0:01

You hear it all the time. The youth

0:04

will solve this. The youth are the solution.

0:06

They will be the change makers. Next

0:08

week, an unusual climate case is

0:10

going to trial in Montana. Led

0:13

by kids. That is a lot

0:15

of pressure when you are 14 or 16,

0:18

as I was getting involved in this case, to hear

0:20

that you are the ones that they are relying

0:22

on for the future. Can the state's

0:25

constitution be used to force action on climate

0:27

change? I'm John Quinnhill, and

0:29

that's next time on The Weeds.

0:31

New episodes drop every Wednesday.

1:06

About 100

1:06

years ago, Indian and

1:08

British archaeologists were on a research

1:10

expedition. They were traveling to

1:13

a site along the Indus River in what's

1:15

now eastern Pakistan. When

1:17

they got there, all they saw were just giant

1:19

mounds of mud. But they started

1:22

digging. So basically when they started excavating,

1:24

they found a city. Rajesh

1:26

Rao has spent the last decade writing about

1:28

the impact of this discovery. They found

1:31

urban planning that probably rivals modern standards

1:34

in some ways, because they found grid-like streets

1:36

with houses on either side. And then they

1:38

found amazing sanitation systems

1:41

in terms of, you know, houses having toilets

1:43

and sewage systems. Archaeologists

1:45

dated these systems

1:46

to 2500 BCE, and

1:49

they were way ahead of their time. There

1:51

wouldn't be extensive urban sanitation

1:54

systems like this for over a thousand

1:56

years. It would probably come down to the

1:58

more recent Roman and Greek.

1:59

civilizations, right? Even there, I think the kind

2:02

of sophistication I would say is not there.

2:05

Archaeologists named this city Mohenjo-daro,

2:08

or Mound of the Dead Men, and it

2:10

was a lot more than just an impressive sanitation

2:13

system. It was a full square

2:15

mile of immaculately constructed

2:17

roads lined with brick walls, multiple-story

2:20

homes, courtyards, even a public

2:23

bath. 4,500 years ago,

2:25

it was likely one of the biggest cities

2:27

in the world. And archaeologists

2:29

eventually found more huge cities

2:32

like this, and over a thousand settlements,

2:34

all thought to be built

2:35

by the same people who they called

2:37

the Indus Valley Civilization.

2:41

This society spanned at least 800,000 square

2:45

kilometers through modern-day India, Pakistan,

2:47

and Afghanistan, a size on

2:50

par with other major ancient empires

2:52

like Egypt and Mesopotamia. And

2:54

then in terms of the number of people, it's one of the biggest,

2:56

so about a million, across all the different

2:59

cities and settlements. And Indus

3:01

people seemed to have traveled even further.

3:03

They're probably the first examples of globalization,

3:06

right? So they were Indus merchants

3:08

who traveled in ships all the way to, it

3:10

appears, the Middle East.

3:12

But as archaeologists kept excavating

3:14

more and more sites, they started

3:16

to get kind of confused. Ancient

3:19

civilizations this big and this influential

3:22

pretty much always had huge, gaudy

3:24

markers of royalty, of hierarchy,

3:27

of conspicuous wealth. But amazingly,

3:29

there were no palaces, no huge

3:32

tombs, or any kind of monumental

3:34

architecture, which was the surprising part for many

3:36

people.

3:37

Archaeologists also didn't find

3:39

depictions of warfare or slavery,

3:42

which are pretty much always found in major

3:44

ancient societies like these. So

3:46

they started to wonder what these people

3:48

might have really been like. There have been speculations

3:51

about, you know, is this like an egalitarian

3:54

society? You know, is it some kind of a

3:56

society where people are more equal

3:59

than unequal?

4:01

We still don't know. But

4:04

with other ancient civilizations like Egypt

4:06

or Mesopotamia, we can answer

4:08

questions like these. Because they actually

4:11

told us. They wrote down their beliefs,

4:13

their stories, and then thousands

4:15

of years later, we're able to read them.

4:18

With the Indus people, though, it's a lot trickier.

4:21

On the one hand, archaeologists have

4:23

found some tantalizing clues.

4:26

These small carved shapes that look

4:29

like writing. So we found about 4,000 artifacts

4:32

with inscriptions. But to this day,

4:34

no one's been able to read them. And

4:37

it's really a pity that their

4:39

script is not the cipher, because we don't quite

4:41

get them, right, in terms of who they were.

4:45

I'm

4:45

Noam Hasenfeld, and this week on Unexplainable,

4:48

the Indus Valley Civilization was one of the largest,

4:51

most technologically advanced civilizations

4:53

in the ancient world. We

4:55

barely know anything about them. And

4:58

thousands of examples of what looks like writing

5:01

have just been sitting there, unread, for 4,500

5:03

years. So who

5:06

were they? And what could they

5:08

have been trying to say?

5:12

So Rajesh

5:17

first got hooked on the Indus mystery when

5:19

he was a kid.

5:23

I

5:27

remember reading about it in a middle

5:29

school textbook. And it was really

5:31

quite fascinating because it

5:33

helps me understand my roots, right, as somebody who was

5:36

born in India. Rajesh didn't

5:38

get right to deciphering, though. He

5:40

studied computer science, and he eventually

5:42

became a professor at the University of Washington.

5:45

That's actually still his day job, where he works on

5:47

AI and neuroscience. But

5:49

then about 15 years ago, he took a

5:51

well-earned break. I decided that

5:53

I needed something that was very different from what

5:56

I do, which is what we do for

5:58

sabbaticals, right, is to essentially

5:59

try to explore a question

6:02

that you haven't explored as part of your

6:04

sort of bread and butter research.

6:06

So he thought back to his middle school days. I

6:09

thought that this was a great opportunity to learn something

6:11

about this field I'm very fascinated

6:13

about and at the same time to ask

6:15

the question can we answer at least

6:18

some questions using the kinds

6:20

of computer algorithms we have.

6:22

Rajesh started his research by looking

6:25

at inscriptions that have been found across

6:27

the Indus Valley. Most of the inscriptions

6:29

are on these square seals so

6:31

these are about an inch, inch and a half

6:33

squared. And these seals kind of look

6:35

like rubber stamps but instead of being

6:37

etched into red rubber they're made

6:40

of a white or brownish stone which

6:42

could be stamped into wet clay.

6:44

And what you'd see are these small etched

6:46

symbols which are along the top of the seal. Some

6:49

of them look like you know what people are called a fish

6:51

symbol. Imagine sort of a vertical

6:54

Jesus fish with like fins

6:56

sticking out. There's also you know another

6:58

symbol that has whiskers you know for

7:00

like catfish have these little things that poke

7:03

out from the front of them. There are

7:05

symbols that look like little stick figures

7:07

that are jars that are interlocking rings.

7:10

When you see them all together these little symbols

7:12

look a whole lot like other ancient

7:15

forms of writing. At the same

7:17

time though the inscriptions on these seals

7:19

are super short.

7:21

These are only on average about

7:23

five symbols in Lent unlike

7:26

language-based inscriptions in

7:28

other parts of the world.

7:29

Writing is just usually way longer

7:31

than that mainly because you can't say

7:33

all that much with just five symbols. Which

7:36

has led some scholars to pump

7:38

the brakes on this whole Indus script thing.

7:40

They've

7:40

suggested that these fish, these stick

7:43

figures, these jars, the rings,

7:46

they might just be pictures. So

7:48

things like European coat

7:50

of arms you know traffic signs, totem

7:52

poles, right? You have many examples of non-linguistic

7:55

symbols. These kind of symbols do communicate

7:57

things but they can't come together to form.

7:59

complex abstract ideas like

8:02

a full language can. So before

8:04

trying to decipher the Indus script, Rajesh

8:07

had to figure out whether it was a language at

8:09

all.

8:12

He started by asking himself a pretty basic

8:14

question. How do you know if something

8:16

is writing or if it's just pictures?

8:19

There are no hard and fast rules here, but

8:22

Rajesh had a few clues he was looking for.

8:24

Qualities that tend to be found across all

8:26

kinds of written languages. So just to list

8:29

a few, first of all, you see linearity.

8:31

Written languages tend to be organized in a

8:34

clear, straight line, unlike, say,

8:36

a coat of arms, which could have symbols all over

8:38

the place. Secondly, we know

8:40

something about the directionality. Languages

8:43

tend to be written in a certain direction, so maybe

8:45

right to left or top to bottom. And

8:48

then finally, whether there is any kind

8:50

of structure that resembles the kind of structures

8:52

you find in a linguistic script. Take

8:55

English, for example. Whenever you see a

8:57

cue, you're usually going to see a U right

8:59

after it. That's a clue that English is a written

9:01

language and not just some random

9:04

grouping of symbols. And Rajesh

9:06

wanted to see if the Indus script checked these

9:08

three boxes of written language.

9:11

First up, linearity. In

9:13

the Indus script, you always find them very nicely,

9:15

linearly written as an order to the symbol

9:17

system. Next up, directionality.

9:20

That's one of the things that everybody agrees on is it

9:22

was mainly written from right to left. There's

9:25

a bunch of ways researchers have figured this out, but

9:27

my favorite is that they noticed symbols

9:29

would often end up getting smushed together

9:32

as they got closer to the left side. They would

9:34

try to squeeze in three symbols instead of just

9:36

one symbol, right? Because they started

9:38

out writing lies and spaced out on the right side,

9:41

but

9:41

then as they got towards the left side, they're like, oops,

9:43

I don't have space. So the

9:45

Indus script passes the first two tests. It's

9:47

written in a line. It's written from right to left.

9:50

But the third sign that something's a written language,

9:53

whether it has the kind of structure you'd

9:55

find in a linguistic script, that's

9:57

way harder to figure out, especially.

9:59

when you have no idea what the language

10:02

is actually saying.

10:05

So Rajesh started looking for patterns.

10:07

And this is where his computer science expertise came

10:09

in handy. So one thing we could do is we

10:12

can calculate something called entropy, which

10:14

is basically a measure of how much flexibility

10:17

there is. So if every symbol can be followed by any

10:19

other symbol, you have very high entropy. If

10:22

every symbol can be followed by only one symbol,

10:24

it's very rigid.

10:25

Rajesh knew that languages are in this sort

10:27

of sweet spot. Linguistic scripts

10:30

occupy a middle regime where it's

10:32

not too flexible, it's not too rigid, it's

10:34

somewhere in the middle. Too flexible and you could

10:36

get random jumbles of letters. Too

10:39

rigid and you wouldn't be able to express all

10:41

kinds of ideas.

10:43

So Rajesh took all 400 individual

10:46

symbols of the Indescript and he catalogued

10:48

them in a computer. Then he took the 4,000 stone

10:51

seals that archaeologists have found. So basically

10:53

all the short words or phrases made

10:56

up of these symbols and he catalogued those two.

10:58

And then we ran some statistical

11:01

analysis on it. So think of it as an AI model,

11:03

right? So we said, okay, learn the patterns in terms

11:06

of which symbol follows which other symbol.

11:09

So remember that English example? Q's

11:11

are almost always followed by U's. Pretty

11:14

rigid. But then those U's are usually

11:16

followed by vowels, like an A for

11:18

quake or an I for quilt. But

11:20

you're probably not going to find any kind of consonant

11:23

next. It's still a pattern, but it's

11:25

a lot more flexible. And Rajesh

11:27

and his team showed that the Indescript does

11:30

a similar kind of thing.

11:31

Our result was to show that the

11:33

Indescript was right among all the languages.

11:39

In the Indescript, after a diamond

11:42

symbol, you're probably going to find two parallel

11:44

lines. After that, there's a good

11:46

chance you'll see some kind of fish symbol,

11:49

but you're never going to see things like stick

11:51

figures or arrows. So

11:53

with this analysis, Rajesh and his team

11:56

showed that the Indescript does have

11:58

these three markers of language.

13:55

But

14:00

at the same time, even if an Indus

14:02

Rosetta Stone were discovered, it

14:04

might not be as easy to decode

14:06

as it was for hieroglyphs. A big reason

14:09

researchers were able to decipher the Rosetta Stone

14:12

is because they already knew so much about

14:14

Egypt.

14:15

For one thing, they already knew the names of famous

14:18

Egyptians. You know, they were able to essentially

14:20

map these very specially marked parts

14:23

of the hieroglyphic inscription to

14:25

names of rulers like Ptolemy and

14:28

Cleopatra.

14:29

Researchers could sound out these words in the

14:31

hieroglyphs, so they could figure out, okay,

14:33

this is a P sound, this is a T sound.

14:36

But with the Indus Valley, we don't have

14:38

this advantage. We don't know any proper names.

14:40

There's no historical names of

14:42

kings or rulers or people or gods.

14:45

And Egyptologists also knew a lot about

14:48

the spoken language.

14:49

Ancient Egyptian was still being spoken or

14:52

used in the Coptic Church.

14:54

Coptic isn't the same exact language

14:56

as spoken hieroglyphs, but it's similar.

14:58

So researchers had a general sense of

15:01

the grammar, the structure, and most

15:03

importantly, the sound, how to actually

15:05

pronounce the words. So when they were deciphering

15:08

hieroglyphs, they had a way to know if their

15:10

work sounded right, something we just

15:12

don't have with the Indus script. We

15:14

don't even know if the script was representing the

15:16

same language or multiple languages.

15:18

And there's one last advantage Egyptologists

15:21

had that Indus researchers just don't.

15:23

Finally, of course, Egyptian hieroglyphics, there were very, very long

15:26

inscriptions on walls and texts and

15:28

papyra and so on. Right. So there's there's lots of

15:30

texts that you could use to essentially

15:33

debug their decoding of the script.

15:36

To be clear, finding an Indus Rosetta Stone

15:38

would be a huge breakthrough.

15:40

But because the Indus Valley civilization is just

15:42

still so unknown, there's

15:45

no quick fix here. But

15:49

there is one potential clue. Scholars

15:52

do have a pretty good guess about what

15:54

kind of language they spoke. The current

15:56

consensus is that they were probably speaking

15:59

the Dravidian.

15:59

language, some kind of proto-Dravidian.

16:02

Dravidian is a family of languages that

16:04

are mostly found in South India, but

16:06

there's reason to believe these languages may have started

16:08

in the Indus Valley and then migrated south

16:11

later on.

16:12

And there are a couple spots where people still speak

16:14

Dravidian languages in Pakistan, right

16:16

near where the Indus Valley civilization thrived.

16:19

So if you really believe in the fact that it may be some

16:21

kind of proto-Dravidian or early form of Dravidian,

16:23

then you can start to guess the

16:25

meanings of particular words.

16:27

Rajesh says that scholars have used this kind of

16:29

guesswork to show that some of these seals might

16:32

actually be names. And he walked

16:34

me through how this works using one particular

16:36

seal. It

16:39

takes a few steps to get there, so just bear with

16:41

me for a sec.

16:42

So you can see that there's these seven

16:44

lines stacked in front of a

16:46

fish symbol.

16:47

Okay, so we've got seven lines and a

16:49

fish. No idea what it means. But

16:52

researchers started thinking about some ancient

16:54

Dravidian words for seven and fish.

16:57

One of the words for seven is Ilu, and

17:00

a word for fish is Meen.

17:02

But then Meen also happens to be another

17:04

word for stars. So it turns

17:06

out that Ilu-Meen is also a name

17:09

for the seven stars, right? Or the

17:11

Ursa Major.

17:12

So Ilu-Meen, this fish with

17:14

seven lines, it could be

17:17

referring to the Big Dipper.

17:19

But even more intriguingly, Ilu-Meen

17:22

could be a name. So it turns out that

17:24

Ilu-Meen is like a name that has been found

17:26

in old Tamil literature, right?

17:28

So the claim is that these are all names that

17:31

are there on these seals.

17:35

So yeah, this is kind of a roundabout way to decipher

17:37

things like this. And it's far from certain.

17:40

But if this seal and others like it really

17:42

are names, it could be huge. Because

17:45

once you know a name, you can look for it in

17:47

other texts, just like researchers did with

17:50

the Rosetta Stone.

17:51

Still, all of this is just a clue.

17:54

And clues like this are basically all we've

17:56

got at this point.

17:57

So when it comes to really understanding

17:59

the full language, we're kind

18:02

of at an impasse,

18:03

even for computer scientists like Rajesh. Unfortunately,

18:06

I think just throwing more computer power

18:08

at us is not going to, I think, change the

18:10

state of affairs. It's not a lack

18:13

of computer analysis or tools

18:15

or power because we don't have very

18:17

long texts, as I mentioned, and we don't have enough of them

18:20

either.

18:20

At this point, our best hope might not

18:22

be in computer science or looking

18:25

at related languages like with the Big Dipper

18:27

example.

18:28

It might be in archaeology, just going

18:30

out there and finding more texts, hopefully

18:33

longer texts. And

18:35

there's reason to believe that those texts might

18:37

really be out there because despite

18:40

the thousands of artifacts we've found,

18:42

we've barely scratched the surface of

18:44

the Indus Valley.

18:45

Only 10% of the archaeological sites have

18:48

been excavated. So the hope is, you know,

18:50

we may get lucky and if we start excavating

18:52

some of those other sites, including some very big sites,

18:54

perhaps we can find artifacts that have

18:57

longer inscriptions. But it's

18:59

also possible, like some of the naysayers are saying, they

19:01

may not even exist.

19:06

This

19:06

possibility is something we

19:08

need to consider. These short inscriptions

19:11

might just be all there is.

19:13

If we deciphered them, we might find

19:16

records of some names or maybe business

19:18

transactions. But would

19:20

that really tell us who they were? What

19:22

they believed? We might

19:25

never really know.

19:27

On the other hand, these kind of short, seemingly

19:30

boring inscriptions are often

19:32

just the start. Take ancient

19:34

Mesopotamia. So much of their earliest

19:36

writing we found is just short, simple

19:39

stuff, like names or records

19:41

of how much beer different workers got for their daily

19:43

rations. But

19:45

archaeologists ultimately found longer

19:47

texts, like Hammurabi's Code, things

19:49

that really did tell us what ancient

19:52

people valued, what they were thinking when

19:54

they laid down the foundation of our

19:56

modern legal system.

19:58

So given how large... and advanced

20:00

the Indus Valley Civilization was, just

20:03

think about what we might be able to find, what

20:05

we might be able to learn from deciphering their script.

20:08

We're basically not

20:10

doing ourselves any justice by not

20:12

knowing about these people, because basically it was

20:15

the largest civilization of its kind in the ancient world

20:17

at that time, but they're silent. And

20:20

hopefully, if archaeologists can uncover longer

20:22

texts, then we really start to learn about

20:25

who these people were, what were the value,

20:28

did they really have a egalitarian

20:30

society. I think we would learn a lot

20:32

about them in that case.

20:36

The Indus

20:36

people were one of the most significant ancient

20:39

civilizations, and they were essentially

20:41

forgotten until the modern era.

20:43

Imagine how differently we would think of the world

20:46

if we were only now reading Hammurabi's

20:48

code, or Confucius's Analects,

20:51

or the Socratic dialogues.

20:53

If we were only now tracing the formation

20:55

of ideas like math, democracy,

20:58

even monotheism,

21:00

these ancient societies shaped how we

21:02

understand the world. And without reckoning

21:05

with the Indus Valley civilization, a

21:07

fundamental building block of the modern world

21:10

is still essentially a black box.

21:13

So I think the Indus civilization has lessons

21:15

to teach us as modern humans, just

21:18

to say, hey, look, we did all these amazing

21:21

things, like practiced amazing sanitation,

21:23

build these great cities. We did not try

21:26

to glorify ourselves, and as far as we know, we haven't

21:28

found these huge palaces. So I think

21:30

there are some examples that they can set for

21:32

us. And giving them a voice through

21:35

decipherment of the script is really a worthwhile

21:37

endeavor.

21:51

This episode was produced by me, Noam Hasenfeld.

21:54

We had editing from Brian Resnick, mixing

21:56

and sound design from Christian Ayala, music

21:59

from me.

21:59

and fact-checking from Serena Solon. Manding

22:02

Nguyen is a lone wolf, Meredith Hodnot

22:05

is a rogue wave, and Bird Pinkerton

22:08

waited for her eyes to adjust to the sudden

22:10

darkness. Suddenly, a deep

22:12

voice rang out.

22:13

Bird Pinkerton, you

22:15

are the one we've been waiting for, the

22:18

one the prophecy foretold.

22:23

If you have thoughts about this episode

22:25

or ideas for the show, please email us.

22:27

We're at unexplainable at vox.com. And

22:30

we'd also love it if you left us a review or

22:32

a rating. Stuff like that really helps

22:34

new listeners find us. Unexplainable

22:37

is part of the Vox Media Podcast Network,

22:39

and we'll be back next week.