the assembly line
– so, suppose that your job is to fasten the water pump on the engine as the engine comes by you,
and they discover later down the line that the water pump isn't fastened properly.
So, what can you do?
Well, in many countries – like the United States – what you do is you refasten it

00:01:03.864 --> 00:01:33.100
so it's fixed. Well, that's fixing the symptom, not the cause.
So, what the Japanese say is, "Let's ask,
'Why was this not fastened properly?'"
So, they'll ask why, and they'll discover the place where it was assembled,
so the place on the assembly line where the engine comes by and the person sees the engine
coming by and picks up the water pump and fastens it on.
OK, that person didn't do it right, but that isn't enough;

00:01:33.100 --> 00:02:00.938
you have to say "why" – "Why didn't that person do it right?"
And then, if you examine that even further,
you might discover that maybe there's a shortage of parts, so maybe there's a confusion
between what this pump requires and what that pump requires.
And usually, on these assembly lines, there are a variety of
pumps and a variety of engines and you have to make sure you match them right.
But you might need a different kind of tool for each one, and you might have picked up the wrong tool.

00:02:00.938 --> 00:02:30.075
So, why would *that* be?
So, we have to ask, "Well, why is there this confusion?"
and "Why are the tools not right?" and "Why this?" and "Why that?"
And, if you finally go on the bottom, you might want to discover that
way back earlier in the process, we should change things.
Maybe we should paint the water pumps different colors
to match the color of the engine, or maybe we should have tags on them,
or maybe we should do this, that or the other.

00:02:30.075 --> 00:03:01.201
There may be solutions to get at the root cause
and that therefore stop the problem from ever occurring once again.
Now, the five whys sounds good.
And you can do that yourself – you can always say "why" when somebody tells you something;
you say, "Well, why is that?"
And they'll tell you why that is, and
you know you've sort of reached the root cause when they say, "I don't know,"
because, actually, you can never really get at the very, very bottom.
So, what we need to do is get at the *lowest level*

00:03:01.201 --> 00:03:37.192
that we can actually do something about.
But that isn't enough!
There are many people who'll criticize the five whys,
saying, "Well, yeah, it may work on the Toyota production line
where you're making automobiles,
but in many cases
it's the wrong approach because what it does is
assumes that for every issue that happens there's a single cause
and "If I could only get down to the single cause, I can eliminate the problem."
Well, in the complex world, that isn't always the case. So, let me talk about aviation safety.

00:03:37.192 --> 00:04:02.731
In the United States and probably in every country when there's a major airplane crash
– and, in fact, in the United States when there's any airplane incident that's a commercial airplane,
then we have a special panel – it's called the National Transportation Safety Board.
And as soon as they hear of an airplane incident,
those – the people who work there – they will have their bags always packed

00:04:02.731 --> 00:04:31.738
and there's a team that's always ready to go. And so, they're told immediately that there
is an incident, and they run for the airport
and they fly to the incident to examine the problem.
Now, their examination can take a *year* or two
because they have to look in great detail about all the different things that possibly could have gone wrong
to cause a loss of lives.
When they issue their report, which is often a year or two after the event,

00:04:31.738 --> 00:05:04.473
they don't say "This was caused by pilot error."
or "This was caused by a faulty altimeter." or "This was caused by a 'boomph!'"
What they usually do is they say,
"There were *many* things happening.
And here's a list of things. And if any one of these had not happened,
there would not have been an accident."
So, if there are, say, eight things that led to the accident,
any one of which could have prevented the accident,
how do you know which is a root cause? All eight are the root cause.

00:05:04.473 --> 00:05:30.663
So, you have to therefore look at the whole system and find out
what the entire underlying set of *possible root causes* is
and work on them.
Now, the public doesn't like that.
So, quite often, when there's an airplane crash,
the newspaper reporters rush out and the people say, "What caused it?
What's the result? Why is it? What can we do?"
And usually, there's something like the pilots made a mistake,

00:05:30.663 --> 00:06:01.208
the pilots erred or
a part failed or
some simple cause.
And, no.
That satisfies people because people really like a simple cause and effect.
Something happened. What *caused* it?
As opposed to the *system* caused it.
And it takes a year or two for people to find out what really went on
and the system components that were involved.
By then, the public is no longer interested.
And so, the report is sort of – yeah, it's a great report,

00:06:01.208 --> 00:06:34.485
and it's read by people like me and it's read by the pilots
and it's read by people in the aviation industry,
but the general public doesn't read it;
they don't know that this has been a wonderful, detailed investigation of the accident,
that here are the many different things that led to the eventual accident.
But that's – if you want to do things well,
you have to not only find the underlying cause, but recognize it as part of a system,
and so there seldom is a single underlying cause – there are most usually *multiple* ones.

00:06:34.485 --> 00:07:00.093
But you're not going to get solutions unless you attack all the multiple ones.
And let me tell you that in aviation safety
this has been extremely effective. It has taken a long time
– it's taken decades – but today we can go for two or three years without a *single* fatality
in aviation accidents in the commercial sector.
*No accidents.*

00:07:00.093 --> 00:07:30.947
People who are afraid to fly and yet there's been no accidents for three years.
And when you take an automobile instead of flying, the accident rate is quite high.
A million people a year die in automobile accidents across the world.
And we can have *zero* people die from aviation accidents across the world.
Now, the problem is that what happens when there's a big aviation accident,
a hundred people may die.
And that's most unfortunate, but it's a hundred people;

00:07:30.947 --> 00:07:46.800
it's not a million people, which is what dies in the automobile accidents.
But you have to understand and appreciate this;
you have to understand *root causes*
and the whole problem of a *systematic systems approach* to issues.