Hitting the Target: Accuracy vs. Precision

Rob Bell:
Hello, and welcome to Sketchplanations The Podcast, fortnightly forays into facts, frameworks and phenomena, as found within the deep archive of explanatory sketches at sketchplanations.com.

I'm engineer and broadcaster, Rob Bell, and joining me once again is designer and creator of Sketchplanations, Jono Hey, and entrepreneur and past winner of The Apprentice, Tom Pellereau.

Tom Pellereau:
All right, lads.

Hello, hello.

That's very impressive how your diction is so clear, Robert.

Rob Bell:
Focus, Tommy, focus.

Which is good because this time we're talking about accuracy and precision and the fact that they're not the same thing.

So in your little write up under the sketch for this one, Jono, you make the point that some things are much more easily explained in a sketch than with words.

So, Jono, do you want to start by explaining what the sketch shows?

And then listeners, please do refer to the sketch as we talk about it.

You can see it as the artwork for this episode, and I'll include a link to see it at sketchplanations.com as well.

Jono, how does the sketch show the difference between accuracy and precision?

Jono Hey:
I think it shows it quite well.

Rob Bell:
I agree.

Jono Hey:
This sketch is essentially, and I'm not the first person to come up with this, but it's essentially four targets, like the sort of targets you might have in archery practice.

The ones that are sat on the ground 20 meters away, there's four of them, and they're basically showing each of the different situations between whether you're accurate and whether you're precise.

So, if I start on the left-hand side, you've got a situation where you're shooting at the target and you're kind of all over the map.

You're not hit in the center and one is in the bottom left and one is in the top right.

Rob Bell:
Some haven't even hit the target.

Jono Hey:
Some haven't, yeah, you're right.

Yeah, some of them are in the back.

There you're not accurate nor are you precise.

The next one over is one which is a situation which is accurate but is not precise.

And in this case, you've been shooting at the target and basically all of the shots are clustered around the middle, but they're not all hit in the middle.

Your average location of your arrows are basically in the center.

Rob Bell:
But there's quite a bit of distance between each of the hits, right?

Jono Hey:
There's a bit of spread around each of them.

Yeah.

So every time you do it, you're slightly different, but you're basically hitting towards the center.

Rob Bell:
Accurate towards the middle but not very precise with where all those arrows are going.

Jono Hey:
Exactly.

And so the next one is a situation where you can see that there's a bunch of shots and they are all clustered really precisely but they are in fact not on the center of the target.

Rob Bell:
I'd be quite proud of that one.

Jono Hey:
Absolutely.

I mean, yeah, because actually precision is very difficult for us to do.

So this is a situation where you were very precise.

You were basically doing exactly the same thing every time.

Rob Bell:
Hitting the same bit at the target every time.

Jono Hey:
Yeah.

But you were miscalibrated and you weren't actually managing to aim towards the center.

Yeah.

So in that case, you weren't accurate, but you were precise.

And then the last one is, boom, you hit the center and you basically hit the center every time.

Rob Bell:
Olympic gold medal won.

Jono Hey:
Atomic clock.

Rob Bell:
Atomic clock.

Lovely.

Jono Hey:
Yeah.

Rob Bell:
Lovely.

Jono Hey:
Exactly.

If you hit the center and then every shot hits exactly in the center, then you're accurate and precise.

I learned about this distinction from a book by Simon Winchester called Exactly, which was all about accuracy and precision and actually how improvements in these led to the modern world and how we make products and the industrial revolution.

I hear this nice way of saying it, which was, accuracy is true to intention and precision is true to itself.

Every time, if you're being precise, every time you do it, you're getting exactly the same result.

But accuracy, you might not be getting the same result every time, but your intent is spot on.

Rob Bell:
That's clear.

Jono Hey:
It can be a bit hard to get your head round.

I can give a bit more context.

But anyway, that's the idea with the picture, with these four different targets, if you can imagine them, and then there's four different scenarios of accurate and precise.

Tom Pellereau:
So for example, my dad throwing the ball at the wall above the window, would you say that it was his precision that let him down?

Jono Hey:
Right.

So he was in the garden and he chucked the ball at the house aiming for the brick on the house.

Rob Bell:
The brick just above the glass window.

Tom Pellereau:
So he's not throwing the ball straight through the glass.

Jono Hey:
He can do that and he did that, but he did it again and it got a different result and it went straight through the window.

Rob Bell:
I would say, Tommy, with all respect to your dad, that is neither accurate nor precise because his intention was to hit the brick above the wall, which he did once, but then he didn't do that the second time.

So there's no accuracy.

But for it to be precise, it would have to hit the same place or very near the same place, which he didn't do either.

Tom Pellereau:
Okay.

Rob Bell:
Jono, do you concur on that?

Jono Hey:
I mean, I think there's a chance that if he was to repeatedly chuck the ball at the house and most of the time hit the brick just above the window, then he could be accurate.

Tom Pellereau:
But in this case, Yeah, he backed both his accuracy and his precision and unfortunately, he didn't have either.

Jono Hey:
Problem with the intention as well.

Rob Bell:
The outcome was really unfortunate.

Tom Pellereau:
Well, for him, it was very funny for me.

Rob Bell:
So I think Jono, in science and engineering, is that where Simon Winchester talks about in his book exactly?

Is he mainly applying it to science and engineering?

Jono Hey:
Well, should I give you a little bit more of the context?

Because it does bear on it.

Rob Bell:
Yeah, go on.

Jono Hey:
So the full book title of the book is exactly how precision engineers created the modern world.

And I think a good example, if I think about it, is imagine you're a shoemaker a long time ago, and you had a shoe shop, and you were good at making shoes, and you made shoes by hand.

And perhaps you got really good at your craft, and you managed to make shoes which were just right for the customer.

But then, every now and then, somebody comes in and you make a shoe and you don't get it quite right, but sometimes you do.

Or then imagine you get somebody else in your shop who's training up and they come and get a shoe.

And so from the customer, the shoe that they come out will be very different because it's made by a different person and to different standards and slightly different ways.

Everything was done with crafts, and crafts depends very much on the person.

And then what happened in around the Industrial Revolution, so in the 1800s, in particular, he talks about this chap called Henry Maudsley.

And he together with building on a number of other inventions, basically he made something called the Screwcutty Lave, which you can go see in the Science Museum, the original one.

And what that enabled him to do-

Rob Bell:
The Science Museum in London.

Jono Hey:
The Science Museum in London, yeah, sorry.

And so was to make screw threads precisely, basically exactly the same every time.

And so before that, you had screws that were cut by hand from iron, which sounds kind of mad.

Obviously, they had tools that they were doing it, but they were basically, you know, somebody's pushing this thing and you're, and you're trying to cut these screw threads.

And so everyone is slightly different.

And so you build your machine.

But if you build another machine out of these same components, they're all going to be slightly different.

And what the sort of innovation he had there was when you were able to produce things, which were exactly the same every time.

So very, very precise.

So not just like a master craftsman gets it right most of the time, but they were precise every time.

It led to being able to do interchangeable parts.

You know, you can have these set of screws, you can have nuts, you can have bolts and all this stuff.

And when all of these are interchangeable, that was essentially the foundation for things like mass production.

Because you can use, I can make these here, and they can work in your machine over there.

And before that, when everybody's making things by hand, and there's all these variations, that wasn't possible.

And he actually invented also, I don't know if you remember using it, when you do an engineering, but there's a thing called a micrometer, which is a very small gauge.

You hold it in your hand a bit like a wrench, and you turn this little screw thread, and it squeezes something and tells you how wide it is.

And then there's often, there's a secondary screw thread, which is incredibly precise.

And he invented that 200 years ago, and was able to show to incredible degrees of accuracy, that had never been possible before, that all these things were the same, and really precise.

And so these inventions in precision, being able to do the same thing exactly again and again, were really critical in being able to make products, as we know them, out of metal that you could reproduce at scale.

Tom Pellereau:
We live in an incredible precision world, and I don't think many of us really realise it at all.

Rob Bell:
No, you're right.

And when we talk about that, the history of it, Jono, as you did there, I think it makes you realise that everything now is expected to be precise and accurate.

Jono Hey:
Well, certainly, if the products we use, and if you're buying parts for a phone that you're making, you expect it to come exactly as spec, and you can do things, you have incredible tolerances and things now.

I was thinking, as an example, though, which I think is different, obviously, he was talking about engineering and making products, an example I was trying to think of, which is perhaps easier to understand, would be food.

If I was to say my accurate meal would be one that was delicious, exactly this is a great meal.

If I went to the local sandwich shop, their quality is probably all over the map.

Sometimes it's good, sometimes it's bad, most of the time it's not great.

There was a little local Italian near where we worked, where every now and then, I had some of the best pasta I've almost ever had, and most of the time, it was good, but every now and then, it was brilliant.

I feel like their intention was like, they knew how to make good pasta, but they just didn't do it every time.

But sometimes they did and it was amazing.

Now, if you go to the not accurate but precise, would be something like McDonald's, right?

So like, it is not gonna be the best meal you ever had, but if you buy a McDonald's wherever, it's basically gonna be exactly the same as wherever else you got it.

So, they're incredible experts at precision.

And I think it's only at like, you know, the really top restaurants where you're basically guaranteed an amazing meal every time because there's all that attention to precision.

Rob Bell:
Yes, yes, yes.

Jono Hey:
It's a different field, but like food is different.

I was thinking if I'm making a curry or something and I've got whatever's in the fridge, it's gonna be different every time.

Sometimes it's brilliant, sometimes it's not, you know.

Rob Bell:
What I've really enjoyed about this is thinking outside the box from away from science and engineering, Jono, as well.

And with food as well, is there something about if you follow a recipe exactly, you're being very precise because you could just you could do that again.

You know, it's 300 grams of this, it's 100 milliliters of that.

So you're just doing that.

You can be really precise with that over and again.

The intention is for it to be a really tasty meal.

And so let's say it came down to the amount of salt and pepper and seasoning.

And the recipe told you to exactly this much and exactly that much precision.

Actually, if you were just to season it to taste, that gives you where you want to be, right?

But you might not be able to repeat that consistently every single time.

But it's moving more towards the accuracy.

Is there something there as well?

Because I enjoy, I've enjoyed really thinking about it in these kinds of terms around the repeatability, the consistency versus the how close you are to the intent of what you're trying to achieve.

Jono Hey:
Yeah, they're all like subtle variations, aren't they?

So that, you know, I think precision is given the same circumstances, you end up with the same output.

And I was saying like a recipe is a set of instructions to get close to a great meal, but like it's still going to be different because your ingredients are going to be different.

You know, your onion is going to be slightly bigger or your tomatoes are going to be in season or not in season or you're missing this bit from your fridge.

And so I sometimes think even though you might follow a recipe, you know, I cook this bit at a slightly higher heat than before for 10 minutes instead of 8 minutes, it's going to be different.

And so unless you get really precise, which is exactly the sort of thing that someone like McDonald's is amazing at, it's still going to be different.

And so as you say, like you're going to season it and it's probably going to come out good season to taste, but it is different.

You're responding to a different situation.

It's like somebody moves the target each time.

Rob Bell:
Yes, exactly.

Jono Hey:
Yeah, you're adapting to the conditions.

Rob Bell:
And another example I quite liked within kind of science, the engineering stuff, is like calibration of tools.

And a great, easy example that I think a lot of people will have done is calibration of, say, weighing scales.

So you could have very precise weighing scales, but they could be consistently out by 200 grams each time.

So they're precise because every time it weighs exactly what it is, but it's 200 grams out.

So your accuracy is out, but your precision is high.

But you can calibrate that tool to make it accurate again, right?

You calibrate your weighing scales to bring it down.

But if you've got shoddy weighing scales where you've got a load of imprecision in there as well, so you're not even getting consistent things, you can't calibrate that back in to be accurate as well.

You might have accuracy-ish, but your imprecision is all over the shop.

Jono Hey:
I think the scales are a great example.

We've got some scales and they just sort of wobble about.

I can't rely on them.

Add small amounts.

Rob Bell:
I saw a little thing on that thing.

And this comes back to the targets and the arrows and or darts, whatever, in your sketch, Jono.

Calibration can fix your aim, but it doesn't steady your hand when you're firing the darts or throwing the darts.

Jono Hey:
Tommy?

Tom Pellereau:
My work relies on creating designs for unique products and then repeating those over and over again.

In most of my products, we sell 10, 20, 100, 200,000 pieces of them.

And the consumer is expecting that every single one is the same.

So I'm so fortunate in the fact that it is now possible, because of all these amazing people in the past who've developed things, that I can make products that are identical.

So for example, my LED face masks, which have been hugely popular, they are completely reliant on having exactly the right wavelength in the LED light that's coming out.

The LEDs that are actually inside all the masks.

Now they are one of the few things which they make thousands of those.

And then there's a machine that moves so fast, which tests the wavelength of every single one.

And then underneath it, it has pots of, depending on how precise or how near to the wavelength we want it is.

And if it's within like plus or minus two, it goes in this pot, plus or minus ten, this pot, et cetera, et cetera.

Because it is still an incredibly difficult art to make all the LEDs identical to, you know, we're talking to one thousandth of a nanometer in terms of what their output is.

Which is incredible precision, but even within that precision, we have to choose a small subset of that.

Rob Bell:
Because you have tolerances, right?

Tom Pellereau:
Yeah.

Rob Bell:
And anything outside of the tolerances that you've deemed acceptable doesn't get used.

Tom Pellereau:
Exactly, or doesn't get used by my product, might get used by a cheaper one, or might get used for something else.

Rob Bell:
So the size of your tolerance is talking about your precision, right?

Tom Pellereau:
Yes.

And the amazing thing is that the huge body of evidence has shown that 633, which is the wavelength of the red that's used in virtual masks, is really important.

And if it's 630, it's not as good.

So my business is hugely dependent on precision and accuracy, and my customers depend on that.

Jono Hey:
Can I give another sort of business example, but not quite so physical?

Rob Bell:
Why please do?

Jono Hey:
We spent a huge amount of time at the companies I've worked at on the interview process and recruitment generally.

One way to think of your interview process is, do you get good people at the end?

Do you get the people that you want at your company who thrive and do well?

And then every time you run your interview process, is it essentially doing the same thing so that you are going to find those people at the end who are going to do well in your company?

And so if you're recruiting over and over again over the years, how good is your recruitment process at finding the right people for your company?

Is this your accuracy?

And then how reliable is it?

If you run it every time, if somebody else is involved in the process, if it's a slightly different department, let's say, are you still going to find out if you run it this year versus next year?

And I think those are quite interesting when you just think of processes generally.

Are you getting what you intend?

And do you keep getting what you intend every time you run it?

Rob Bell:
That's good.

Jono Hey:
And so I think if you think about it in that sort of way, a lot of things can be related to this idea of accuracy and precision.

I think a very classic one, and there's a related sketch, a very similar drawing, but these two ideas of reliability and validity, which I came across in the education school.

And a classic one there is if you do an IQ test, are you actually testing someone's intelligence?

And then if you were to do multiple IQ tests, would you get the same IQ at the end of it?

Would you get the same result?

They're called reliability and validity in those examples.

Are you measuring what you intended to measure?

And if you, given all the circumstances are the same, somebody's doing this again, will they get the same result?

And actually, so often in a lot of social processes, we get different things.

And obviously, where I learned about accuracy and precision here was about engineering.

We essentially resolved that and can now make things that are both accurate and very precise.

But in a lot of these social contexts, we really struggle both with the accuracy and with the precision.

I guess it's not always clear that your interview process is going to get the right people in, nor is it clear that every time you do it, you're going to get the same people coming out the other end.

Interesting concepts to think about.

Rob Bell:
I think the non-physical examples get quite interesting in this, Jono.

So I just watched the Oscar-winning documentary called Mr.

Nobody vs.

Putin.

Okay.

It's about a school teacher who just filmed everything basically within the school in Russia.

What I wanted to talk about where I think this applies is that there's one teacher in there who, it feels to me like he grew up during the Soviet period and so he has been indoctrinated with Soviet mantra.

When the war in Ukraine started, Putin enforced on schools across Russia that those Soviet mantras come back and teachers were forced, they had scripts to read to their pupils, the pupils had scripts that they had to respond back to.

So it's all very, all very indoctrinated if you like.

And there was this one teacher who was just really in that mindset.

And so I was thinking about one can be very consistent in their beliefs or something on what they believe to be correct.

And who am I to say, who am I to judge this?

But you watch it and you feel, well, that is that is, you've just been indoctrinated to say all this stuff.

So you can be consistent in what your beliefs are.

But in actual fact, you're wrong.

And that could be factually wrong.

But because of whatever you've been told or whatever you've been told, you are consistent about your beliefs in it, but your accuracy.

So your precision is high, but your accuracy is shot because it's wrong.

Jono Hey:
It could be factually wrong.

Education is a really interesting space for this.

You put people through a course, you've got different teachers.

Do they all learn the same thing?

As you end up, if you try and increase the precision around it, like how repeatable it is, then you end up teaching to the test.

Or everybody's learning exactly the same material, which may or may not be right for that set of students you've got in front of you.

I think you want the creativity of the teacher to be able to adapt to the students in front of you.

But the more you try and standardize it, more precise and repeatable you make it, who knows if you're getting the right outcomes at the end of it.

Rob Bell:
When you start to dig into this, it gets quite interesting.

It's an interesting thought exercise, I thought, to apply this into lots of different areas.

Another one I thought about is, and I've had this recently about buying a house, and about gut feelings.

So you can have a really strong gut feeling about something.

By definition, I think that would be accurate.

But your gut feeling is an accurate representation of what you're feeling.

But it's not necessarily precise because you might find it difficult to explain the detail of what you're feeling.

But you know if it's accurate because I'm feeling this in my gut.

Does that make sense?

Jono Hey:
Yeah, that makes sense to me.

I think it goes quite to the, almost like the McDonald's example.

But there was another sketch which has come up before, which is this idea from Roger Martin called the Knowledge Funnel.

And he has this thing that things, particularly he's talking about business, but go from being a mystery to being a heuristic, to being an algorithm.

And so the mystery is like, we don't know why this happens, but sometimes we get it right.

And the heuristic is like, okay, we've got some rules.

Let's say that we're going to follow roughly these rules and it gets a bit more reliable.

And only like in food do people like McDonald's get it down to an algorithm.

But at the beginning, you might have that intuition.

And you're like, I think it's this, but I couldn't say why it's this.

Rob Bell:
Yes.

Yeah, yeah.

Jono Hey:
And the job there, if it's a business and you're trying to give people a great experience every time or make the product the same every time, is to figure out what it is and turn it into something that's repeatable.

Rob Bell:
And precise.

Yeah.

Jono Hey:
Yeah.

Which is an interesting challenge.

Tom Pellereau:
There's potentially an interesting aspect also of culture too, in the fact that, for example, bread making used to be kind of very inaccurate, very imprecise to say.

And then during our younger years, certainly, and it's still today, if you want an identical loaf of bread, you can go into any shop and get the exact same King's Mill, for example, in the UK, right?

But we're kind of culturally kind of going more towards there.

But I quite like going to the bakery on a Saturday and getting an artisan loaf from one of the many bakeries now.

Then you kind of end up spending like almost a tenner on a loaf of bread, whereas you could go and get one that is absolutely perfect from the co-op or the test goes around the corner for sort of a pound ten.

We've become so precise in some areas that I think potentially culturally we're almost like, but I don't want that precision anymore.

I like it when it's a bit random.

Rob Bell:
That's a nice observation, Tommy.

So kind of harking back to what Jono was talking about early and the industrial revolution bringing in mechanisation of processes in order to have accuracy and precision.

There's now a real demand to go back to the crafts, craftsmanship of individuals.

Tom Pellereau:
Yeah, and you see that in beer and food.

And a part of me feels like people having long beards is kind of a harking back to that.

Because when you have a beard, it's a bit more unpredictable, a bit more that, you know, when everyone's clean shaven, it's sort of a bit more precise, so to speak, maybe.

Rob Bell:
I love it.

Tom Pellereau:
Culturally, we go around in circles, maybe.

Jono Hey:
Well, you did know that there is a sketch called The Beard Cycle, which is quite ridiculous, but it was more about fashion as much as anything.

Rob Bell:
It plays out, though, in society, Jono.

Explain The Beard Cycle.

You've mentioned it.

Jono Hey:
No, yeah, only that, you know, after a while, if everybody's got beards, you know, somebody who's clean shaven looks different and interesting.

Rob Bell:
Yeah.

Jono Hey:
And then if everybody's clean shaven, the guys with the big beards are more interesting and different.

Tom Pellereau:
Yeah.

Jono Hey:
That's the thing about your craft loaves, Tom, I think it's quite interesting.

Like fruit and vegetables is an interesting one, because obviously, you know, nature is a bit different every time.

But you can definitely walk into some supermarkets now.

And, you know, carrots are all like within one or two centimeters and they're all straight and they're all exactly the same color.

And, you know, in a way, the joy of a Nobly Nali or something, yeah, it's like everything's a bit different, you know, and it looks a bit sort of weird and suspicious that all the apples are exactly the same size and color and without blemish and everything like that.

And we don't necessarily like it every time, actually.

It feels weird.

It is weird.

Rob Bell:
Isn't it funny, though, how now society has deemed it acceptable for us to pay more?

Tom Pellereau:
Yeah, isn't that clever?

Jono Hey:
You could sell some products to them.

They're slightly different wavelengths.

They don't quite fit.

It's a bit weird.

Charge a bit more for them.

Rob Bell:
What else would anyone like to add to the precision and accuracy discussion before we round off?

I'll go with one.

What would you rather have?

A doctor who's precise or accurate, or a surgeon who's precise or accurate?

Tom Pellereau:
Depends what they're doing.

Jono Hey:
Yeah.

Most of the time, you want both.

The book says it's accurate and precision is what you need.

That is the innovation.

Rob Bell:
That's the epitome of needing accuracy and precision, isn't it?

Jono Hey:
Surgery is an interesting example.

You'd have an amazing surgeon who could maybe do amazing with poor tools.

How do you make it so that everybody gets the right surgery and done to a high standard every time?

That's difficult.

You have to be accurate and precise.

Rob Bell:
What about a car?

Jono Hey:
Yeah, you probably want accurate and precise.

I wrote in the sketch True to Intention and True to Itself, which I find it intriguing, but it's sometimes a bit hard to get my head around.

Another way to think about it, which I quite like is accuracy is doing the right thing, and precision is doing it right.

Tom Pellereau:
And robots, that I've seen this, can be incredibly precise at repeatedly doing the same thing, but it can be way off accurate, hilariously.

Rob Bell:
AI, I think, is probably...

I get bored of talking about AI, but I think it's important that we bring it up because it's always here.

I feel like AI could be similar to that, Tommy, where I don't know, is it precise?

Tom Pellereau:
I think it can be pretty good at both.

Jono Hey:
Well, I actually think one of the intriguing things about AI is that it's not always precise.

You can ask the same question and get different answers, which is kind of fascinating because we're not used to computers doing that at all.

You expect that computer is going to give you, you put in ones and zeros in one end, it's going to come exactly the same.

Our electronics and our software has been like that since day dot.

And so it's so strange, isn't it, that you can say, tell me again and it will give you something different.

Rob Bell:
But it could also not be accurate as we know.

Jono Hey:
Yeah.

Interesting.

To give a really topical example, I was thinking of if you're shooting missiles, like one of them is actually aiming towards the right target and the other one is hitting the target where you're intending.

And there's a case recently where they hit the spot where they were aiming extremely precisely, but it wasn't the right place to be targeting in the first place.

There's a very literal, accurate and precision examples, but we've got extremely precise for the weapon, but you still need to do things accurately.

Rob Bell:
Yeah.

Good.

Well, listen, I make that precisely the time that we should stop, but whether that's accurate or not is anyone's guess.

Go well, stay well.

Goodbye.

Jono Hey:
Till next time.

Thank you.

Rob Bell:
All music on this podcast series is provided by the very talented Franc Cinelli.

And you can find many more tracks at franccinelli.com.