0:00

In May of this year Rand Fishkin

0:02

received an email from a leaker

0:06

And I had never emailed with him before I

0:08

didn't know who he was The

0:11

email was labeled confidential Google

0:14

Rand is a public figure in internet marketing

0:17

and this leaker was claiming to have access

0:19

to Internal documentation about

0:21

something very important and

0:23

up until this point very

0:26

secretive these are wild

0:30

accusations I don't know if

0:32

you agree with this Scott, but maybe no

0:34

one thing impacts the shape of the

0:36

modern internet more than Google

0:38

search If you

0:40

remember the internet before Google

0:43

you would strongly agree with

0:45

that sentiment Especially if

0:47

you use some of the alternative search

0:49

engines that used to exist. I

0:51

remember asking Jeeves I Forgot

0:54

about Jeeves actually right a lot of people

0:57

forgot about Jeeves Today Google

0:59

is over 80% of search traffic

1:01

with 8.5 billion searches a day

1:03

2 trillion annually There's nonsense

1:05

numbers at this point a hundred thousand

1:07

a second the average person listening

1:10

You probably Google more times in the day than

1:12

you eat meals What

1:14

Google serves dictates what

1:16

information people consume how

1:18

Google ranks websites? Dictates what websites

1:21

live and die Optimizing

1:23

for that system is now an entire industry

1:27

and For a long time

1:29

all we've really known about that system is what

1:31

Google tells us Kind of

1:33

for good reason because you don't want it to get

1:35

any more gamified than it already is But

1:38

really everything about what Google tracks and

1:40

doesn't track what data they use and

1:43

what they don't that all

1:45

comes from statements from Google their

1:47

PR people their executives You

1:50

know what we know about Google search is what Google has

1:52

told us There

1:55

have never been any documents about the API

1:57

leaked to confirm or importantly

2:00

contradict those public statements.

2:03

Now, Rand's sitting there looking at

2:05

something that potentially does at

2:08

this email from this leaker claiming to

2:10

have a copy of a bunch of

2:12

internal documentation about the Google Search API.

2:16

When we hold these leaks up against those statements,

2:21

Rand started to find some interesting stuff. When

2:24

I finally got on the phone with him, he

2:26

pulls up the trove of documents

2:28

and my mind exploded.

2:32

The thing for me is were

2:35

the statements made in the

2:37

same kind of timeframe as

2:39

the documentation that we've received? The

2:42

other thing is why

2:44

did Rand not just

2:46

keep these documents private?

2:50

Somebody gave him the key to the treasure

2:52

chest. Google's intentionally kept

2:54

their search algo, a black box, and

2:57

modifies it every time somebody gets close

2:59

to figuring out ways to cheat it,

3:02

creating the industry you speak of with

3:04

thousands and maybe millions of SEO people

3:07

and companies in the world these days.

3:09

They even offer their own certifications, which

3:11

don't tell you how to do it.

3:16

If you were given these

3:18

leaked documents, you would

3:21

be maybe one of the only people outside

3:23

of Google that knew how to cheat the

3:25

system. He chose to

3:27

expose them publicly rather than build

3:30

an industry out of it. Which

3:33

I guess kudos to you. It kudos to

3:35

Rand on that one. To the temporality, you

3:38

bring up a great point. So did Google.

3:40

I think that for the people combing

3:42

through these thousands and thousands

3:44

of documents, figuring out

3:47

that timeline is the

3:49

rat's nest to untangle right now. As

3:52

for Rand, using his new found

3:54

knowledge for personal gain, you

3:56

know I didn't ask him about that.

3:59

Maybe I should. A

4:02

lot has happened in the weeks since he got that

4:04

email, and I wanted to talk to him about it. Google

4:07

is in the final stages of a big antitrust

4:09

case with the DOJ, it concerns Google search, and

4:12

whether they're engaged in anti-competitive practices, it

4:14

is not the only case of this

4:16

kind they have been involved in. Because

4:19

Google, technically their parent company alphabet, owns

4:21

websites that compete on the internet. So

4:24

how Google ranks websites matters not just

4:26

to all of us, but to them.

4:29

And that does potentially create

4:31

a conflict of interest. There

4:34

is a rising frustration

4:36

with particularly

4:39

Google's self-preferencing behavior. Essentially

4:42

Google doing things inside of Google

4:44

search where they own 95% of

4:46

the market to benefit other parts

4:49

of Google. I

4:51

wanted to know more about all this, about the

4:53

leak, Google's response, the

4:56

experience of getting wrapped up in this story. So reached

4:59

out to Rand, fun chat, appreciate

5:01

his time. This

5:03

is my conversation with Rand Fishkin, co-founder,

5:05

Spark Toro, and Snack Bar Studios

5:08

about the Google API leaks here

5:11

on hacked. Rand,

5:27

thank you so much for sitting down to talk with

5:29

me about this. Yeah, my pleasure Jordan, thanks for having

5:32

me. So I think we're the average person a lot

5:34

of this can seem in the weeds. For normal internet

5:36

users, to what extent is

5:38

the internet shaped by Google's search algorithm?

5:42

I can actually tell you this exactly because

5:44

I did some data

5:46

analysis of clickstream panel data, which

5:48

is essentially, it's kind of like

5:50

Nielsen TV set boxes from the

5:52

1960s, but fast forward

5:55

to 2024 and collect data

5:57

about every URL that's visited by 10% of the

6:00

users. of millions of devices through

6:02

a panel and 70% of all

6:05

internet traffic is sent by Google. Wow.

6:11

Pretty brutal. Brutal,

6:13

why do you say that? Well, I

6:16

am someone who believes that

6:18

monopoly power tends to stifle

6:21

innovation, creativity, and opportunity. And

6:23

I think that Google's stranglehold

6:26

on internet traffic

6:28

and on what

6:31

content people see and

6:33

don't see really

6:35

limits what

6:38

is created. Right? So for content creators, I'm

6:40

sure you know this world well. For

6:43

content creators, if you're in the

6:45

video game space, what is able

6:48

to be surfaced on Steam or the Nintendo

6:50

Switch Store or the Xbox Store or the

6:53

PlayStation Store, that plays a huge role

6:56

in what game designers choose

6:58

to create. Similarly,

7:01

when you make things for the internet,

7:04

whether it's a YouTube video or an

7:06

article about poison dart

7:08

frogs in Central America or the

7:12

best mustache wax for curly hair,

7:16

you're going to change what you

7:18

do based on what Google tells

7:22

you is important and how you can

7:24

potentially get traffic to your page.

7:26

And so, you know, we

7:29

kind of end up with this Google shaped internet. We're

7:31

talking because there was a trove of

7:33

leaked documents concerning the search algorithm that

7:36

shapes that Google shaped internet. And these

7:38

this leak came into your possession. Take

7:40

me through that story starting with I

7:42

think the email that you got. Yeah,

7:45

I pulled it up on my computer here

7:47

because I couldn't quite remember exactly

7:51

how it went. So I got an email

7:53

from a guy in

7:55

Georgia, Tbilisi, Georgia, Georgia,

7:57

the country. country,

8:00

not Georgia, the state. And

8:02

I had never emailed with him before. I didn't

8:04

know who he was. The

8:07

email was labeled confidential Google.

8:12

And he says, Rand, I know you've been out

8:14

of the SEO industry for a while. SEO is

8:16

search engine optimization. That's sort of the practice of

8:18

ranking web pages in Google and getting traffic to

8:21

them. And I used to run a

8:23

company called Moz, which is in

8:25

the SEO software and education space. And

8:28

he says, you are the first person to highlight

8:30

the influence of click data on search results. From

8:33

what I've heard, Google went so far

8:35

as to manually demote your experiments and

8:37

publicly make statements that are far from

8:39

truthful, including reputation destruction. There

8:42

were several Google representatives who over

8:44

the years said particularly

8:47

harsh things about my work and

8:49

research when I was at Moz. And

8:53

anyway, this email goes

8:55

on to sort of cite all

8:57

these examples. And he claims the

8:59

emailer claims to have proof of

9:01

how Google ranks pages, proof that

9:04

Google has lied publicly dozens, if

9:06

not hundreds of times to

9:08

news sources of all kinds,

9:11

proof that potentially they lied

9:14

to Congress when they when

9:18

their CEO Sundar Pichay talked

9:20

about how they use

9:23

data potentially. There's

9:26

even some suggestion in here that there were

9:28

lies about the Department

9:30

of Justice case that was prosecuted last year. These

9:34

are wild accusations,

9:37

right? I mean, if you get an email like

9:39

this, you know, especially I'm

9:41

six years out from running Moz from, you

9:43

know, being outside the SEO industry, I kind

9:46

of look at this and go, well, I

9:48

have to say this person sounds credible,

9:51

but also incredibly far fetched.

9:54

And so extraordinary claims require extraordinary

9:56

proof, right? So I, you

9:58

know, I write back to this person. and sort of

10:01

say, okay, thanks for telling me all this stuff. Like,

10:03

what are your goals here? And

10:06

they say, I think you should be the

10:08

one to publish this leak data and I

10:10

wanna show it to you. So

10:14

we schedule a phone call. This is how

10:16

unexcited about it I was, Jordan. It

10:19

was, I received the email on May

10:21

5th. On

10:23

May 23rd, I canceled the scheduled call

10:30

with the guy, because I wasn't feeling so well. And

10:33

then we rescheduled for

10:35

later in May. When

10:38

I finally got on the phone with him, he

10:40

pulls up the trove of

10:42

documents and my mind

10:45

exploded. I

10:47

mean, this is, here's

10:49

essentially what this guy showed

10:51

me was the

10:56

API documentation. API is sort

10:58

of how you make programmatic,

11:00

calls at scale. It's

11:04

the API documentation internal

11:06

to Google's search engineering team.

11:09

So this is, imagine you and I work on

11:11

Google search engine and imagine

11:13

we're programmers there and we are

11:16

trying to make Google web search

11:18

even better. These

11:20

are the list of all the types of

11:23

data that we can

11:25

call in order to build

11:28

or modify an algorithm, a

11:31

ranking system, right? To choose which pages

11:33

appear which, before which other ones. And

11:36

not just by the way, not just Google web

11:38

search, YouTube is in here, Google Android searches in

11:40

here, Google Maps is in here and local, Google

11:43

News is in here. All

11:45

the different flavors of things that

11:47

Google searches publicly for human beings. And

11:51

as he showed me these, we're

11:53

talking about 2,500 documents containing 14,000 different.

12:00

attributes or features that you can call,

12:02

right? So, you know,

12:04

if somebody says, oh, Google search is simple, it's just X,

12:06

Y and Z. You can be like, yeah,

12:08

yeah, X, Y and Z. It's 14,000 X, Ys and Zs. Literally,

12:13

14,014 X, Ys and Zs are what they used as

12:17

of March of 2024. So

12:20

we get off the call, you know,

12:22

and I'm kind of losing my mind going through this. And

12:25

the first thing I do is hit up

12:27

a few people quietly in my network. A

12:30

few people who used to work in

12:32

Google as engineers, software engineers, three

12:36

people in particular. The first one that I reached

12:38

out to said, I don't

12:40

wanna talk about this. And I'm

12:42

not willing even anonymously to broach

12:44

the subject. Okay, but the other

12:46

two said, yes,

12:49

I, you know, happy to take a look.

12:51

They took a look and they came back to me and basically

12:53

said, yeah, this absolutely looks

12:55

legitimate. I didn't personally have access

12:57

to this document when I was at Google. You

13:01

wouldn't have access to this unless you were

13:03

on that specific search engineering team. But

13:06

this is absolutely

13:08

Google formatted, you

13:11

know, internal speak throughout some,

13:13

there's almost no way this could have been

13:15

faked. And

13:18

then I talked to an expert in search ranking

13:20

systems who I had known from my time in

13:22

the industry, a guy named Mike

13:24

King. Mike runs an agency in New York

13:26

called I-PoleRank. And he

13:29

and I have been friends for many years. Mike

13:31

is ludicrously talented,

13:34

just extraordinarily detailed in his

13:36

research. He's been working on a book about

13:38

information retrieval, which is the science of how

13:41

search engine works, search engines work.

13:43

And so he's got

13:45

this absolute plethora of

13:48

relevant experience

13:51

around this stuff. So I show him the leak. I

13:54

called him, I called him, it was a Friday

13:56

night. He's out with his kids in Brooklyn at

13:58

the park. He's like, okay, okay. Wait, what are

14:00

you telling me? All right, let me

14:02

just go home and look at this. And

14:06

I think he stayed up all night and most

14:08

of the weekend working on the leak. And then

14:10

on Monday night, he and I both published blog

14:14

posts describing this leak, sharing

14:17

what was inside them. Obviously

14:19

the early analysis was very

14:22

incomplete, but already there were dozens

14:24

of features that were extraordinarily

14:27

interesting, contradicted many statements Google

14:29

had made in the past. And

14:31

when we published that, Jordan, the

14:34

internet exploded. I mean, hundreds

14:37

of thousands of visits just to my blog post, I'm

14:39

sure to Mike's as well, you

14:41

know, interview requests from two

14:45

dozen publications, you know, everyone from the

14:47

Verge, the New Yorker to New York

14:49

Times and Washington Post and Wall Street

14:51

Journal and everyone else you can imagine.

14:55

And, you know, Cara Swisher talked about

14:57

it on her podcast and, you know,

15:00

it was the top of Hacker News, it's the top of

15:02

Tech Meme. It

15:04

was an insane two weeks after that.

15:09

And since then, you know, people have

15:11

been analyzing this leak because it's public. Anyone

15:13

can see it. You can go right now and

15:16

look at the 14,000 inputs that

15:19

make up Google's search ranking algorithm. That

15:21

had never been possible in the

15:24

last quarter century. It's mind blowing. I

15:26

want to dig into what we learned

15:29

about the API that we didn't previously know.

15:31

And maybe just start, because I found this

15:34

one interesting. Google makes

15:36

Chrome. Google also

15:38

sells ads. Google representatives

15:40

have long stated that they don't

15:42

use any information about users in

15:44

Chrome for ranking, which is very

15:47

important for selling advertising. And

15:49

that always seemed kind of shocking to

15:51

not use this massive trough of privileged

15:54

data that you could potentially be gathering

15:56

in your biggest business line over here.

16:00

These leaked documents maybe tell

16:02

a story that that separation of church and

16:04

state isn't quite so separated. Can

16:06

you, to start with what's in here, tell

16:08

me a little bit about that. Yeah, yeah,

16:10

there's no separation at all. I

16:13

mean, when you look at how Google

16:15

measures, for example, one

16:17

of the things that would happen when any

16:19

human being performs a search is, let's say

16:21

you are looking for, this happened to me

16:24

recently, I was looking up the

16:26

Aquarium of the Pacific, which I think is

16:28

in Long Beach, California. I wanted

16:30

to find out how long they were

16:32

running their frog exhibit. So

16:35

they've got a new frog exhibit that just

16:37

launched. It showed up in my Google News

16:39

Feed. I was like, ooh, I wanna go

16:41

see Poison Dart Frogs when I'm down in

16:43

California over the summer. And so

16:45

I do this search

16:48

and I click on the first result, which

16:50

is about the event, but it doesn't say

16:53

how long it's running. So then I click

16:55

back to Google's results. I

16:57

scroll down a little bit until I find a

17:00

press mention, right? Someone talking about them in

17:03

the news and saying, oh, the exhibit is

17:05

planned to be permanent. I click that one

17:07

and it, okay,

17:10

great. So I don't have to worry about when it's

17:12

gonna happen. Here's what Google's

17:15

documentation says. If you

17:17

click on a search and then you

17:19

click the back button and

17:21

you choose another result, that

17:24

suggests to Google that

17:26

the other result is probably more relevant and

17:28

deserves to be higher up in the rankings

17:31

than the one you left. You

17:33

left a result and your search

17:35

was unsolved, and then you bounced

17:38

back to the search results and chose a

17:40

different result. And once you went to

17:42

that one, then your search was resolved. This

17:45

is called pogo sticking. It

17:47

has long been used in

17:49

information retrieval literature. And

17:52

here it is right in the documentation. You

17:54

can observe that Google is not measuring this

17:57

through Google analytics, which many people speculated for

17:59

a long time. They're not just

18:01

measuring it by looking at what happens on their

18:03

search results page. They are

18:05

looking at the billions of devices

18:07

that use Chrome, Google Chrome, as

18:10

their browser to be able to

18:12

measure this. And

18:14

this is only one of hundreds

18:17

of uses of Chrome data

18:20

inside the ranking systems. As

18:23

another example, which I find particularly fascinating,

18:25

being someone who was in the industry,

18:28

many people know that one of the ways

18:30

that you rank higher in Google is get

18:32

lots of links pointing to you, right? If

18:34

lots of other pages on the internet link

18:36

to your page, that tends to suggest to

18:38

Google that you are more important than someone

18:40

who has very few links. So

18:44

inside the leak, you can see

18:47

that Google uses Chrome data, traffic

18:49

data to demote or increase the

18:51

value of links that come from

18:54

pages that either don't receive or

18:56

do receive traffic. For

18:59

example, if you are linked to by an article

19:02

in The Economist that got a lot of traffic,

19:05

that link is probably worth much more

19:08

than a link from, you know,

19:11

randomwebsite.net that gets

19:13

no traffic. That wasn't

19:15

always true, by the way. Google

19:18

used to be very manipulatable. Back

19:20

when I started in the industry, you could

19:23

get a bunch of links from a bunch

19:25

of different scammy sketchy websites and rank nearly

19:27

anything anywhere you wanted to. But

19:30

Google, Google managed to find a

19:32

really clever solution to this using

19:34

traffic data from Chrome. Google

19:37

has long stated and you gesture towards this,

19:39

but Google has long stated that they need

19:41

to balance what information they

19:44

make available publicly about how search

19:46

works, just frankly, because

19:48

the more that's public, the more is

19:50

gameable from everyone from full-blown scammers all

19:52

the way over to professionals in optimizing

19:54

for search engines. Talk to

19:56

me about that balance that we're seeing in

19:59

these leaks between transparency. and kind of

20:01

making the internet even more

20:03

broken in certain ways than it already is. Yeah,

20:07

I'm gonna throw out there my suspicion that

20:09

10 years

20:12

ago, maybe 15 years ago, if

20:14

a document like this had

20:16

leaked, it would have been

20:19

quite damaging to Google's ability

20:21

to organize

20:23

the web and make its information useful. I

20:27

will grant that. And I think

20:29

that's because Google just wasn't that

20:31

sophisticated back then, right? The systems

20:33

for ranking were gameable.

20:38

They really were. I

20:40

look at this leak today and

20:42

everything I've observed in here suggests

20:44

to me that Google is nearly

20:47

bulletproof. Go, go

20:49

spam all you want. I

20:52

don't think you're gonna break through. This system is

20:55

not only sophisticated and elegant, but it

20:57

is crafted in such a way that

21:00

in order to game the system, you

21:03

would have to be really useful to

21:05

real human beings and a lot of them.

21:08

If you are, is it really spam

21:10

anymore? Right? Like

21:13

if you don't make things that achieve

21:15

real popularity, that real people link to

21:17

from their websites, that real news sources

21:19

talk about and pick up, that get

21:22

traffic, that once they start ranking, even

21:24

if you were to game all the

21:26

other signals, once they start ranking in

21:28

Google, if it doesn't successfully answer lots

21:30

of searchers queries, you're

21:32

gonna fall out of the rankings and someone else

21:34

will rise. So I really,

21:37

I don't see a downside

21:40

to Google sharing this. I think if

21:42

some conspiracy theory 10 years from now is

21:45

like, oh, actually it wasn't a leak, they

21:47

put it out there intentionally and I've got

21:49

the email to prove it, that

21:52

wouldn't totally shock me because very

21:54

frankly, this is useful information for...

22:00

not getting scammed by sketchy

22:02

SEO providers. But it

22:04

is not a roadmap that is

22:06

going to tell you, Oh, man,

22:09

if I just put, you know, the number

22:11

seven in my title tag 12 times, you

22:13

know, I'll rank at the top. There's nothing

22:15

like that. Sure. There's

22:17

no like name your business triple A plumbers because

22:19

the three A's come at the start of the

22:21

telephone book. Yeah, right. It's not it's not the

22:23

white pages in 1985. Sure. Exactly.

22:27

Has Google publicly commented on the

22:29

documents that were leaked to you?

22:32

So I did get a private

22:35

email from a Googler the

22:37

night I published it, that was quite upset with

22:40

with one characterization of how I described an

22:43

event, which and I did change it in

22:45

the post. And then

22:47

I believe it was the next week. Google's

22:53

made up made a public statement through sort of

22:55

a PR person that the

22:57

leak was authentic. But they

22:59

urged people not to misread,

23:04

you know, potentially incomplete data. And, you

23:06

know, in fairness, the

23:09

leak does reference some of the

23:11

references in the features do reference

23:13

other data sources that

23:15

we can't access. For example, there's

23:17

a there's a list

23:20

of a white list of election

23:22

approved election news providers, right?

23:25

So that if you were to, you

23:28

know, it's January 6 2020.

23:30

And you and you, you know,

23:32

you're an American and you search for who won

23:34

the election, you know, was there a dispute? Is

23:36

there any evidence that the election was problematic? Google

23:40

wants to make sure that the accurate

23:42

truth is represented in their results, and

23:44

not someone who,

23:47

you know, is misrepresenting that and you

23:49

can imagine certainly that in the in the

23:51

political spectrum, it would not

23:53

be difficult to replicate all the signals that

23:55

you might need including popularity and news references

23:58

and links and clicks and all that. I

24:00

have the quote here from the Google

24:03

spokesperson. It was that we would quote

24:05

caution against making inaccurate assumptions about search

24:07

based on out of context, outdated or

24:10

incomplete information, which as you

24:12

said is importantly not saying these leaks aren't

24:14

real. What do

24:16

you read that statement to mean? Well,

24:19

I don't think it means anything because

24:21

it doesn't even say that these documents

24:23

are outdated or

24:25

that they're inaccurate. It just says

24:27

we caution against generally any information

24:29

that is out of context or

24:32

outdated. These documents are in

24:34

context. In fact, all of the features are

24:36

not all. Many of the features

24:38

are very well described, right? Such that

24:40

if you and I were new engineers

24:42

who joined the Google search team, we

24:45

could read these documents and be like, oh, okay,

24:47

I get what that means. That's when

24:50

I call the Chrome data that tells me

24:52

what percent of people click the back button

24:54

after searching for this. And this is the

24:56

one where they

24:58

have this thing called squashed and unsquashed

25:00

clicks. Squashed clicks,

25:03

they describe it as referring to

25:05

clicks that their spam system thinks

25:07

are not real human beings and

25:09

real devices. And so they

25:11

don't wanna count those clicks, that kind of

25:13

thing. That's what I'm talking

25:16

about. That's what I'm talking about. Scott,

25:19

why do you love Notion? I love that you just toss this

25:21

to me because I love it so much. Because

25:24

I know you love Notion. Because

25:26

I'm reading this data and this

25:28

advertising notes out of Notion. I

25:31

love it because it's just a great place

25:33

to put things. It's a

25:35

great place to structure data. It's a

25:37

great place to build small apps. It's

25:40

a great place to use

25:42

contextual AI to facilitate

25:45

my work and personal life. I store

25:47

everything. Now, I literally have

25:50

Notion documents that store all of my bikes

25:53

and my wife's bikes and every part on

25:55

them. So that when I have to order

25:57

maintenance pieces for them, I know exactly what

25:59

model. of, you know, rear shock it has.

26:02

I use it for

26:04

so many things. So I can't tell you why

26:06

I love it. I just love it. It's just

26:08

a feeling something you feel in your heart. When

26:11

you get a really good piece of software that

26:13

combines your notes and docs into one place that's

26:15

simple and beautifully designed with the power of AI

26:17

is all built right inside of it. Not another

26:19

separate tool in a different browser or tab you

26:21

know you don't have 75,000 tabs running live. You

26:23

just got notion we used it just the other

26:26

day. We use it every day. Yeah, I was

26:28

going to say. There's a huge part of our

26:30

workflow. Just the other day it's like I have two

26:32

instances of it in front of me right now. Notion

26:36

is a place where any team can write, plan, organize

26:38

and rediscover the joy of like it makes work feel

26:40

a little bit more playful and that's really,

26:42

really cool. It's

26:44

a workplace design not just for making progress, but like, you

26:47

know, getting inspired like you're in the same room together. It's

26:49

also like the big thing for me is that it's

26:51

like a it's

26:54

like a app building environment like you can

26:56

build data driven applications so quickly and easily.

26:59

Like I know lots of famous

27:01

content creators that use notion to

27:03

like manage their workflows and projects

27:05

when they're making new YouTube videos

27:07

or podcast episodes. It's

27:10

just a great place to put data

27:12

access data structure data move

27:15

processes. It's just it's just so good for so

27:17

many things. And you know what are

27:20

fine fine listeners can try notion

27:22

for free when they go to

27:24

notion.com/hacked. That's all lowercase later letters

27:27

notion n o t i o

27:29

n.com/hacked. You can start turning ideas

27:31

into action. And when you use

27:34

our link that hacked link you're supporting our show. So

27:36

when you invariably do go to sign

27:39

up for notion because it rips notion.com/hacked.

28:00

We did and we made it with Shopify and

28:02

it was a genuinely delightful experience. Why? Because

28:04

Shopify is the global commerce platform that helps

28:06

you sell at every stage of your business

28:08

whether you're at that like just launching a

28:10

shop online stage or first real life

28:12

store stage all the way to just like, oh my

28:15

God, we sold a million orders stage. Shopify,

28:17

they got your back. We are not at

28:19

that. Did we just sell a million order

28:21

stage? And that's sad. So if you like

28:24

to buy something, visit store.hackpodcast.com and check out

28:26

how great Shopify is. It

28:29

powers over 10% of all e-commerce

28:31

in the United States and Shopify

28:33

is the global force behind big

28:35

companies, not like us, but like

28:37

Allbirds, Rothies, Brooklyn, and millions of

28:39

other entrepreneurs. It's easy to

28:41

use. It's very functional. It

28:44

integrates with everything. It's great. If

28:46

you want to do online commerce, check out

28:49

Shopify if you haven't already because it's massive

28:51

and you should have checked it out by

28:53

now because it's the biggest company. And whether

28:55

or not you're like a giant company like

28:57

Allbirds or just a wee little merch operation

29:00

like ours, Shopify's award winning help is there

29:02

to support your success every step

29:04

of the way because businesses that grow, grow

29:07

with Shopify right now, you can

29:09

sign up for one dollar per

29:11

month trial period at shopify.com/hacked. That's

29:13

all lowercase. Go on

29:15

over to shopify.com. I mean,

29:17

let's do this slash hacked now

29:20

to grow your business no matter what stage you're in. Scott,

29:23

one more time. That's that. You

29:25

are shopify.com slash

29:29

hacked. Ransomware

29:32

supply chain attacks and zero day exploits

29:34

can strike without warning, leaving your business

29:36

is sensitive data and digital assets vulnerable.

29:39

But imagine a world where your cybersecurity strategy

29:41

could prevent these threats entirely that

29:43

right there. That's the power of the

29:45

threat locker zero trust endpoint protection platform.

29:49

Plus cybersecurity is a non-negotiable

29:51

to safeguard organizations from cyber

29:54

attacks. Threat locker

29:56

implements a proactive, deny

29:58

by default approach to.

30:00

cybersecurity, blocking every action,

30:02

process, and user, unless

30:05

specifically authorized by your team.

30:08

This least privileged methodology mitigates

30:10

the exploitation of trusted applications

30:12

and ensures 24-7, 365 protection

30:15

for your organization. The

30:19

core of ThreatLocker is its Protect

30:21

Suite, including application allow listing, ring

30:23

fencing, and network control. Digital

30:26

tools like the ThreatLocker, DetectEDR,

30:28

Storage Control, Elevation Control, and

30:30

Configuration Manager enhance your cybersecurity

30:32

posture and streamline internal IT

30:34

and security operations. To learn

30:36

more about how ThreatLocker can

30:38

help mitigate unknown threats in

30:40

your digital environment and align

30:42

your organization with respected compliance

30:44

frameworks, visit threatlocker.com.

30:47

That's threatlocker.com. I'm

30:52

Dr. Megan Sacks. And I'm Dr. Amy Sloshberg.

30:55

And we're the host of the podcast Campus

30:57

Killings. Our show covers some of the most

30:59

sinister crimes to take place on or around

31:02

school campuses, or the cases we discuss have

31:04

a school-connected theme. And with the new school

31:06

year comes an all-new second season of Campus

31:08

Killings, which will debut on September 16, 2023.

31:12

But if you want to listen to Campus

31:15

Killings now, you can binge all the episodes

31:17

from season one, available everywhere you listen to

31:19

podcasts. Is

31:24

there anything else we learned about Google search from these documents

31:26

that we haven't talked about? We talked

31:28

about Chrome. We talked about the PogoStick stuff. Is there

31:30

anything else we learned in these documents that the average

31:32

internet user might want to know about? Ooh,

31:35

average internet user is a good qualifier

31:37

there. I

31:39

think that

31:42

there are a tremendous number of things

31:44

that should be extremely interesting to anyone

31:46

who creates or

31:48

publishes content on the internet and wants that content

31:50

to do well. The number

31:52

of things that apply to you if that's

31:54

not who you are are limited.

31:58

But I will say one of those.

32:00

things that I think folks

32:02

should probably keep in

32:05

mind and be aware of is that when

32:08

Google's public representatives make statements

32:10

about how Google works

32:13

and those get quoted potentially uncritically

32:15

in the press, this

32:18

document suggests that was probably a mistake,

32:20

right? That Google's public statements

32:23

about, especially about how search works and

32:25

what they care about and what's important

32:27

and what will affect your rankings

32:29

and won't, you probably

32:31

should take those with a grain of salt because

32:35

this documentation suggests that somewhere

32:38

between dozens and hundreds of times in the

32:40

last 20 years, Google has

32:42

been directly misleading or straight

32:45

up lying about those things. In

32:49

my blog post, I urged especially

32:52

industry commentators, podcasts

32:54

like yours, folks like Kara Swishu covered

32:56

at The Verge, Search

32:59

Engine Land, all of these publications, I urged

33:01

them to take a critical

33:03

view of statements that are made

33:05

publicly by Googlers because they

33:08

are in the best interest of Google potentially,

33:11

but they're not always accurate

33:14

and sometimes directly provably wrong. And

33:16

I think we should treat them

33:18

a little bit more like we treat statements

33:20

from politicians, right? I think the job of

33:22

a journalist is don't tell

33:24

me that Jordan said it's raining and

33:27

Rand said it's sunny outside. Go

33:29

outside and tell me what the weather is. Since

33:33

you brought it up, what would that

33:35

former group, people who regularly publish content

33:37

in the internet, what's the headline for

33:39

them? Oh, God. The

33:43

headline for them is you should probably

33:45

follow and pay close attention to people

33:47

who are studying and analyzing and extracting

33:49

value from this leak because there

33:51

are hundreds

33:54

of takeaways that are

33:56

both actionable and probably

33:58

different to things. you've

34:00

done in the past or learned as best

34:03

practices. Gosh,

34:05

just today, Cyrus Shepard, who

34:07

I used to work with at Moz and

34:09

who's an expert in search, he was actually

34:11

one of Google's quality raters for

34:13

a couple of years, which is

34:16

quite interesting. But he

34:18

noted that there was a

34:20

finding from another party inside

34:22

the Google leak that

34:24

something called content effort

34:28

is scored inside the

34:30

Google leak. It's a factor that

34:32

essentially human quality raters, as they

34:34

visit websites, you know, there's

34:36

these thousands, tens of thousands of people who

34:38

work for Google through a contractor, they visit

34:40

websites and they're supposed to write about them,

34:42

right? And sort of score them and say

34:45

whether they're good or bad and

34:47

all sorts of features about them. And one of

34:49

the things they're asked to do is say, did

34:51

it look like a human being spent a lot

34:53

of effort manually to

34:56

create something uniquely valuable, right?

34:58

Differentiated and valuable to people

35:01

with this resource. And

35:04

that score now appears

35:06

to be using a large language model

35:08

AI. So it basically takes

35:10

the input from all these quality raters,

35:12

builds a metric, you know,

35:15

a sort of algorithm, and now it's

35:17

scored through an AI system. And

35:19

I found that totally fascinating, right?

35:22

Essentially you've scaled up what

35:24

quality raters used to do manually and done

35:26

it with an AI. And that's being used

35:29

according to the documents in the ranking system.

35:31

Wow. I wanna talk about the AI thing

35:33

because I think there's something really important there, but the

35:36

anonymous source, just to go back to that. Since

35:39

then, someone has come forward saying

35:41

that they are that anonymous source,

35:43

Erfanazimi, I believe is their name.

35:45

Yes. Yeah, so

35:47

Erfan decided, I think, it

35:52

was only about two or three days after the

35:54

leak was published. I think

35:56

he sort of saw that the reception

35:58

was generally favorable and not. Not

36:01

attacking, not critical of

36:03

the source or of

36:05

the credibility of the

36:07

data and he came forward

36:09

as the anonymous leaker. He's

36:13

since done a few interviews

36:15

and talked publicly about

36:17

it. He's

36:19

a real interesting guy. We don't agree on 100% of

36:21

things but Jordan, I actually find him to be

36:23

quite a lovely human being. He's

36:26

a sweet and empathetic and sensitive side and I

36:29

think a really strong sense of justice too. What's

36:32

your sense of why

36:34

he chose to leak these documents? Because as you said,

36:36

it could have gone either way. The public reception could

36:38

have been one thing. Google's reception could have been another

36:40

thing. Why do you think he chose to leak these?

36:42

To be honest, I think his

36:45

stated reasons are accurate. I've seen nothing

36:47

in his behavior before or since to

36:50

suggest that he had anything other

36:53

than a deep frustration and

36:55

anger that Google had misled

36:58

people in the field of content creation

37:01

and the field of marketing and

37:04

the technology world and press overall

37:06

but potentially even some of the

37:09

legal cases against Google. I

37:12

think he wanted the record to be

37:14

set straight and I think he felt an

37:16

obligation to make this

37:19

data available to people.

37:24

That's something I share for 17 years of my life. My

37:28

whole mission was how do

37:30

I make search more transparent? That

37:33

was Moz's whole goal. I think that's what

37:35

built that company was making

37:37

this dark underbelly. When

37:40

I started in SEO, Jordan, it

37:43

was seen as a scam. People

37:45

thought that everyone in SEO was just

37:47

sketchy and terrible and that

37:50

they were manipulative. It

37:53

took decades to

37:55

make it a mainstream marketing practice that

37:57

every company now invests in. almost every

38:00

company in the world has someone who

38:02

thinks about or works on SEO. It's

38:05

no longer seen as sketchy or spammy,

38:08

it employs millions of people worldwide, obviously

38:11

because Google sends 70% of the

38:13

internet's traffic, right? Outgoing

38:16

click traffic. And so this

38:18

is kind of a, it's a little bit

38:20

of a dream come true to be able to share

38:23

this, even though I'm out of the field. And

38:25

I think for Erfan, being

38:27

still in the field, he

38:29

really wanted people to know

38:32

the truth.

38:35

You were talking earlier about the usefulness of Google,

38:38

and I'm fascinated by this, sort of the

38:40

rise and fall of how we find information. I

38:42

think there's a feeling amongst some,

38:45

a lot of people right now that

38:47

Google is becoming increasingly less useful for

38:49

finding authentic human created information, or it

38:51

can be in certain use cases. If

38:54

you want something written by a person

38:56

without any commercial sort of motivation behind

38:58

it, basically Google is just a tool

39:01

for appending Reddit to your search post.

39:04

If you want to find some good, what

39:06

did these leaks tell us about the usefulness

39:09

of Google search for people looking to do

39:11

more than just, you know, window shop on

39:13

the mall that the internet is becoming? Oh,

39:16

you know, to be honest, I'm not sure

39:18

that the leak reveals anything

39:21

in that particular direction. Instead,

39:23

I would say that what

39:26

you'd want to look for in these cases is

39:29

when people, statistically speaking,

39:31

you know, let's take a panel of tens

39:33

of millions of people like like the Dado's

39:35

panel. And let's look

39:37

at, you know, how many searches did they

39:39

do over the last

39:41

two years each month? And

39:43

was that number growing or shrinking? And

39:46

when they do searches, where do they go after they

39:48

search? They stay on Google? Do

39:51

they click on a paid

39:53

ad? Do they click on what's called

39:55

the organic results, right? The SEO results,

39:58

which are unpaid. Do they?