0:00

Support for Recode

0:02

Media comes from Oracle. AI

0:04

could be one of the most important computer

0:06

technologies of our time. But

0:09

AI needs a lot of processing speed and that can

0:11

get expensive fast. Now

0:13

you can upgrade to the next generation of the cloud. That's

0:16

Oracle Cloud Infrastructure, OCI

0:18

for short. OCI is a single

0:21

platform for your infrastructure, database,

0:23

application development, and AI needs.

0:26

See how customers like Uber and Cohere are

0:28

training their AI models at twice the speed

0:31

and less than half the cost. Take

0:33

a free test drive of

0:34

OCI at oracle.com. That's oracle.com.

0:44

This is Recode Media with Peter Kovtun and

0:46

that's me. And today we do not have a

0:48

guest on our show. Instead, we have

0:50

a lot of guests. That's because I'm sharing

0:53

the first part of something cool that

0:55

I got to make. It's called the Twitter Fantasy.

0:58

It's a four-part series about Twitter's origins,

1:00

its development, its future. This

1:02

is part of our Land of the Giants series. We've been making

1:04

these since 2019. I have done one on Netflix

1:07

and Apple. There's a lot of other good ones as well. It

1:09

is one of my favorite things I have done in my career,

1:12

full stop. The idea is to tell you

1:14

about tech companies and industries that are reshaping

1:16

the world. But we're not just

1:18

doing a business history that focuses on CEOs

1:20

and entrepreneurs. And that's what we're

1:22

doing with this season. We are definitely talking to

1:25

and about Twitter's leaders. There's a

1:27

lot of that in this first episode. But

1:29

we're also talking to the users that made the platform

1:32

and people who can describe Twitter's impact

1:34

on culture and politics around

1:36

the world. And yes, there is

1:38

going to be some Elon in here, but not

1:40

too much Elon. Anyway, if you listen

1:42

to Recode Media, you may have already listened to

1:44

this episode. Apologies, we will get

1:47

you fresh stuff next week, I promise. And

1:49

if you listen to Recode Media and you haven't listened to

1:52

this show, I really think you'll

1:54

like it.

2:01

Jason Goldman was supposed to be an astrophysicist.

2:03

Instead, he ended up dropping out of grad

2:05

school and working at internet companies. I'd

2:08

been extremely online before that was even the thing.

2:11

I was like a nerd from way back in

2:13

the early 90s. Goldman got a job

2:15

at Google and eventually handed up at a

2:17

dream gig for a young tech worker in

2:19

San Francisco. He had a product

2:22

at Twitter. One problem,

2:24

though. Twitter barely

2:26

worked at all during this time. If

2:29

you used Twitter back then, you will remember the

2:31

Fail Whale. This is a cartoon

2:33

that showed a flock of birds carrying a whale. And

2:36

you got it when you opened the Twitter homepage if the

2:38

platform was down. Which happened

2:40

a lot of the time. Most

2:43

of the time, Twitter went down unexpectedly. Every

2:46

so often, it went down on purpose, like

2:48

in June 2009. I wrote

2:50

a blog post saying, hey, we're taking

2:52

a downtime. We'll be down from 2 to 3

2:55

a.m. on Saturday morning, West Coast time,

2:58

in order to do this critical database upgrade,

3:00

which was critical. We're telling the world

3:02

we need to take a break for

3:04

a couple hours to fix the thing because

3:07

we're failing all the time. Yeah, because we're

3:09

failing all the time. Lots of people on

3:11

Twitter would complain whenever the service went down. But this

3:14

time, the Twitter team got an unusual request

3:17

from the State Department. We'd like you not

3:19

to do this if you can. One understands

3:21

important, but there's this thing

3:23

going on and this is really bad timing for it. Could

3:26

you do it another time? The reason

3:28

that the U.S. State Department wanted Twitter to

3:30

stay up was that Iran had just gone through a

3:32

presidential election. And there were

3:34

very believable accusations of fraud and

3:36

protesters had taken to the streets. And

3:39

people in Barack Obama's administration thought

3:41

protesters were using Twitter. They

3:44

felt like going down would

3:47

negatively affect the protesters' ability to

3:49

organize. When

3:51

Twitter got that request from the State Department,

3:54

it was not a tech giant. It

3:56

was a three-year-old's accidental company,

3:59

staffed on the phone. almost entirely by inexperienced

4:02

young men. But when the Obama

4:04

administration looked at Twitter, it didn't see

4:06

a bumbling startup. The State Department

4:08

saw Twitter as a bastion of free speech,

4:11

a place where journalists and dissidents and regular

4:13

people shared information from all over the

4:15

world. This may be hard

4:17

to remember, but back in 2009, it

4:20

was totally normal for serious, important

4:23

people to talk about Twitter like this, all

4:25

kinds of people. Here's

4:27

a 2009 quote from Mark Fiefely, a

4:30

former official at the National Security

4:32

Council.

4:34

Quote,

4:34

without Twitter, the people of Iran would

4:37

not have felt empowered and confident to

4:39

stand up for freedom and democracy, end

4:41

quote. Twitter, Fiefely

4:44

argued, should win the Nobel Peace Prize. The

4:47

Obama administration's request worried

4:49

Goldman and Biz Stone, one of Twitter's

4:52

co-founders. If they did something the State

4:54

Department asked them to do, did that mean they

4:56

were taking sides in an Iranian election?

4:58

We didn't want to pretend that we knew what was

5:01

going on, particularly with respect

5:03

to geopolitical affairs. Biz and I had

5:05

a real talk about this at the time,

5:07

like a real like, can you believe this is

5:09

what we're doing right now? Like, can you believe it? Like,

5:13

we're not that smart. Goldman

5:15

and Stone did what Washington asked and Twitter

5:17

stayed up. And over in Iran, well,

5:20

that's the thing. This role was

5:22

minimal and it didn't, it was impromptu. It

5:24

was mostly word of mouth that organizers

5:27

use in order to bring people in the streets. Baba

5:30

Krahimi is a professor of communication, culture

5:32

and religion at the University of California,

5:35

San Diego. In 2009,

5:37

he went to Iran to study online political life.

5:40

And when the protest kicked off, he ended up doing lots

5:42

of TV interviews from Tehran. He

5:44

found that TV networks thought what the State Department

5:46

thought.

5:47

The Twitter was driving events inside the

5:49

country. Protesters

5:52

aren't only sending video and photos,

5:54

but live updates from the demonstrations

5:57

using a website called Twitter.

5:59

I remember one Fox News reporter

6:02

asked me, you know, what's your view of, you

6:05

know, Twitter in Iran? And at that time,

6:07

I actually didn't know much about Twitter because I

6:09

haven't really used it at that time. I

6:12

didn't see any Iranians use it. I didn't encounter

6:14

it much. I kept talking about Facebook,

6:16

but Twitter was somehow pushed in our conversations.

6:20

And much of it had to do with, again, this

6:22

basic question, how Iranians

6:25

are using internet or Twitter.

6:28

That was the big question everyone wanted to know.

6:30

To be clear, people outside Iran

6:32

were certainly using Twitter. They were trying to focus

6:34

attention on what was happening inside Iran.

6:37

The State Department, the media, Iranian

6:39

expats, they all thought Twitter was central

6:42

to protests in Iran because they were

6:44

on Twitter. And that's where they learned about the protests.

6:47

But to protesters on the ground, Twitter

6:49

was nowhere near as important as it looked from

6:52

the outside. I

6:54

remember this period really well. I

6:56

was spending a lot of time reporting on Twitter, the company

6:58

back then, and I was using Twitter

7:01

a ton. And I was one of those people that

7:03

felt like Twitter was important. But

7:06

also, I couldn't tell you exactly why that

7:08

was. Turns out the folks at the State

7:10

Department were in the same boat. And

7:15

to me, that is the story of Twitter, one

7:18

we've been playing and replaying for more than a decade.

7:21

We know it's a big deal, but we don't really know

7:23

how big a deal it is or why.

7:26

And we keep getting it wrong, all of us. Even

7:29

now, it can still command our attention

7:31

in inexplicable ways. Twitter

7:33

has become kind of the de facto town square.

7:37

It's important to the function of democracy.

7:39

It's important to the function of the United

7:41

States as a free country and many other

7:43

countries and to help freedom in the

7:46

world. A year ago,

7:48

in 2022, Elon

7:50

Musk became the latest and most significant

7:52

person to fundamentally misunderstand Twitter.

7:55

He bought it for $44 billion, and

7:58

we all know what happened next.

7:59

I'm Peter Cofkin.

8:10

And you're listening to

8:12

Land of the Giants, the Twitter fantasy.

8:22

It's pretty reasonable to say we are at the end

8:24

of Twitter as we know it. For

8:26

one thing it's now called X, which virtually

8:28

no one calls it, so I won't either. More

8:31

important, Twitter in some form will probably

8:33

still linger on for a while.

8:35

But despite what you've heard, Twitter's not dead

8:38

or dying because of Elon Musk.

8:41

I'd argue the platform had dug its own grave long

8:43

before he showed up to throw dirt on it. Long,

8:46

long before. You can go all the way back to

8:48

the very beginning. Twitter

8:50

began life as an accident. It

8:52

started as a side project of a side project,

8:55

and that afterthought quality has stuck around all

8:57

of its life. The choices people

9:00

in and around the company made in its earliest days,

9:02

and in some cases choices they didn't make, helped

9:05

permanently shape the company's trajectory. It's

9:08

a course that ends with the world's richest man buying

9:10

it and instantly regretting it.

9:12

Here's the quick backstory. Ev

9:15

Williams, a young, idealistic Nebraska

9:17

native, moved to San Francisco in

9:20

the 1990s to join the first web phone.

9:23

Williams co-founded a company to build workplace

9:25

productivity software, but the

9:27

only product they shipped was a side project called

9:30

Blogger. Blogger let anyone

9:32

with a keyboard and internet connection publish

9:34

their thoughts on a blog.

9:36

And Google bought Blogger

9:38

in 2003. Williams

9:40

went on to co-found another startup called Odeo,

9:42

and that was going to do for audio what Blogger

9:45

did for writing. But in 2006,

9:48

Apple made it easy to move podcasts

9:50

onto an iPod, which is exactly

9:53

what Odeo was trying to do. So there

9:55

was no reason for Odeo to exist. What

10:00

are we doing here? That's

10:03

Blaine Cook, one of the engineers who was working

10:05

at Odeo. Since Odeo now needed

10:07

to pivot, it organized a hack day. That's

10:10

a day for their developers to drop everything, dream

10:13

up a new idea and build a rough version

10:15

of it. One of those developers

10:17

was Jack Dorsey. He had an

10:19

idea for something he called status. He

10:22

borrowed the idea from instant messenger apps like

10:24

AIM. Instant messaging clients

10:27

had this idea of status. So you could say,

10:29

like, I'm available or I'm offline, but then you could

10:31

put in some other

10:33

information about what you were up to. Like you might

10:35

say, you know, making dinner or something like that. And

10:38

so Jack's idea was really just take

10:40

that course, status

10:42

update thing and then make it

10:44

something that persists that you can actually share.

10:46

Shortly after that status got

10:48

renamed Twitter. Word for the

10:50

sounds. Birds make. And that prototype

10:53

of Twitter, the MVP product that

10:55

gets built at a day, more or less. What

10:58

is the reaction internally? Oh, my God, this is

11:00

it. This is a gazillion dollar company. Or

11:02

is it other interesting? I think

11:04

more on the oh, that's interesting. Part

11:07

of the reason Twitter wasn't an obvious winner. It

11:09

worked using SMS text messages. And

11:12

at the time, in order to send the text from most

11:14

phones, you needed to tap out the letters

11:17

using the number pad. Tap a key once

11:19

for a twice for B, three times

11:21

for C. This was deeply annoying.

11:24

And it was not something most people ever bothered

11:26

with. Sending and receiving text

11:28

messages also cost money, which

11:31

meant that early on, Twitter lost money

11:33

on every tweet. This is important.

11:35

We're going to come back to it later. Some

11:41

people got Twitter right away. Here's

11:43

Jason Goldman. Ev signs

11:45

me up for Twitter over SMS,

11:48

and we are having dinner on our

11:50

way to go on a trip together, to go

11:52

to Vegas together. And so all my first

11:55

tweets are from this trip that we

11:57

take together to Vegas. And

11:59

it's like me, like. I'm gonna go play poker

12:01

and I'm saying like, you know, he's hanging out by the pool and

12:03

like I'm like Oh like this is like this ambient

12:06

awareness of like what my friends are doing I

12:08

don't have to like text with them or make a plan I just

12:10

know what their day is like and I

12:12

just like I don't know it just like immediately made sense to

12:15

me what this Was gonna be Remember this

12:17

is before the iPhone and before the millions

12:19

of ways our phones let us track and communicate

12:21

with each other Facebook existed

12:24

but that was where you posted pictures and the party

12:26

went to last weekend Twitter gave

12:28

you present tense Accentable Intel

12:30

about what your friends were doing now Williams

12:34

hired Goldman to help decide what Odeo's

12:36

parent company should do next my

12:39

first decision or like the first email I send

12:41

basically as director of product strategy is we

12:43

shouldn't work on anything besides Twitter Odeo

12:46

spun off Twitter as its own company Williams

12:49

didn't want to run it day to day. So the founders

12:51

held a meeting to decide who would be CEO

12:54

It's principally Jack's idea and

12:57

so Jack becomes the CEO and Basically,

13:00

everyone else ends up going to work there Jack

13:02

Dorsey had never run a company He

13:04

had zero management experience at

13:07

the time. He was considering quitting tech and

13:09

becoming a fashion designer Jack

13:11

was always a bit of a sci-fi John

13:13

Borthwick is an investor who funded several apps

13:16

that Twitter ended up buying So he got

13:18

to observe the early Twitter team and

13:20

Dorsey at close range Just

13:23

very hard to read Just hard

13:25

to know exactly where's that and

13:28

the hard to know how much he's actually

13:30

thinking Dorsey

13:33

loved nightclubs and one of his primary use cases

13:35

for Twitter was to send his friends updates about

13:37

his plans from a noisy bar So

13:40

it made sense that Twitter's big coming out party

13:42

happened in 2007 at South by Southwest That's

13:45

the music and culture festival that takes place in Austin,

13:48

Texas every spring There's a tech

13:50

part too. In fact, then it had a real spring

13:52

break vibe for a certain kind of tech person

13:56

Who's going who what kind of nerds

13:58

are going to Austin in? March. Some

14:01

real nerds, man. Some real nerds. It's

14:03

like people who like writing on the internet, sort

14:05

of in that like kind of early internet culture

14:08

space. And then at the end

14:10

of the tech part, you start seeing all of

14:12

the cool people in leather jackets with

14:14

guitars and like, you know, musical instruments

14:16

show up. You're like, all right, the actually cool people are here

14:18

now. It's time for us to go home. But

14:20

while the tech nerds swarmed Austin, the Twitter

14:23

founders had a captive audience. They

14:25

set up a big screen in the lobby with a live feed

14:27

of tweets about the conference, and they gave

14:29

attendees instructions about how to sign up. And

14:32

the tech guys at South by use Twitter

14:34

pretty much as intended. They

14:37

told other tech guys about where to find free

14:39

drinks. The whole structure

14:41

of South by is the yes, there's

14:43

the panels, but it's really about

14:45

like the social scene. And

14:48

you would see this emergent behavior

14:51

where people would be in the bar,

14:53

the scene would be kind of dying down. All

14:55

of a sudden, you'd see a bunch of tweets that like something else

14:57

was popping off down on the street. And you

14:59

would just everyone I was like, Oh, we're all moving

15:01

over. I saw the tweets and things going on. And everyone move

15:03

over. It was just like this idea that you

15:06

could see people moving because

15:08

of the tweets that they read. And

15:10

it was just like this dominant back

15:12

channel for the

15:15

conference itself. Twitter

15:17

ended up being the big startup coming

15:19

out of South by that year. They even

15:21

won an award for best new blog.

15:24

Jack Dorsey gave the acceptance speech, and

15:27

he was very on brand. I would like

15:29

to thank everyone and 140 characters less. And

15:32

I just did. Thank you. Thank you.

15:35

So the founders South by proved that Twitter

15:38

could make a crowd of people move like

15:40

a single all knowing organism. And

15:42

that was one early idea of what Twitter was

15:45

going to be. But it wasn't the only

15:47

vision of what Twitter could be. Around

15:49

this time, there were lots of competing ideas even inside

15:52

the company. For Ev Williams,

15:54

who'd founded blogger a decade earlier, and would

15:57

go on to found medium, Twitter

15:59

was another place that would make it easy for

16:01

ordinary people to say what's on their mind. Here's

16:05

Williams in an interview from 2010.

16:07

The way we described it on the homepage was a way

16:09

to keep up with friends, family, co-workers by

16:12

answering the question, what are you doing?

16:14

Another version of what Twitter could be was not

16:16

a platform at all, but a protocol

16:18

like email. No one owns email,

16:21

but it's one of the basic languages of the internet.

16:23

It allows people in different parts of the world using

16:25

software and devices made by different companies

16:28

to talk to each other. Here's Jason

16:30

Goldman again. Jack really

16:32

believed, I think, in this protocol notion of

16:35

the product and this public good

16:37

notion of the product, like describing Twitter as like the

16:40

public town square or like this protocol layer

16:42

of the internet. But protocols don't

16:44

usually make tons of money. And

16:46

at that moment, right after South by Southwest,

16:49

Twitter started looking like it might make some

16:52

people very rich. Yahoo

16:54

offered $12 million for the company. The

16:57

founders turned the offer down. They thought

16:59

Twitter was worth $100 million at least. Venture

17:02

capitalists agreed. Twitter

17:05

started raising round after round of VC

17:07

money starting in 2007. They

17:09

needed that money to add servers, to hire

17:11

people, to pay for those SMS

17:14

fees. But taking that money

17:16

meant the dream of an open protocol was

17:19

dead. Once you've agreed to

17:21

become a VC company, the VCs

17:23

have a very specific contractual

17:26

structural idea about

17:29

like how this is going to play out, which

17:31

is they're investing to get an outsized

17:34

return 10x or more in

17:36

the next seven plus years, or

17:39

it goes to zero. Like those are the outcomes they're

17:41

optimizing for. It's like, like, we're not really

17:43

interested in a double here. Like we're not really interested

17:46

in like we put in $5 million and

17:48

it comes back $10

17:48

million. How seriously

17:51

were those discussions happening?

17:53

That maybe it could be something else? It wasn't

17:55

happening at a sufficient level of detail to obviously

17:57

avoid future pain. I

18:00

think this is a really critical kind of hallucination

18:04

on all of our parts. Like, we all kind

18:06

of saw what we wanted from this thing. Twitter

18:09

was a little bit like Jack

18:11

in that you could look into its eyes and see

18:13

whatever you wanted to see. That's

18:15

investor John Borthwick again. Some

18:18

people would look at it and say, oh god, this

18:20

is like short form email. Others

18:22

would look at it and say, this is the future

18:25

of communications. Others would look at it

18:27

and say, this is the public square. Twitter

18:30

didn't actually state what they were.

18:32

And so they could be many things

18:35

to many different people. For

18:38

a while, the Twitter board looked into Jack

18:40

Dorsey's eyes and saw a visionary

18:42

co-founder that could run a visionary

18:44

company. According to hatching Twitter,

18:47

Nick Bilton's definitive history of the company's early

18:49

days, Dorsey hadn't run anything before

18:51

and it turned out he wasn't good at running Twitter back

18:53

then. So, Twitter's board replaced him

18:56

with Ev Williams, his co-founder, in 2008.

18:59

This was the beginning of a carousel of CEOs

19:01

at the company. A

19:04

lot of Twitter's early adopters lived in San Francisco

19:06

or Silicon Valley and they worked in tech. Right

19:09

behind them were journalists like Jay Wortham, who

19:11

now writes for the New York Times magazine and hosts the

19:13

podcast Still Processing. Back

19:16

then, Wortham worked at Wired.

19:18

I was a research fact-checking intern, berm,

19:20

berm, berm. And

19:23

I think the early Twitter offices,

19:25

we could see from the Wired offices,

19:27

I think it was all in downtown

19:29

San Francisco. For Wortham,

19:31

like a lot of early users, Twitter

19:33

served as a kind of surrogate social life.

19:36

Now I'm a little bit more of an extrovert, but when

19:38

I was younger, I was such an introvert. It was a much

19:40

easier way for me to talk to people, to

19:42

share about myself, to sort of learn about the world.

19:45

The thing that Twitter replicated for

19:47

me that really hooked me

19:50

was the feeling of being in an AOL chat

19:52

room and just always seeing

19:54

what people were talking about. And it's the most

19:57

intoxicating thing and they know that.

19:59

Twitter offered a chance to join the conversation

20:02

and that distinguished Twitter from other

20:04

social platforms.

20:06

Most people were not able

20:08

to share their thoughts in

20:10

real time in any meaningful way, right? And

20:13

if you're posting on Facebook, it's probably to

20:15

a very small limited group of people that

20:17

you've agreed to be, quote, friends with. Twitter,

20:20

on the other hand, I mean, it's just a free for

20:22

all. So if you're someone who has

20:24

any opinion or any thoughts

20:26

about anything, all of a sudden

20:29

you have this currency that

20:31

you didn't have anywhere else and nobody could

20:33

keep you from that. It was a way

20:35

that you could be heard and

20:37

seen that we've

20:40

never really experienced before.

20:42

Before long, actual bona fide

20:44

celebrities started showing up on the site

20:46

and jumping into chats with normies.

20:48

And I think that was also part of the early

20:51

intoxication, right, like it was removing

20:54

all these barriers between people that normally

20:56

existed and that was the buy-in. I

20:58

remember

20:59

there was a rock guy named

21:01

Sebastian Bach, the lead singer

21:04

from Skid Row just joined a conversation

21:06

I was having once. He might even dunk down

21:08

me. I'm like, this is amazing. I don't even care

21:10

about Skid Row or Sebastian Bach, but he's

21:12

famous. He's now participating in a conversation

21:15

that I'm having. That's wild. It's wild,

21:17

yeah, it's wild. And I do

21:20

think in the beginning, people

21:21

talked a lot about Twitter as like, this

21:23

cocktail party you never really knew who you were gonna

21:25

meet. Soon some of those real life

21:27

celebrities were making pilgrimages to the Twitter

21:30

offices in San Francisco. Arnold

21:32

Schwarzenegger came by, Sean Combs,

21:34

well that's Puff Daddy to me and maybe to

21:36

you, showed up and asked to become their

21:39

chief marketing officer. If

21:41

there was one moment you could point to when Twitter

21:43

as celebrity magnet really got going, it

21:46

was Ashton versus CNN. In

21:48

April 2009, the actor Ashton

21:50

Kutcher challenged CNN to

21:52

a race to be the first to count on Twitter with one

21:55

million followers. And the winner

21:58

got to say they were the winner. CNN

22:00

reported on the race as if it were real news

22:03

and it became a viral story. Ashton

22:06

won by the way, he live streamed the moment

22:08

he hit 1 million followers.

22:11

The whole freaking world is watching.

22:13

Come on people, we can do this.

22:21

I think the fact that it was CNN versus

22:23

Ashton, it was a perfect illustration

22:25

of the flattening of the graph that

22:27

Twitter allowed. Like it was a place where you

22:30

could become famous or talk to famous people.

22:32

What's it like in the company when that

22:34

is happening? And it's, you're only a couple

22:37

of years out from being, this is for nerds

22:39

to tell the other nerds where they're having a margarita.

22:42

Right. It would have been a lot more enjoyable

22:44

if I was not one of the primary people

22:47

on the hook for the thing to work.

22:50

And it fundamentally did not work

22:52

most of the time. Like the product

22:55

just didn't work. Like

22:57

it was just technically broken for

22:59

the better part of two years. Coming

23:02

up it's Twitter versus the fail whale.

23:05

But even when the service was up and running, it

23:07

still had an existential problem.

23:16

Support for Recode Media comes from Oracle.

23:20

AI is storming every industry.

23:22

Billions of dollars are being invested in it. Buckle

23:25

up. It might be one of the most important

23:27

computer technologies of our time. We've

23:29

covered this in previous episodes. The problem

23:32

though is that AI needs a lot of speed and processing

23:34

power. So how do you compete without

23:37

cost spiraling out of control? You

23:39

could upgrade to the next generation of the cloud.

23:42

Oracle Cloud Infrastructure.

23:45

That's OCI. OCI is a single

23:47

platform for your infrastructure database, application

23:50

development and AI needs.

23:53

OCI has four to eight times the bandwidth

23:55

of other clouds. It offers one consistent price.

23:58

OCI is a single. for

24:00

your infrastructure, database, application

24:03

development, and AI needs. OCI

24:05

has four to eight times the bandwidth of other

24:08

clouds. It offers one consistent price

24:10

instead of variable regional pricing. Of

24:12

course, nobody does data better than Oracle. Oracle

24:15

does data. So now you can train

24:17

your AI models at twice the speed and

24:19

less than half the cost of other clouds. If

24:22

you want to do more and spend less like

24:24

Uber and Cohere, take a free

24:26

drive of OCI at oracle.com

24:29

slash Peter. That's oracle.com

24:32

slash Peter. Oracle.com

24:34

slash Peter.

24:39

Having

24:43

celebrities like Ashton Kutcher and Oprah

24:45

Winfrey sign up for Twitter was great publicity,

24:48

but it put a ton of stress on the company's tech

24:50

infrastructure. Infrastructure that was

24:53

never designed to handle tens of millions

24:55

of users. Just because

24:57

of the nature of the way it was built and then how we

24:59

tried to scale it, it was kind of spaghetti.

25:02

Dick Costolo came on as Twitter's chief

25:04

operating officer in 2009. He found a platform

25:08

that was constantly being rebuilt. Twitter's

25:11

code would spring a leak and the Twitter people

25:13

would patch it quickly but temporarily.

25:15

I mean, there was a service in

25:17

the architecture called the Ashton Kutcher

25:19

service, which had been built when Ashton

25:22

was getting so many followers and it rerouted

25:24

fanning out the tweets to anyone

25:26

who followed Ashton Kutcher through a separate service.

25:28

There was all this tourist stuff that was like not

25:31

the right way to write code at all. Twitter

25:35

needed a stronger tech backbone. More

25:37

servers, more engineers to rebuild the plane

25:39

in mid-flight, and that was going to take even

25:42

more VC money.

25:45

In 2009, Twitter completed a funding round that valued

25:47

the company at a billion dollars.

25:50

And here it's important to place Twitter in the history

25:52

of social networking startups. After

25:55

the first web bubble collapsed in 2000, investors

25:57

stayed far away from consumer tech. But

26:01

a few years later, there was buzz around web

26:03

startups that connected people with each other

26:05

and that relied on those users to post for

26:08

free things other users would

26:10

find interesting. And that meant zero content

26:13

costs. These things

26:15

grew fast. They didn't require much money to

26:17

start. And every year, someone made

26:19

one that was more popular. First,

26:21

there was Friendster, which got overtaken by MySpace,

26:24

which Rupert Murdoch bought for $580 million. And

26:28

MySpace got surpassed by Facebook, which

26:30

was supposed to be worth billions. Investors

26:34

saw a pattern. Maybe Twitter would

26:36

be the next big winner. Here's

26:38

investor John Borthwick again. There's

26:41

this cycle that people get

26:43

into, which is expectations following

26:45

that competitor, that Facebook competitor, raising

26:48

more capital expectations against

26:50

that. And that people are basically

26:52

chasing that tail. We want to be Facebook.

26:54

Facebook's really big. In order to get really big,

26:56

we need to hire a bunch of people. We need to raise money

26:59

to do that. And then that then ratchets

27:01

up our expectations. We have to go down that path.

27:03

Right. It wasn't crazy to expect

27:05

in 2010 or 2011 that

27:07

one day Twitter would have a bigger valuation than Facebook.

27:10

Sean Garrett became Twitter's first PR

27:12

guy around this time. Did people feel

27:14

that way inside the company? At the time,

27:17

like, there was no reason to believe that

27:19

Twitter can be more powerful, especially in a

27:21

mobile world, than Facebook. Twitter

27:23

was growing fast, faster

27:26

than any other online service. When

27:28

Oprah Winfrey signed up in April 2009, Twitter had

27:30

about 20 million users. The

27:34

year later, there were 100 million. Twitter

27:37

is still small compared to Facebook, but

27:39

it was catching up fast. And

27:42

even Facebook founder Mark Zuckerberg feared

27:44

Twitter might beat Facebook, or at

27:47

least be a real competitor. Zuckerberg

27:49

tried to buy the company in 2008. And

27:52

after Twitter turned him down, Facebook

27:54

spent years trying to compete with Twitter. But

27:57

even if Twitter was growing faster than anything

28:00

else on the internet, there was still one

28:02

big problem with this scenario. When

28:04

Dick Costolo joined his COO

28:07

in 2009, Twitter was barely generating any

28:10

money. You don't have to be profitable

28:12

if you're a tech startup. You're supposed to be growing

28:14

at all costs, and it's okay to

28:16

make zero dollars when you start. The

28:19

generous term for this is pre-revenue.

28:22

But you do eventually have to start showing the world

28:24

how you're going to make real money someday.

28:27

Investors

28:30

saw a single pathway for revenue, selling

28:33

ads. But advertising was

28:35

never a great fit for Twitter, according to John

28:37

Borthwick.

28:38

The nature of the real-time web is

28:41

that it makes it incredibly hard

28:43

to, if not impossible, to

28:45

be able to target intentful behavior.

28:48

And so the most lucrative forms of advertising

28:51

that we've seen on the internet are ones where people

28:53

basically are placing intent into

28:55

a search engine. You want to buy a foldable

28:58

kayak? So you type foldable kayak

29:01

into Google, and then Google serves you ads

29:03

for foldable kayaks. That is the best

29:05

ad business ever. The second

29:08

best business is Facebook's ad business. You

29:10

have billions of users who tell you a lot

29:12

about what they like and care about, and you can maybe guess

29:15

that they're interested in foldable kayaks. Google

29:18

and Facebook are marvels of modern

29:20

advertising, and they are money printing machines.

29:23

But Twitter didn't have the I would like to buy

29:25

this thing now feature that's built into Google, and

29:28

it didn't have Facebook's scale. Twitter's

29:30

investors wanted this revenue model. The

29:33

founders, including CEO Ed Williams,

29:35

were not so sure. Remember, Williams had

29:37

founded Blogger and sold it to Google,

29:40

the world's biggest advertising company.

29:43

Here's Jason Goldman again. We did have ambivalence

29:45

about it. I would say that as ambivalence

29:47

about it is one of the things that primarily got

29:49

him fired at the end of 2010. This

29:52

was Twitter's second leadership change in two

29:55

years. It replaced Williams

29:57

with Costolo, who had been Twitter's COO. Like

30:00

Williams, Costolo had built an early web

30:02

business and sold it to Google. Unlike

30:05

Williams, he was actually focused on making

30:07

money. We just raised money at a billion

30:09

dollar valuation. We got to start taking

30:11

it seriously as a business. But Twitter

30:13

didn't want to rely on banner ads, which were the

30:15

standard format at the time. The

30:18

ad networks, and specifically

30:20

Google's and Facebook's ads, were

30:23

mostly what you would consider

30:25

in the right-hand rail and the demilitarized

30:28

zone of content on the web. In

30:30

the middle of your screen, you had a feed. On

30:32

the top and on the right rail, you had ads. That

30:35

works okay on a laptop. On

30:37

mobile, of course, there is no

30:39

right rail, there's the feed. If you

30:41

use Twitter at all in the last 10 years, or

30:44

really any app that has a feed, you

30:46

know the solution. Put the ads

30:49

in the feed.

30:50

The ad is a tweet.

30:51

It can be replied to, it can be retweeted, it

30:54

can be favorited, etc. Then

30:56

we can have a set of tweets that are

30:58

content tweets, and some tweets that are ad tweets,

31:00

and the ad tweets we can insert into the feed. And the ads

31:03

are content. The ads are content, and those

31:05

were the first in-feed ads.

31:07

This is hard to remember now that every

31:09

ad is part of the feed. But at the time,

31:12

promoted tweets really were something different.

31:15

They were a new solution to the problem of how to

31:17

get ads in front of eyeballs. And they

31:19

worked. So much so that Facebook,

31:21

and later Instagram, and everyone else did

31:24

exactly the same thing. So

31:26

Twitter was finally making a little bit of

31:29

money. But if it wanted to make real

31:31

Facebook-sized money, it was gonna have to

31:33

keep growing at a crazy pace. And

31:36

that was no longer happening. At

31:38

the beginning of 2009, Twitter was growing

31:40

faster than anything else on the internet. But

31:43

by the end of that year, it had already begun leveling

31:46

off. The company was having to work harder

31:48

and harder to find new people to sign up. This,

31:51

it turns out, may be the most existential

31:53

Twitter question of them all. What

31:56

if this thing just isn't that interesting

31:58

to most people? So

32:00

why wasn't Twitter for everyone? Let's

32:03

start with the obvious. To make a tweet, you

32:05

have to write something. And for a lot

32:07

of people, writing is no fun.

32:10

Twitter is really hard to use.

32:12

Jay Wirtham loved it, but Wirtham is a writer

32:14

and most people aren't.

32:16

For all the supposed simplicity

32:19

of the service, you just make an

32:21

account and you can write whatever you want. It's really

32:23

hard to figure out what works. Like if

32:25

you're just tweeting any thought

32:27

that comes into your head, it's not really

32:30

going to probably land.

32:32

Wirtham is describing this from the perspective

32:34

of someone who wants to share stuff on the internet.

32:36

Just like Twitter's earliest employees

32:39

and users. Those people assume most

32:41

people would feel the same way. Jason

32:44

Goldman again. When we did Blogger,

32:46

we fully thought that everyone was going to have a blog.

32:49

So then when we do Twitter, I'm like, everyone's

32:51

going to do this. Like, this is clearly going to be the one

32:53

that everyone does because it's short. It's even

32:55

more seductive than blogs. Like you know, you

32:57

could just say, I'm eating a taco. Everyone's

32:59

going to do this. It turns out not everyone's

33:02

going to do this. Not everyone wants to be a content creator.

33:05

There are some people who really do

33:07

want to be content creators. Like

33:09

journalists who famously like the sound of

33:11

their own voice, who love getting likes

33:14

and retweets from their peers even more. Or

33:16

like certain celebrities or politicians

33:18

who imagine they could use Twitter to interact directly

33:21

with fans and followers. Or

33:23

eventually, a rocket company CEO

33:26

with a bottomless supply of takes and means.

33:29

These folks were the super users, the top 10%

33:31

of Twitterers who wound up producing 80% of

33:34

the content on the site. And gradually,

33:36

Twitter accepted that reality and landed

33:39

on the idea that Twitter was a great place to read

33:41

other people's tweets, even if

33:43

you didn't want to write them yourself. But

33:45

the next problem was getting the right tweets in front of

33:47

you when you first tried Twitter so

33:50

you could see how great it could be. Here's

33:53

Michael Sippy, who is VP of Product

33:55

under Costolo.

33:59

ton of fun. But

34:02

coming in, it

34:04

takes a lot of work

34:06

to actually get that follow

34:08

graph correct. And so we spent a lot

34:10

of time and a lot of technology to

34:12

essentially help you tune that

34:15

initial set of accounts that you would follow.

34:18

And that meant a bunch of different

34:20

hypotheses that we had. One is like, well,

34:23

we should do things like Facebook, it's a social

34:25

network. So let's go scrape your

34:27

address book and make sure that there are friends that

34:29

are there and like you can go find the

34:31

problem is that we did

34:33

not have the density of Facebook or the growth of Facebook.

34:35

So like, your friends weren't posting enough.

34:38

And then that's not good enough.

34:39

And if in your first

34:42

hour, day, three days of usage,

34:45

you're not finding value in the product, and

34:47

you're not willing to put in the work to go find

34:49

those people to follow,

34:51

you're gonna churn.

34:53

Churn is a term you hear a lot

34:55

when you talk to Twitter veterans about the company's

34:58

problems. Churn is when someone tries

35:00

your service and then leaves. And Twitter

35:02

had a lot of churn. But

35:04

for a long time, Twitter acted like it couldn't afford

35:07

to care about churn and had to care about growing.

35:10

Here's Jason Goldman. I think we were too

35:12

focused for too long on growth. And

35:14

this is during the era where growth hacking became,

35:17

you know, a concept and this idea that you

35:19

just needed to get more people in the top of the funnel

35:21

and pay less attention

35:24

to the churn that is happening when

35:26

people bounce out because they simply don't get it. There's

35:29

a pretty good chance that you've used Twitter, because

35:31

you're listening to this podcast about Twitter. But

35:34

most people have never tried Twitter. And

35:36

lots of those who have don't stick around. It's

35:39

so hard to get them to come back. That's

35:41

Michael Sippy again, the former head of product.

35:44

You can get kind of occasional use of a product,

35:46

right? So you'll hear people like, well, I'm

35:48

not really only on Twitter during the NBA finals, because

35:51

that's when it gets fun. I'll come on for

35:53

the Olympics. I'll come on for the World Cup. I'll come on

35:55

for the elections, because that's my interest area.

35:57

And that's what it is.

35:59

Twitter did try to make itself more appealing

36:02

to new users and to make it stickier for the

36:04

ones that showed up. And when you look back

36:06

at Twitter's history, there is a long list of kudawudda

36:09

shuddos. The cost of those

36:11

big idea was to make Twitter more like

36:14

Facebook. Everyone loves to go to Facebook

36:16

and look at their friend's photos. That's like a massive

36:18

use case. We don't have media at all. Twitter

36:21

needed photos. And in 2010,

36:23

2011, the best place to find photos

36:26

after Facebook was Instagram,

36:28

the super buzzy photo sharing app. We

36:31

had talked to them about buying them before

36:33

I was CEO. And then when I was CEO

36:37

in February 2012, we tried to buy

36:39

them aggressively. We

36:41

couldn't convince them to do it. A

36:43

few months later, Mark Zuckerberg convinced

36:45

the Instagram team to sell to Facebook instead

36:48

for a billion dollars. This is a

36:50

humongous amount of money at the time. Twitter

36:53

did end up buying Vine, an app that shows

36:55

looping six second clips. It

36:57

was intriguing, but no one at Twitter knew exactly

36:59

what to do with it. Michael Sippy.

37:02

How do we take Vine technology and put it into

37:04

the Twitter product? And what's that mix? And how does

37:06

that work? And what does distribution look like? And how

37:09

much Twitter branding is there on Vine? All

37:11

of those conversations, I think, drove that team

37:13

nuts. Inside a different kind

37:15

of company, Vine could have eventually become

37:18

TikTok, the most powerful entertainment

37:20

app today. But by the

37:23

end of 2016, Snapchat and Instagram had eaten

37:25

Vine's lunch. And Twitter closed

37:27

it down. Twitter tried other

37:29

ways to build out the platform, to give people

37:31

other stuff to do besides writing and reading

37:34

tweets. It launched Periscope,

37:36

a live video service. It launched its

37:38

own music service, except it wasn't really what

37:40

users wanted out of a music service because it didn't

37:42

really play music. You're supposed to

37:45

use it to follow musicians or something. There

37:47

was lots of talk about turning Twitter into a place where you could

37:49

buy and sell stuff. To

37:52

Costolo, though, the big one that got away is

37:54

always going to be Instagram. Do

37:56

you replay the Kuda Woda? What

37:59

happens? I think we

38:01

would have won. The fact that they

38:03

had built this gorgeous,

38:05

beautiful photo sharing application

38:08

that was becoming the dominant photo sharing

38:10

application, which is why we wanted to buy it

38:12

and why Mark did buy it. We

38:15

owned, if you will, the text space

38:18

and would go on to acquire Vine,

38:20

which was number one in the app store for weeks

38:22

and weeks and weeks right before Instagram launched video

38:25

and sort of knocked it out, would

38:27

have been positioned to have these three apps that are

38:29

perfectly suited to their media types.

38:32

We just didn't get Instagram and, you know,

38:34

that probably would have changed the course of the internet

38:37

in some important way. Maybe,

38:42

or maybe none of those features really mattered.

38:44

People who were on Facebook wanted

38:46

to use Facebook, not Twitter. In 2009,

38:49

Zuckerberg had been so freaked

38:51

out by Twitter that he wanted to own it or

38:53

bury it. But eventually Zuckerberg

38:56

and crew learned to stop worrying about

38:58

Twitter. In Nick Bilton's book,

39:00

Hatching Twitter, Zuckerberg describes

39:02

Twitter's team as lucky incompetence.

39:06

It's as if they drove a clown car into

39:08

a gold mine and fell in. This

39:11

wasn't a unique assessment and there's a lot

39:13

of evidence that supports it. Take Twitter's

39:16

CEO musical chairs, for instance, in the 15 years

39:19

before Elon Musk bought the company, Twitter

39:21

had five different CEOs, two

39:24

of whom were Jack Dorsey. Lots

39:26

of startups swap out leaders at different times,

39:29

but that's the kind of instability that indicates

39:31

a company fundamentally can't decide

39:33

what it should be. To

39:36

be fair, Twitter did make

39:38

it to an IPO in 2013, where Twitter's

39:41

private shares turned into public stock and

39:43

made tons of money for its investors and

39:45

its early employees. But ever since

39:47

then, Twitter's stock lurched and stalled

39:49

as various managers tried, unsuccessfully,

39:53

to convince Wall Street that Twitter was going to be

39:55

the next Facebook. Twitter

39:57

did build an ad business worth nearly $5 billion.

40:00

billion a year. That sounds like a lot

40:03

until you compare it to Facebook, which does $113

40:06

billion a year, or Google,

40:09

which does $224 billion. Another

40:14

reason Elon Musk could buy Twitter in 2022

40:17

is that unlike tech companies like Facebook

40:19

and Google and Snap, the founders

40:21

didn't control the company's stock so they couldn't

40:24

fend off an unwanted buyer. But

40:26

the main reason is that Twitter had been public

40:29

for nearly a decade and investors didn't

40:31

care about it anymore. They'd heard

40:33

Twitter's story over and over and they weren't buying

40:36

it. So when the world's richest man

40:38

showed up and wanted to pay way more than

40:40

Wall Street thought it was worth, they

40:42

had no choice but to take the deal. So

40:46

let's spell this out. Deciding

40:49

to take venture money, deciding

40:51

to be an ad business, deciding

40:53

to compare yourself to Facebook and losing

40:55

out on that comparison, those

40:57

are all decisions that led to the real-life

40:59

scenario where Elon Musk owns Twitter.

41:03

And why should we care about what happens

41:05

to Twitter? Objectively speaking,

41:07

it's a tiny company compared to real tech

41:09

giants.

41:10

But...

41:11

Twitter, for better or for worse, has completely

41:14

shaped and participated

41:16

in a cultural conversation for the last 10 years,

41:19

you know, and probably will for the next 10 years.

41:22

And I think that alone is worth just

41:24

looking at. Like, what is this thing that

41:27

had so much power in shaping national

41:29

politics from elections to social

41:31

movements to even our understanding of gun

41:33

violence? I mean, Twitter has played a massive

41:36

role in that and there's no denying it.

41:38

This series is called Land of the Giants.

41:41

But Twitter was never a giant, not

41:43

by market cap, not by user numbers. But

41:46

it could exert an influence totally disproportionate

41:49

to its size or its value. Figuring

41:52

out how exactly Twitter commanded that

41:54

power and what exactly that

41:56

influence was, those are questions

41:58

that have flummoxed a lot of smart people. people. And

42:01

that's why we're here, to try our best to provide

42:03

some answers. Because there is one thing

42:06

I'm certain about. Twitter, for

42:08

all its flaws and in all its glory, could

42:11

only exist in a certain time, and that time

42:13

is winding down. So while

42:15

it's still here, let's talk about why Twitter

42:18

mattered, and what that tells us

42:20

about what comes next.

42:32

Audio

42:35

clips from Al Jazeera, a TED

42:37

conference, south by southwest, and Marketplace.

42:40

Land of the Giants, the Twitter fantasy, is

42:43

a production of Vox and the Vox Media

42:45

Podcast Network.

42:47

Matt Frasica is our lead producer. Alua

42:49

Kemi, Alade Sui, is our producer. Megan

42:52

Cunane is our editor. Charlotte

42:54

Silver is our fact checker. Brandon

42:57

McFarland composed the show's theme and engineered

42:59

the episode. Art Chung is our showrunner.

43:02

Shachk Kerwa is our executive producer. I'm

43:05

Peter Tofk. If you like this episode, as

43:07

always, please share it. Follow the show

43:09

by clicking the plus sign in your podcast

43:11

app.

43:18

Support for Recode Media came from Oracle.

43:21

AI is everywhere right now, but

43:23

it also needs a lot of processing speed, and that

43:25

gets expensive fast. It might help

43:28

to upgrade to the next generation of the cloud.

43:30

That's Oracle Cloud Infrastructure, or

43:33

OCI. OCI is the single

43:35

platform for your infrastructure database, application

43:37

development, and business.