字幕 (368)
0:00There is this famous Google interview
question that everyone gets wrong.
0:04You're shrunk down to the size of a nickel
and put into a blender.
0:08The blades will start spinning in 60 seconds,
so what do you do?
0:13I would like to think I could duck down
and miss the blades
0:16Break the thing at the bottom maybe?
0:20Ask nicely for the blender
not to be turned on.
0:23Tie my clothes together, and then like
use it as a rope I guess?
0:27If I was lighter I could maybe catch a draft up?
0:29- Just accept defeat.
- I mean, I'm the size of a nickel
0:32what quality of life is that?
0:35I'm going to run to one wall,
push on it, run to the other wall. Push on it.
0:39Run to the other wall, so
I'm going to tip the container over.
0:41Honestly, the thing I would think about
0:43is trying to get to the very center
of the blades.
0:46It's spinning around me, but
0:47the actual RPM is probably not that high,
if I'm standing in the middle.
0:51I tie my clothes to one of the tips
of the blades as it’s starting up.
0:56Yeah, I get it to swing
me around and then I woooo!
0:59But these answers don't cut it.
1:03Now, I first heard about this problem
in this book.
1:06It describes how each year Google received
1:09about 3 million applications,
but they would only hire 7000 people.
1:14That's a 0.2% acceptance rate.
1:16So one way to screen out
millions of applicants was to use brainteasers,
1:21and the interviewers
would make these up for fun.
1:25We didn't get a list, at that time of
what questions to ask.
1:29We would share questions among each other.
1:31Some of them gained traction,
questions like:
1:33How many golf balls can fit in a 747?
1:37Or, how much should you charge
to wash all the windows in Seattle?
1:42But the blender question
really stuck with me, and I'm not alone.
1:47- Just lay back and enjoy that breeze.
The best model in the world is only going to
run maybe 10 or 11 hours,
1:54so we're getting out and when we do, we're better off for it because whatever doesn't kill you makes you stronger.
1:58- The question has been hotly debated
in Reddit comment sections.
2:06There are so many different answers,
but which one is the best?
2:12I try to hide underneath the blades,
I guess.
2:14Hide under the blade, probably
2:16Great first reaction,
2:18but maybe that doesn't solve your problem
entirely.
2:20Now you're just stuck inside of a spinning
blender, so maybe you want to escape.
2:23Can I climb the walls?
2:24Are there defects in the walls that are
sufficiently large for me to grip on to?
2:29Do I have Van der Waals forces that are
strong enough to connect me to the wall?
2:33Am I, like, essentially a tiny gecko?
2:36A gecko can stick
to the wall of a blender,
2:39even though neither its foot, nor
the glass are charged.
2:42The gecko's foot has to be pressed
firmly against the glass,
2:45so its atoms are within a few nanometers
of the glass atoms.
2:49Then, at any moment one atoms electrons
aren't uniformly spread about the nucleus,
2:54they might be slightly more on one
side than the other.
2:58This makes the atom momentarily
slightly positively charged
3:01on one side and slightly
negatively charged on the other.
3:05The glass atom next to it experiences
the pull of this charge imbalance,
3:09and so a similar charge
imbalance is induced on the glass atom,
3:14and therefore the electrons in the glass
atom are drawn
3:16to the nucleus of the gecko
atom, and vice versa.
3:20So there is a very weak attractive force
between neutral atoms.
3:24This is known as a Van der Waals force,
and it's what makes geckos stick to walls.
3:30It's the same force
that holds graphite together.
3:32There's no actual bonding between
the layers of graphene in a pencil.
3:36It's like a stack of paper.
3:38The layers only stick because of Van
der Waals forces.
3:42Now these forces are pretty weak.
3:44But since we're so small, maybe
they'd be enough to help us climb out.
3:49This I can almost certainly say,
we wouldn't be sticking at that scale,
3:52and that's because the Van der Waals
type interactions are still small.
3:57And I studied climbing, so that of these types of
scale, cockroach and gecko, you know,
4:02it turns out that you have to get
very special to do that.
4:05Geckos have millions of tiny
branches on their feet that increase
4:09their surface area
and allow them to mold to surfaces.
4:13Our hands aren't like that,
but ants and cockroaches
4:17don't rely on Van der Waals forces,
and they can still climb up walls.
4:22So maybe a miniature human could too?
4:25The mechanism of a cockroach foot,
4:27and I used to know all that
cockroach feet, is absolutely gorgeous,
4:31Same with an ant, by the way.
4:32There's two little claws, the tarsal claws
4:35those are things
that slap down on a surface and really do slap
4:38when climbing meters a second... slap and
engage, despite having no adhesion,
4:43they have very sophisticated,
frictional attachment.
4:47Those claws can grip almost anything,
even glass.
4:51While glass feels smooth to us,
4:53It's actually covered in tiny surface
imperfections.
4:56At insect scale,
these features are significant.
5:00- Ants basically have climbing gear.
- Oh yeah.
5:03They're like, using these little
like axes basically to pick their way in.
5:06We don't have the attachment disks
or whatever that would be,
5:10or like the special claws
or the Van der Waals forces.
5:14Well, we have claws, if you're that scale,
our fingers are claws.
5:16We have only really got... we have
two claws, really.
5:19And then our feet
aren't great at climbing, I don't know.
5:23Well, again, at that scale though,
I don't know, right?
5:26Imagine putting a little sharp, spike into your foot
5:29and sharpen your
shoes, wear high heel shoes.
5:33You'll be good to go.
5:35So now I'm climbing in heels.
5:38But there's still a problem.
5:39I mean, I'll have to be pretty careful
placing each hand and foot slowly.
5:44It's going to take longer
than 60 seconds to get out.
5:47And in that time,
the blades will have started spinning.
5:51One mistake and I'm a smoothie.
5:55So Google was looking
for a different answer.
5:58Now we’re going to the physics building.
Maybe they know?
6:01Yeah. I really got nothing. I’m stumped.
6:04This is so embarrassing.
6:05We’re in our last year of our degree, we should
definitely know this!
6:08I feel like I could probably swing running
around the sides and yeeting myself out.
6:13Okay, if we're just talking about the entropy,
6:14it should increase at some point.
So some sort of chaos should be...
6:17None of the system will stay un-disrupted...
6:20Take that as a limit to infinity
and I'll be chilling...
6:22Like using that logic
if I just like extrapolate...
6:25Now that's still too big for me
to Quantum Tunnel or anything like that.
6:29- Whoa whoa whoa whoa.
6:30I mean, that is really overthinking it.
6:33It's actually not that complicated.
6:36- Do you want to hear the best answer I've heard yet?
- Sure.
6:43-Does that work like that?
- Jump where?
- Out of the blender, just go up.
6:46- So whoever told you that is...
- ...crazy, right?
6:49Yeah, does that makes sense to you?
6:51No, it doesn't, but...
do you want to hear why that works?
6:54Yeah, tell me how it works.
6:55Jumping out of a blender seems impossible
because at nickel size,
6:59the wall of a blender is 15x
your height.
7:01It'd be like leaping over an eight story building.
7:05But watch these clips...
7:17A horse, a dog and a squirrel.
7:19They all jump to about the same height.
7:24This is exactly what Alfonso Borrelli,
the father of biomechanics,
7:27looked at in the 17th century.
As he put it, in the same conditions,
7:31smaller and lighter animals
make bigger jumps relative to their body.
7:35if the other conditions are equal,
and indeed the limbs and the other organs
7:40are in the same proportion,
the dog will jump as far as the horse.
7:44Now, sure, there is variation.
7:47A species whose survival depends on
jumping will be optimized for it,
7:51while others, like turtles and elephants,
they don't jump at all.
7:55But when you consider the huge variations
in size,
7:58I mean a horse is 1500 times
heavier than a squirrel.
8:02It's incredible that they jump to
around the same height.
8:06And it's not because squirrels
are super muscly or something.
8:10Horses and squirrels have similar muscle
to weight percentages,
8:14and insects have even less muscle
relative to their weight.
8:18Why do you think an ant can lift
50 times its own body weight?
8:23Like, is it any more muscular? No
you guys hit the gym.
8:25Come on. Like
you're more muscular than an ant.
8:28So how are small things so strong?
8:32Well if you look closely at a muscle. It’s made up of tiny units called sarcomeres.
8:37They act like miniature springs.
8:40How far a muscle compresses depends on
how many of these springs are in series.
8:45But the strength of a muscle depends
only on how many are working in parallel.
8:51The thicker the muscle, the more springs
in parallel, and the greater the strength.
8:57Therefore strength depends
on the cross-sectional area of a muscle.
9:01And as animals shrink,
this cross-sectional area
9:04scales down
with the square of their height.
9:07But an animal's weight
is proportional to its volume,
9:10so that scales
with the cube of their height.
9:14So as you scale down, weight decreases
faster than strength,
9:18and as a result, smaller animals have much
higher strength to weight ratios.
9:23I mean, you could probably lift,
your own weight,
9:25like if you were to put your own body
9:27weight on your back and squat that, you could now lift...
100 hundred times.
9:33And for us, stuck in that blender,
that extra strength
9:36relative to our weight
means we could jump right out.
9:40Your surface area
decreases with the square.
9:42You'd be like a little superman.
9:45- I see, okay!
- That's really cool.
9:47So I could jump, like,
literally out of a blender.
9:50You could jump out of a blender.
9:51But in movies and games
9:53where people are shrunk,
they almost never show it like that.
9:56Honey, I Shrunk the Kids It was one of my favorite movies
when I was a kid.
10:01Tiny people struggle picking up scissors.
10:03They almost get crushed by raindrops.
10:06If it was scientifically accurate, they’d actually be overpowered.
10:10Most people don't think of this
when they first hear the question.
10:14The answer almost seems too simple.
10:17When you ask the right questions,
you define the problem.
10:20There's some really obvious solutions
that work, and that's
10:24actually true for a lot of problems
in the real world too.
10:27Now I'm all for obvious solutions,
but from the start,
10:30the answer of jumping out
didn't sit right with me.
10:34Even this idea of like, I'm
going to jump out of the blender like that
10:37doesn't make sense to me,
because jumping is not just like,
10:40okay, how strong
you are relative to your weight.
10:42It's also timing
and your kinetics and all that.
10:45So like, how long can you be in
touch with the ground?
10:48How much can you apply that force in
one burst like over a really short period?
10:52Would it be
fair to say you're overthinking things?
10:55You got to suspend your disbelief
somewhere.
10:58I think if you like, factor in all the
potential challenges a human would have.
11:02Just like if they just all of a sudden
that size, they don't have
11:05like time to practice using their legs
and stuff in that new environment.
11:09Like, I don't give them very good
chances of jumping out.
11:11Sometimes there are people
who make everything more complex
11:14than it needs to be,
and that can be problematic.
11:17I would like to see like,
11:18you know, realistic modeling of,
we scale me down 100 times.
11:23Like, can I jump higher?
11:25I want to see someone do
those physics equations, yeah,
11:28you could jump higher, but you couldn’t jump
100x higher, you know?
11:32So that's why we got the researchers
at Georgia Tech's biomechanics lab to investigate.
11:37While, they're figuring that out,
11:39let me tell you about today's
sponsor, Incogni.
11:41Every time you browse the internet,
sign up for a newsletter, or even just buy
11:46Your personal data is collected,
it is stored.
11:50And sometimes sold
without your permission.
11:53So how do you escape?
11:56Well, with your permission, Incogni
contacts data brokers on your behalf
12:00and requests using proper legal language
12:03that they delete
your personal information.
12:05So instead of spending countless hours
trying to track these companies
12:08down yourself, Incogni
does all that hard work for you.
12:12I started using Incogni
and within just a week
12:15my inbox was less overrun
and my phone even got fewer spam calls.
12:19After a month,
my dashboard showed even more removals.
12:23Companies I had never
even interacted with, but who
12:26somehow had my personal information.
12:29If you want to try the easy way
to escape the clutches
12:32of data brokers, visit
incogni.com/veritasium.
12:35You can click that link down
in the description or scan this QR code.
12:39Make sure to use Code Veritasium
to get 60% off
12:42your annual subscription
to take control of your data today.
12:46That's incogni.com/veritasium.
12:49Now let's see how that
simulation is coming along.
12:52Okay, so we have our simulated blender.
12:55We’re 2 centimeters tall and we have to jump at least 30 centimeters to get out.
13:00I was like, well, what about me?
13:02Like, I'm pretty,
you know, embarrassingly non-athletic.
13:06What if I do this? So I did it right here
next to my desk.
13:09My partner sort of measured my jump height,
13:11and I know how much I weigh
and all that stuff.
13:13So what would it look like for me?
13:15If we have a person
that weighs 84kg, is squatting
13:1915cm and has a jump height of 27cm.
13:24That person, if they were scaled down
13:27to 1% of their original size,
13:30they would jump 42cm high.
13:33The simple simulation
shows a jump height of 42cm.
13:38So you would make it out.
13:40But we need to add in air resistance.
13:43Since our cross-sectional area is now
100 times larger relative to our weight,
13:48drag should have a greater effect
at nickel size.
13:51If it was 42cm jump height
before for the jumper, with drag...
13:56considering drag, then it's about 39cm.
14:01So we do decrease in jump height a little bit.
14:04But that drag calculation
is assuming you jump perfectly vertically.
14:08But what if you're a bit uncoordinated
and you flip onto your side mid jump?
14:13Well, then you're exposing
ten times the surface area
14:16and that increases
the amount of air resistance.
14:20Like if somehow you flipped,
and you're still moving up like this.
14:23Like what is the air resistance then?
14:28that means 22 centimeter jump height.
14:30Oh. So we don't we don't.
14:34If you start getting overconfident
and you wanted to do,
14:37like, a backflip while you're at it,
then you're going to mess it up. Yeah.
14:42Don't backflip out of the blender
is a good piece of advice.
14:46Don't try and show off.
14:47You're trying to not get chopped up.
14:49Just like, just go headfirst.
14:51It's not so much getting out
of the blender.
14:54It's what happens next.
14:56You've got two nickel sized men free
in the world. Think of the posibilities.
15:01The simulation came back,
15:02You can jump out of the freaking blender.
15:07I'm glad we went through this,
this exercise.
15:11Do you want me to do another month
of research on this?
15:13No. You know, like you've done it.
15:15You've done enough.
You've done enough. I'm convinced.
15:21I feel like jumping is
an unsatisfactory answer.
15:24It was unsatisfactory
when you mentioned it in the first place.
15:27And you went through
and you got the simulation,
15:29you got the model, and you're like, look,
you know, our little guy can jump 40cm.
15:33Are you convinced now?
15:34And I'm like, I guess.
15:37But like, my spidey sense was tingling.
15:41- Oh was it now!
- There is something going on...
15:44You're telling me
that I have to apply a force
15:49in 1/1000 of a second,
and I have to undergo
15:56I'm not going to survive that.
15:58So what I'm getting now is that, like,
my intuition was good.
16:03I think everyone's intuition was like,
you can't jump out of a blender.
16:07I think they're right.
16:10well, that's overthinking it,
but that's the whole point of the brain
16:13teaser is to overthink it is to get to
that point where you're thinking about it
16:18in the detail of like,
what would actually be feasible.
16:22A whole lot of things would go wrong.
16:24Our hearts have to generate
a certain amount of pressure
16:27to get the blood, you know, going up
to our head and going all the way down.
16:31If you take the human heart
and shrink it down,
16:33it's not going to be able
to generate the same kinds of forces.
16:36I think it would be a catastrophe,
in a smaller size. Controlling
16:39air pressure inside these countless sacks
inside of our lungs.
16:43There's an exquisite balance there.
16:45Now, you try to take that same design
and squeeze it down.
16:50I would be skeptical that you'd be able
to keep the passageways open.
16:54You wouldn't even be able
to think this through, because you just
16:57wouldn't have the brain structures that we have.
17:00You can't fit 86 billion neurons in a nickel sized volume.
17:04You can't scale cells down either.
17:06That's the thing. Like cells are cells.
17:09I mean, jumping out would be, to me,
seems like your only option, but I don't
17:12think you're going to be able to jump
17:14because you can't breathe
and your heart can't pump blood,
17:18so you just keel over and die
before you can make your jump.
17:23Okay, so if you're a biologist,
you think we die.
17:26If you're a physicist, you can decide
whether we'd be little supermen or,
17:30as I believe, incapable
of fully harnessing our extra strength.
17:34What did Borelli know?
17:35He didn't even have blenders.
17:38He doesn't know the stress.
17:39But if you're an interviewer at Google,
you might not even care what the answer is.
17:45I think one of the misconceptions
that candidates have is
17:49when I'm asked this question, it's
because they want to see
17:53if I can solve this problem.
17:55That's actually not quite right.
17:56There are five attributes
people are looking for.
17:58There's addressing ambiguity,
18:01There's breaking down the problem,
being creative, being smart,
18:05and then communication.
18:06- So I guess like none of those five are
whether it's correct.
- Right.
18:12We’re the idiots who went
and tried to figure out what's the best
18:15of those answers.
Um, yes!
18:20Google realized that asking these types of questions
didn't make much sense.
18:25Laszlo Bock, the senior vice president
of people operations at Google, said this:
18:30On the hiring side, we found that brain
teasers are a complete waste of time.
18:34How many golf balls
can you fit into an airplane?
18:37How many gas stations are in Manhattan?
18:39A complete waste of time.
They don't predict anything.
18:42They serve primarily
to make the interviewer feel smart.
18:45But I just feel like there's that moment
18:47where you're like,
so are you going to admit you're wrong?
18:49And I'm like, nyah, you know, I think
this is further to like, I'm not wrong.
18:55This question is crazy.
18:57And I think it goes to your very point.
18:59Your very point,
which is that like brain teasers like this
19:02are not good ways to assess whether people
know what they're talking about.
19:07So although brain teasers aren't useful
to assess
19:10job applicants,
they are useful for something.
19:13I mean, every time we ask this question
to people on the street,
19:17to physics students and to scientists,
they lit up.
19:21They had to try to see the world
from a new perspective.
19:24And it's exactly this way of thinking
that has led to
19:27some of the biggest scientific
discoveries.
19:29Einstein used thought experiments
to come up with his theory of relativity.
19:34Euler's solution
to the bridges of Königsberg puzzle
19:37is what inspired graph theory.
19:39And when Schrödinger wanted to illustrate
his problems with quantum mechanics,
19:43he imagined a cat in a torture box.
19:47The blender question is admittedly silly,
but silly
19:50questions can yield profound answers
and show us new things.
19:55I think in order to learn something new,
you have to be willing
19:58to embrace the ridiculous
and just go with it.
