Full transcript
0:00Have you ever heard of the " Nenê da
0:02Brasilia " ?
0:03A woman who defied her predetermined
0:05fate and became the leader of the drug
0:08trade in the largest city in Latin
0:10America ?
0:10Mother of the community or cold-blooded
0:12killer ? It's impossible to be
0:14simplistic when trying to define Nenê
0:16da Brasilândia .
0:17Nenê's story is intertwined with the
0:19history of Brazil . You need to know
0:21Nenê da Brasilia .
0:23This is Nenê da Brasilia , a podcast
0:25produced by Rádio Novelo . Listen on
0:28Amazon Music . Hi .
0:32This podcast is a production of Rádio
0:34Guarda-chuva , journalism for those who
0:36like to listen . If you open Google now
0:42and type " hot water , " all the results
0:44on the first page will be related to
0:47the substance heated H2O . If you open
0:50Google now and type " hot dog , " none of
0:52the results on the first page will have
0:54anything to do with a heated dog . This
0:58may seem silly , but if you think about
1:00it , it's astonishing . Just as it's
1:04astonishing for the program to know
1:06that when you type " biogetúlio , " " bio "
1:09means biography . And when you type " bio
1:12fullvest , " " bio " means biology . Or the
1:16program knows that if you type a
1:18generic term like " university , " it has
1:20to show you not only a list of higher
1:22education institutions , but the most
1:25relevant higher education institutions
1:27in your geographic location . All of
1:30this is astonishing . But today Google
1:33is so efficient and so ubiquitous that
1:35we don't think about it anymore , just
1:37as we don't think about a hydroelectric
1:39power plant when we flip a light switch
1:41when we get home . And if on the one
1:44hand we don't perceive the astonishment
1:46of this computational intelligence , on
1:47the other hand we also don't realize
1:49how profoundly it has changed not only
1:51the internet , but the world we live in
1:53and very likely the future of humanity
1:54as well . The idea that would give rise
2:05to the Google system appeared by chance
2:07in the 1990s , when a Stanford computer
2:10science student named Laurence Edward
2:12Page was looking for a topic to work on
2:15. He started thinking about a system
2:18that would allow people to comment on
2:19websites . It was quite a problem ,
2:24because large websites would probably
2:26have many comments . It would be
2:28necessary to list these comments
2:29somehow and define which would have the
2:31greatest importance . The idea of L Page
2:33to solve this issue came from academia ,
2:35because even today the way to measure
2:37the relevance of an academic article is
2:39through citations . The more an article
2:42is cited by other articles , the more
2:44important it is . The idea of L Page was
2:47to do this on the internet using
2:49something very similar to citations :
2:51links .
2:52This technology would give websites a
2:54score according to their relevance on
2:57the internet .
2:58This is the journalist , researcher , and
2:59entrepreneur Guilherme Felite .
3:01And how will he judge relevance here ?
3:03He will consider a technical term
3:05called backlink .
3:06Guilherme Felite has covered business
3:07and technology for over a decade . Today
3:10he hosts the Tecnocracia podcast and
3:12has a company called Novelo Data ,
3:14specializing in data analysis . In other
3:17words , how many times that site is
3:19linked from other sites . So you count
3:22how many links that site has to other
3:25sites and then you give a score to each
3:27of them , and each link has its own
3:30value .
3:30The importance of each site is also
3:32defined by the number of links pointing
3:34to it . If the public agency's portal
3:37has a link pointing to your website , it
3:38will have more weight than a link on
3:40that blog you created with your cousin
3:42that hasn't been updated since the
3:44early days of the internet . And the
3:48idea was good , but there was a problem .
3:51It's very easy to know where the links
3:53on a page are pointing . Just click on
3:55them . But it's much more complicated to
3:58know which links are pointing to a page
4:00. In other words , links only work in
4:02one direction . To solve this problem ,
4:05Larry Page needed help . The
4:07person who helped him get this off the
4:09ground was another guy named Seg Brin .
4:11The solution to the problem was
4:13basically to use the computer in Larry
4:15Page's dorm room at Stanford to copy
4:17the entire internet and thus map the
4:19entire tangle of links connecting all
4:21the hypertext documents . And okay ,
4:24we're talking about the internet in the
4:261990s , but even so , at various times ,
4:28this process used all the university's
4:30bandwidth , which probably didn't make
4:32his classmates very happy , but the
4:34important thing for our story is that
4:36the result was unexpectedly good . In a
4:39short time , Bren and Page realized that
4:41the idea of figuring out a way to add
4:43comments to websites had become
4:45insufficient . They had discovered a way
4:47to solve one of the biggest problems of
4:50an internet that was growing
4:51chaotically , spread across hard drives
4:53all over the planet : an efficient
4:55search system that would help ordinary
4:57people find what they needed . Something
4:59that may seem commonplace to you today ,
5:02but wasn't at the time .
5:04Page Hank's automated approach , even
5:06when it was an academic project at
5:09Stanford , was already yielding better
5:11results than Yahu Alta Vista and all
5:14those others that were available . In
5:24the 1990s , search engines ranked pages
5:28mainly according to their content . If
5:32you typed " hot dog , " they would search
5:35for pages where the words " dog " and "
5:37hot " appeared most frequently and
5:38prominently . This caused all sorts of
5:42distortions . If you typed " daily
5:44newspaper , " for example , you would
5:46probably end up on a page called " Daily
5:48Newspaper of , I don't know , Cabro do
5:50Norte . " It was unlikely that it would
5:52reach relevant pages like those of Fora
5:54de São Paulo or the state of São
5:56Paulo , simply because those pages
5:57didn't have the word " newspaper " in the
5:59title . Furthermore , as the internet
6:07grew , search engines that worked based
6:09on page content had more work to do
6:11scanning everything and finding
6:13satisfactory results . However , this
6:16same growth improved hypertext
6:18interconnection , making Brin's PG link
6:21analysis system more efficient .
6:23Then , in '96 , Google began operating
6:26within Stanford's servers . This name ,
6:29Google , wasn't the first one considered
6:32. They had initially thought of " white
6:35box , " but ended up abandoning the term
6:38because it sounded too much like " wbox ,
6:41" a slang term for the female sexual
6:43organ . In the end , a friend suggested
6:47they call it Go . It sounds similar , but
6:49it's not . If its letter G O G O L is a
6:53mathematical term that designates the
6:56number 10 raised to the hundredth power
6:59. Then they started doing some tests ,
7:01someone typed the wrong term and there
7:03was the colorful little word that has
7:05now become a verb in many different
7:07human languages . Currently , the company
7:14that controls Google is called Alphabet
7:16, but for the sake of clarity and
7:18conciseness , most of the time we'll
7:20just call it Google . In a short time ,
7:27the search engine was already using
7:29half of all the bandwidth reserved for
7:31Stanford University . So many people
7:33were doing searches and so much
7:35scraping . And then , what's the next
7:40step for two young American academics
7:42who discover something brilliant in the
7:44field of technology ? Rent a colleague's
7:50garage .
7:51Ugh . Rent a garage . Second step , please
7:56. Try to sell Google , because after all
7:59, we're talking about two guys who were
8:02researchers , and if you've ever met a
8:04researcher , 98 % of them have absolutely
8:06no business acumen to run a company in
8:09that respect , because they weren't
8:11trained for it either . So , in '99 , they
8:14offered to sell Google to another rival
8:16search engine at the time , which was
8:18Exite . They sat down with a guy named
8:21George Bell , an executive at Exite , who
8:23had been hired to be what the folks at
8:25Silicon Valley call adult supervision .
8:28The nerdy kids go there , discover
8:30something fantastic , but when it comes
8:32to dealing with money , they call in a
8:34suit with experience handling money .
8:38They sat down with two computers
8:39connected to the internet , one open to
8:41the Google system , which at the time
8:43was still called Backrub , the other
8:45open to the Exite search engine . Then
8:48they typed the word " internet " into
8:50both search engines . The first results
8:53from Exite were a list of Chinese
8:55websites , where the word " internet "
8:57spelled in English appeared floating in
8:59a sea of incomprehensible characters .
9:02The first results from Google led to
9:05pages that taught how to use internet
9:07browsers . It was basically the ideal
9:10result . And what did the adult
9:13supervisor do in the face of that
9:14technological marvel ? He wrinkled his
9:17nose . The problem ? Exite made money
9:21from ads displayed on the site , meaning
9:24people had to stay on the site to see
9:27the ads . And since Brin's Page system
9:30was very efficient and very fast , the
9:32adult supervisor at Exite was afraid ,
9:34believe it or not , of losing money
9:36because people wouldn't have time to
9:39see the ads . According to American
9:42journalist Stephen Levy in his book *
9:45Google : A Biography * , when they sat
9:47down for that meeting , the two Stanford
9:50students already had a price in mind
9:53for their invention : $ 1.6 million .
9:57Today , a mere two decades later ,
10:00Alphabet , the company that controls
10:03Google , is worth $ 1.4 trillion .
10:06Remember that 1 billion is 1000 million
10:09. 1 trillion is 1000 billion . And if
10:14that didn't give you a good idea of
10:16what the price of all the companies
10:18that sprang from Brin's first Page idea
10:20means , well , the value is almost equal
10:22to the sum of everything Brazil
10:24produced in 2020 , when our gross
10:27domestic product was $ 1.45 trillion .
10:30I'm Tomas and Averini , and episode 90
10:52of * Escafandro * has already begun .
11:01Danieli , we're going to talk about how
11:03Google invented a new type of
11:05capitalism and changed our lives
11:07forever . Before we begin , I need to
11:10remind you that Escafandro radio is
11:12proudly supported by its listeners
11:13through a crowdfunding campaign . And
11:16lately we've had a campaign within a
11:19campaign , the R $ 5 campaign . The idea
11:23behind it is to get more supporters ,
11:25even if those supporters can only
11:27contribute smaller amounts . This way ,
11:30we spread our income stream across a
11:32larger community and increase the
11:34project's stability . Because with a
11:36broad base , if one listener has to stop
11:39subscribing , or 10 listeners , or 100
11:41listeners , it has little impact on our
11:43work . This allows us to have more
11:46predictability , more planning , which
11:48allows us to continue pursuing our goal
11:51of always delivering better , more
11:53exclusive , and more relevant episodes
11:56for you . Of course , besides helping the
12:03project grow , improve , and reach more
12:05people , when you become a supporter ,
12:06you get a series of perks . You'll have
12:09access to an area with some complete
12:11interviews , you can listen to
12:13unreleased episodes a day in advance ,
12:15you can record the ending of an episode
12:17, and with higher donations you'll
12:19receive autographed books or a
12:20customized mug with the Escafandro
12:22radio logo . In short , if you like the
12:26project , those R $ 5 won't make a
12:28difference to your monthly budget ; come
12:30join the luminous humans who fund the
12:32project . To do this , the process is
12:35quick and safe . Just go to the Catarse
12:37website . Choose the amount you want or
12:43can support . Catarse.me/escafandro . And
12:50if you're already among luminous humans
12:52, like Juliana Trajano , Paula Gabriela
12:55da Silva , and Marcelo Vilas Boas Mota ,
12:57thank you very much for that . The
13:06meeting with the Exit team was kind of
13:08the last straw for Bram and Page , who
13:10realized they were going to have to run
13:12that company alone . And that basically
13:15left two problems ahead .
13:17The first was finding an executive who
13:19could lead this process ,
13:20an adult supervisor , remember ? ... and
13:22leave the field open for the two
13:24founders to do what they wanted , which
13:27was product development .
13:28The first idea they had was so
13:30outlandish that it ironically
13:32highlighted the importance of having an
13:35adult supervisor . For Page and Brin ,
13:38the only human worthy of their Google
13:41CEO was Steve Jobs . Then people said , "
13:45Look , it seems this guy is kind of
13:47satisfied with his job as the owner of
13:50Apple . " Even so , it took a lot of
13:53convincing to dissuade them from the
13:55idea . In the end , they ended up hiring
13:58a guy named Eric Schmitt , who , in
13:59addition to being the CEO of a company
14:01specializing in networks , understood a
14:03lot about engineering and communication
14:05.
14:06And Eric Schmitt was going to have a
14:08role in structuring some points , some
14:10issues that don't necessarily involve
14:13the technological side . And that's
14:15effectively what Eric Schmitt did , and
14:17he did it brilliantly . Google only
14:19grows to the size it does because Eric
14:22Schmitt was the person who knew how to
14:24create these divisions , how to lead
14:26these people , talk to governments , talk
14:29to candidates , and so on .
14:31Right ? First problem solved . Second
14:33problem , Guilherme Felite , please . The
14:36second was to guarantee a source of
14:37revenue , because after all , you had a
14:39growing volume of people visiting
14:41websites , you had a growing volume of
14:43requests being made , but you still
14:44weren't generating money . You were
14:46basically burning through investment
14:48capital in servers , employees , and
14:50other things like that .
14:52One of the first investors to pour a
14:54huge amount of money into Google was
14:56none other than Jeff Bezos , founder of
14:59Amazon . Which just shows us that the
15:01world is small , but Silicon Valley is
15:03even smaller . In any case , however rich
15:06the investors were , they wanted to see
15:09results . And the first idea they had
15:11for this was the same idea that all
15:13successful website owners have : placing
15:17ads . And they started doing this in the
15:20most traditional way possible , in a
15:22process that even involved a formal
15:24request from the advertiser to be sent
15:26by fax . Again , according to Steven Levi
15:29, and I'm going to use several stories
15:31from his book , Sergei Brin , who had a
15:32reputation for being a " Semoquirana " ( a
15:33derogatory term for someone who is
15:35easily manipulated or misguided ) , found
15:36that fax thing a bit strange . And the
15:38head of sales had to guarantee that yes
15:40, they would have enough ads to cover
15:43the cost of an apparent fax . And for a
15:48while , this scheme worked with Google
15:49sales executives , taking clients out to
15:51dinner , explaining the tool and showing
15:53an example of a sponsored link that
15:55appeared at the top of search pages at
15:57the time , differentiated by a yellow
15:59background . And the system did work ,
16:03but it worked poorly . They were using
16:06an analog tactic created for billboards
16:08and print newspapers for a digital
16:10medium that was completely different .
16:13Until , in October 2000 , Google took its
16:15first step towards trillions . They
16:19realized that the approach had to
16:21change because , on the one hand ,
16:22searches were unique and highly
16:24segmented . On the other hand , the cost
16:27they had to show an ad was very low .
16:29And putting these two ends together ,
16:31they saw the opportunity to create a
16:33little tool that at first seemed naive ,
16:36but that changed the paradigms of
16:38advertising . A little tool called ADWs ,
16:43from ad , Word from word .
16:47So , basically , you want to link your
16:51product to search results . You're not
16:54going to do what some search engines
16:57like Google did , which was basically
16:59accepting money to place a result as
17:01organic . What you're going to do is
17:04place paid results on the page
17:06indicated as paid results .
17:09The first stroke of genius with AdWords
17:11was creating a DIY system where the
17:13client basically needed to provide a
17:16valid credit card number . The second
17:18was offering a segmented environment
17:20that allowed for very low prices . The
17:23idea was that they could attract small
17:25businesses that had never thought about
17:27advertising in the digital advertising
17:29market . And who would be the first
17:31advertiser in AdWords history ? Google ,
17:34of course . In October 2000 , they
17:36published the simple and brilliant
17:39advertisement : " Got a credit card in 5
17:41minutes ? Advertise on Google now . "
17:45Although the ad was only shown to a
17:47portion of the audience within minutes ,
17:50someone had already registered , and
17:53less than half an hour later , another
17:55user who typed " Live Lobsters " into
17:58Google search saw an ad from a company
18:00called Lively Lobsters . From there ,
18:07this system evolved , reaching a
18:09sophisticated model where the price of
18:11each ad was determined in instant
18:13virtual auctions .
18:15These models changed significantly in
18:17the early years of Edwards ' launch ,
18:20where ads were more distinct from
18:23organic results . Recent research
18:25suggests that the average user , when
18:28accessing the site , cannot distinguish
18:30between ads and organic results . Uh-huh
18:33.
18:33Perhaps there's a somewhat standard
18:35confusion there because when the user
18:37clicks on the ad , Google earns . When
18:40they click on organic results , Google
18:42doesn't .
18:43In any case ,
18:44the business is so successful that it
18:47remains its main source of revenue to
18:50this day . ( Quote from Sergei Brin about
18:53the moment the money started coming in .
18:56) Honestly , back in the early days of
18:59the internet , I felt like a fool . I had
19:03a brand new internet company , just like
19:04everyone else . It wasn't as profitable
19:07as others , and things were quite
19:08difficult . When we started making money
19:11, I felt like I had a real business .
19:13When the guys realized it was
19:15incredibly successful , they said , " Okay
19:17, but we have such a good system that
19:19we don't need to restrict it to just
19:21our own websites . We can take our ads
19:24to websites on other platforms that
19:27aren't managed by us . "
19:29In 2003 , Google launched AdSense . While
19:33AdWords sold ads within its own program
19:35, AdSense inserted those ads on
19:37websites that partnered with Google . If
19:41you had a blog about interior design ,
19:43wrote a post about the importance of
19:44flowers for interiors , and sold an ad
19:46through AdSense , it was very likely
19:48that your readers would stumble upon an
19:50ad box from a flower shop .
19:51Basically , you break down a part of
19:53AdWords , but you kind of pulverize it
19:56across billions of websites on the
19:57internet . Even today , it generates
20:00billions of dollars annually for those
20:03guys , but search is still the biggest .
20:05And here , maybe you , my communist
20:07listener , are a little bothered because
20:09you clicked on a program about a
20:11technology company and you're listening
20:14to us talk about advertising . Well ,
20:17that's perhaps the most important point
20:19of the episode , because given the
20:21financial results of AdWords and
20:23AdSense , Google's vocation has changed .
20:26It has ceased to be primarily a
20:28technology company and has become
20:30primarily an advertising company .
20:32Google and Facebook , they have no shame
20:35whatsoever in saying that they are
20:37advertising companies to this day . None
20:40. I interviewed Eric Schmitt in 2011 ,
20:42something like that , and he said it in
20:45no uncertain terms : " We are very proud
20:48to be an advertising company , because
20:50what pays the bills , pays for the
20:53champagne pools , the houses in the Alps
20:55and all that stuff , is the volume of
20:58advertising that was directed from
21:00other sources in the 80s and 90s to
21:03Google and Facebook servers . " And so ,
21:05if the goal was to be an advertising
21:07company , Google was going to do
21:09everything possible to be the most
21:11efficient advertising company in the
21:13world . And to do that , it needed to
21:15know everything it could about the
21:17potential consumers it had access to .
21:19And to do that , it needed to have more
21:22access to more potential consumers . And
21:25then Google starts buying and
21:27developing tools that , on the one hand ,
21:30greatly help people's lives , but on the
21:33other hand , also compile data about
21:36that person's life in a way that allows
21:39you to create a profile of that person
21:42and link that data to know that the
21:44person who's looking for nonsense might
21:47also be interested in aviation , reading
21:50historical romances , and riding their
21:53bicycle on Sunday mornings while the
21:56sun is out . You gradually build an
21:59advertising system that doesn't just
22:01rely on search intent and will have a
22:04series of other data sources that
22:06complement this profile . That's how it
22:10becomes a company .
22:12This is sociologist Sérgio Amadeu da
22:14Silveira ,
22:14who is able to have data on who has his
22:18email address , data on who is searching
22:22, and data from this tracking network .
22:26He is a professor at the Federal
22:28University of ABC and researches
22:30digital networks and the political
22:31implications of artificial intelligence
22:33. When you arrive at a news website ,
22:36the banner will be the banner that
22:38relates to the search you did on Google
22:41, you understand ? You were searching
22:43for a refrigerator , so the banner ad
22:45will be for a refrigerator . This
22:47started to change advertising .
22:48The decisive step in this direction
22:51happened in 2007 when Google spent $
22:543.1 billion on the biggest acquisition
22:58in its history , which wasn't YouTube .
23:01DoubleClick , Google bought this company
23:03. If you're not in the tech field ,
23:05you've probably never heard of this
23:07acquisition or this company , mainly
23:09because DoubleClick is the kind of
23:11company that doesn't like to be in the
23:13spotlight , preferring to operate in the
23:15invisible parts of the internet . It's a
23:18company specializing in a simple
23:20digital trap called a Cookie .
23:23It's a small digital file called a
23:25Cookie . By placing this Cookie on your
23:28computer , when you're accessing a
23:31website that has an agreement with the
23:33company that placed the Cookie , that
23:36company knows it's you because you have
23:41that numbered Cookie on your computer ,
23:43in your browser .
23:44So it identifies you , and then what
23:46happens ? Your entire browsing history ,
23:49or almost all of it , is being tracked :
23:52the time you logged in , how long you
23:55stayed , whether you went to another
23:57page on the site , how many pages you
24:00viewed , whether you left the site for
24:02another site that's part of the cookie
24:05network , it will also identify you . So
24:08you practically have a digital tracker .
24:11And then there's another purchase
24:13that's absolutely fundamental , but
24:15which kind of went under everyone's
24:17radar at the time , which in 2007 they
24:19paid an undisclosed amount to a startup
24:22called Android . Android becomes Android
24:24, an operating system , and then
24:25everyone says : " Wow , this is brilliant ,
24:27they're going to have the Windows of
24:29mobile phones . " Android , in the end , is
24:32an inexhaustible source of data about
24:34people's habits .
24:36My phone is Android . Inside I have
24:38Google , Chrome , Gmail , Google Photos ,
24:41Google Drive , Google Lens , Meet ,
24:44YouTube , and Google Maps . Nine Alphabet
24:47apps that are constantly collecting
24:49data about my life . Knowing that you
24:52need to connect your Google account on
24:54Android , on YouTube — I mean , most
24:57people connect their accounts — it's
24:59very easy to track people across
25:01different platforms and create a
25:03profile that will give you much more
25:05than just the certainty of nonsense or
25:07Britney Spears . The algorithm is
25:10complex enough to use clicks you've had
25:13on blogs or YouTube searches , or a
25:16series of other data inputs , to show
25:19you ads that the algorithm deems
25:21relevant in different ways as well . It
25:24doesn't have to be a sponsored link
25:26that will appear in the search results .
25:28It could be an ad on Google , on YouTube
25:30, it could be something that will
25:32appear on some other website that isn't
25:35even Google's because of AdSense .
25:37And perhaps the biggest problem in all
25:38of this has to do with the fact that
25:40Google knows how to create products
25:42that become indispensable to our lives .
25:44Even those of us who know about these
25:46dangers , we look at some of the
25:47products and say : " Wow , this is so good
25:49that for a few seconds you forget about
25:51that privacy problem . " I think the main
25:53example for me is Google For . Google
25:56Photos solved a problem that everyone
25:58had before everyone else . But anyway ,
26:01if you have an Android and you want to
26:03be a little shocked , search on Google
26:06itself , including the amount of
26:08information Google has about your
26:10Android .
26:11The first link I chose to click took me
26:13to an article on the website Tudo
26:15saying that Google collects 900 data
26:17points from Android every day , even if
26:20the user does nothing . But in Google's
26:23defense , I have to say that the first
26:25suggestion from the search was a page
26:27from them talking about data and
26:29privacy . This led me to my own Google
26:32account page , where I discovered that
26:35they collect my data and no less than
26:3746 applications . I also discovered that
26:40you can click the button and export
26:42this data . What I obviously did was , to
26:44which Google replied : " OK , that's fine ,
26:47but it will take a while , probably
26:50hours , maybe days . " In the end , it
26:52didn't take that long . In less than an
26:54hour my files were ready to be
26:56downloaded . I downloaded them and ended
26:59up in a bureaucratic anticlimax lost in
27:03a sea of folders and subfolders full of
27:06files . Most of them in extensions that
27:09my computer programs couldn't open . It
27:13has an identifier number for each
27:15person . And it can have this duplicated
27:18because , for example , it hasn't
27:20identified you yet . So it keeps
27:23collecting this data . When it knows
27:25it's you , it probably combines this
27:28data with what it already has of your
27:31accumulated profile , which is data
27:34whose size we don't even know in
27:37kilometers , in bytes , but it's a lot .
27:42It collects data insistently all the
27:44time . All the time , because our
27:47identity for it is fluid , right ? It
27:49doesn't know I'm a Corinthians fan . So ,
27:51every now and then it learns something
27:54and keeps changing your profile ,
27:56increasing the possibility of
27:57assembling samples . In fact , it doesn't
28:00... They sell the data , right ? They sell
28:01access to companies that want to reach
28:04a specific profile .
28:06The company also doesn't know what it
28:08has , what data is being used to sell
28:10its ads , right ?
28:11There are different levels of companies
28:14with access and different structures ,
28:17right ? And so , by becoming this
28:21planetary digital population devouring
28:23data ,
28:24it will inaugurate , according to
28:27researcher Shosana Zubov , what would be
28:30a data-driven capitalism , right ?
28:34The philosopher and professor emerita
28:36at Harvard Business School , Shosana
28:38Zubov , is the author of a book called *
28:40The Age of Surveillance Capitalism * .
28:42Quotes for her definition of
28:45surveillance capitalism : One , a new
28:47economic order that claims human
28:50experience as free material for
28:52disguised commercial practices of
28:54extraction , prediction , and sales . Two ,
28:58a parasitic economic logic in which the
29:01production of goods and services is
29:03subordinated to a new global
29:05architecture of behavior modification .
29:07Three , a disastrous mutation of
29:10capitalism marked by concentrations of
29:12wealth , knowledge , and power without ...
29:14Precedents in human history . Four , the
29:18structure that underlies the
29:20surveillance economy . Five . A threat as
29:24significant to human nature in the 20th
29:26century as industrial capitalism was to
29:28the natural world in the 20th century .
29:32Six . The origin of a new instrumental
29:35power that claims dominance over
29:38society and presents surprising
29:40challenges to market democracy . Seven ,
29:44a movement that aims to impose a new
29:46collective order based on total
29:48certainty . Eight , a new expropriation
29:52of critical human rights that can be
29:55best understood as a coup from above , a
29:58destruction of the sovereignty of
30:00individuals .
30:02According to Xanasubof , Google has
30:04become the pioneer , the discoverer , the
30:07developer , the experimenter , the main
30:09practitioner , the example , and the
30:11center of diffusion of surveillance
30:14capitalism . She compares what Google is
30:16in this system to what Ford GM was to
30:19managerial capitalism based on mass
30:21production . It now realizes that it can
30:24be a platform in various markets . What
30:26is a platform ? It is a data company
30:29because it wants to bring together the
30:31supply of those who have something to
30:34offer with the demand of those who need
30:36it . Think of Uber . Uber Strictly
30:39speaking , you don't need to own a car
30:40at all . Uh-huh .
30:41It brings together those who need to
30:42work with a car and those who need to
30:44get around in urban spaces . That's it .
30:46Just like Netflix , you don't need a
30:47film production team . You
30:49don't need one . If you want to , you can
30:50, right ? But if you don't want to , you
30:52can't . But look , it dominates the
30:54supply and demand of a segment or the
30:56entire market . And it's not an
30:58intermediary , because it , for example ,
31:00the search engine , has a lot of
31:02websites offering information , a lot of
31:04people needing information , there's
31:06supply and demand . But when it delivers
31:09the information to the demand , it
31:12prioritizes , it interferes , it
31:15processes the information . It's like
31:18Uber today . If Uber wants to send all
31:20the drivers to the north side of a city
31:22, it imposes . It has the data , it has
31:25the information , it has the access to
31:27manipulate things . An example of Uber
31:30is manipulating prices . Yes .
31:32When you're , for example , late at night
31:35. There
31:36may even be drivers available , but it
31:39will manipulate that price . Because he
31:41knows you need that car at that moment ,
31:43right ?
31:43Perfect . He knows , and often he already
31:45knows your pattern .
31:46Here it's important to make a caveat .
31:49This hypothesis I've put forward is
31:51narrated by several users , and there's
31:53a lawsuit in the United States accusing
31:55Uber of manipulating prices . But just
31:57like in the case of Google , the
31:59algorithms are closed , and nobody knows
32:00exactly what happens on the hard drives
32:02scattered around the planet . One of the
32:05digital models , and the one that makes
32:07the most money and is most successful
32:09today , is this ; it's this model of data
32:12collection , data processing to be able
32:14to interfere . I call it behavior
32:16modulation because it's not ... uh , when
32:19we use the term manipulation , it sounds
32:21like we're talking about something that
32:24has no alternative , something that old
32:26functionalist sociology used to talk
32:28about mass communication , right , that
32:30you just have to say false information
32:33and everyone will believe it . It's not
32:35quite like that , is it ? I say
32:36modulation because there's a field that
32:38it controls over you , but if you want ,
32:40you can get out of it , right ? You have
32:43your own baggage . Cultural , its
32:45information , right ? But it tries to
32:48control your gaze . The main thing these
32:51platforms do , especially communication
32:53platforms , is control our gaze , because
32:56they don't create the content , right ?
32:58Google itself , on YouTube , it sets up
33:00the first page you'll reach , it's an
33:02algorithmic system that builds it .
33:04Uh-huh . Now , when we talk about this ,
33:07about behavior modulation and about
33:09this thing that Americans call a "
33:11rabbit hole , " right , where you enter
33:13and become increasingly immersed in a
33:16biased way of thinking that creates
33:18these far-right bubbles and so on ,
33:20we're used to associating this with
33:22YouTube , right , which is part of the
33:25Google group . Does this type of
33:27modulation happen in other spaces ,
33:29other Google platforms ?
33:30It does modulate in the search engine ,
33:32right ? Certainly . Uh-huh . And the most
33:35curious thing is , when I click on the
33:38link it gives us , it knows it's me
33:40clicking and it improves its neural
33:43network , right ? Because it mainly uses
33:45neural networks there , which is a
33:48specific type of machine learning
33:50algorithm , right ? It creates Models to
33:54receive us and meet our interests ,
33:56right ? It also modulates when it now
33:59has these personal assistants , right ?
34:02The smart speaker , it also starts to
34:06try to narrow reality so that we have
34:10the options it gives us . Making a pun
34:13here , it's not an augmented reality
34:16that these platforms do , it's a
34:18diminished reality , right , for the
34:21advantageous options they offer us ,
34:24right ?
34:24What you're saying is very important ,
34:26because it has to do with search , is
34:27that right ?
34:28Search . But that's how this whole world
34:31of platforms is generally working ,
34:33right ?
34:33But I think this is extremely important
34:36for us to talk about , because it
34:39perverts the origin of Google , right ?
34:42Because Google's objective was to make
34:43an efficient search .
34:44Yes . Now it's not .
34:45And from the moment you do a search
34:47that you have the objective of
34:49collecting data and selling ads , that
34:51search ceases to be efficient for the
34:53purpose of being a search , right ?
34:55Perfect , perfect . It loses quality , but
34:58Google knows that there's even the verb
35:01" to google , " right ? Go there and do a
35:04Google , right ? This eagerness of
35:12platforms to collect data is so intense
35:15that they are constantly inventing new
35:17ways to collect this data : the like
35:19button , the emojis you put in the
35:21comments , those hundreds or thousands
35:23of unknown people who , all of a sudden ,
35:25because some programmer created a shiny
35:27little square on your screen , you
35:29started calling friends .
35:32Knowing the number of friends I have is
35:36impossible . But they invented a little
35:39figure that's a link that says : " Ah ,
35:41you're a friend , you're an acquaintance
35:44. " You ... Then a bunch of metrics that
35:46were impossible before started to
35:48appear . But notice : these are metrics
35:51and data that are invented , created ,
35:54and what interests them is what grabs
35:58attention , what creates spectacle . I
36:01don't believe that what YouTube does is
36:04privilege radical content , as they say ,
36:08no . It privileges spectacular content ,
36:11what grabs attention , what sparks
36:14curiosity , the absurd ,
36:16and what generates engagement too ,
36:17right ? That's
36:18it , but that will generate engagement .
36:20If you're watching a scene It's absurd ;
36:23you get indignant , some people don't ,
36:25but he knows who gets indignant and who
36:28doesn't .
36:28He knows because many people react ,
36:30comment , click the thumbs up , click the
36:33thumbs down , go back to watch specific
36:35segments again , forward it to friends .
36:38But according to Professor Sérgio
36:40Amadeu da Silveira , it's likely that
36:42these platforms also have other ways of
36:44knowing . In the United States patent
36:46offices , for example , there are several
36:49software registrations that attempt to
36:51discover the user's mood through more
36:53subjective signals , such as typing
36:55speed . We , in fact , already talked
36:58about this here in episode 14 , " Your
37:01Brain on Instagram . "
37:03People say , " Oh , they put us in bubbles
37:04. " I don't really like that term "
37:06bubble " because it's misleading ,
37:08because what they put us in are samples
37:10. For example , if a company wants to
37:13reach a guy of a certain age who voted
37:15for Lula or who didn't vote for Lula ,
37:18who is a Corinthians fan or a Palmeiras
37:20fan , it will negotiate that , that
37:22audience to be " The target segment is
37:25being targeted , you understand ?
37:27And then the algorithm will probably be
37:29monitoring that person and seeing if
37:32it's getting it right . Then it puts you
37:34in that sample if you clicked on the ad
37:36; otherwise , it removes you from that
37:37group and tests again to put you in
37:39another one . It must be saying that all
37:40the time . That's
37:41right . What you said is very important .
37:43You know why ? People don't get it . ' Oh ,
37:46but how does it know I'm doing it ? '
37:48It's because every click I make on the
37:50internet generates a digital trail . And
37:54this trail is captured ; it's valuable .
38:00And you
38:01know what's needed to store this
38:02digital treasure ? Space and natural
38:07resources . It needs to store it in
38:10structures that are data centers , right
38:12? These structures are gigantic , right ?
38:17And they are already an environmental
38:19problem , right ? They consume a lot of
38:21energy , a lot of water . Because in a
38:22data center , for example , at Google ,
38:25some must have more than a hundred
38:27thousand machines , more than 100,000
38:30computers , self-processing servers . And
38:32that consumes an absurd amount of
38:34energy . " This consumes a completely
38:37absurd amount of water , right ? One of
38:41their data centers consumed 1/3 of the
38:44water consumed by a city in the United
38:48States , where they had their data
38:51storage center , in a single year . In
38:552021 , a Google data center in Cidade de
38:58Deus , Oregon , consumed over 1 billion
39:00liters of water per month . 1/3 of
39:04everything the city consumed . This
39:06absurd amount of water is used to cool
39:09the machines that store all your
39:11never-deleted emails , your photos of
39:13kittens in the cloud , but also an
39:15endless number of files that say you
39:18like nonsense , read gossip about Brit
39:20Spears , and just adopted a dog . At the
39:23beginning of the 20th century , we used
39:25to say , and I used to say it too , that
39:27digital technology would reduce
39:29environmental impact because it would
39:31eliminate the use of so much paper . We
39:34said these things somewhat naively .
39:37Digital technology does , in fact , allow
39:40for a certain reduction in
39:41environmental impact , without a doubt .
39:44But the data-driven business model ,
39:46which ... I call it datafication , and
39:48this amplifies the environmental impact
39:51. Why ? You have to store this vast
39:54amount of data , and the data that
39:56Google has , it won't pass on to Amazon ,
39:58or Microsoft , or Facebook , because they
40:00are competitors , they are platforms ,
40:04and not even to the IBGE ( Brazilian
40:05Institute of Geography and Statistics ) ,
40:06which could , for example , create a
40:07public policy to improve something ,
40:08right ? No , it won't go through the IBGE
40:10; on the contrary , it will take the
40:12data from the IBGE , from the
40:13municipalities , from everyone , from the
40:15universities , and store it with them .
40:25This data can be used for various
40:27purposes . It can even be used to
40:29manipulate elections . There is no
40:32longer any doubt , for example , that
40:34both Donald Trump's campaign and
40:36Russian digital militias used social
40:38networks to modulate behavior and
40:41manipulate the American electorate in
40:432016. But by all indications , the
40:45biggest target of Big Tech is the same
40:48target of Ford , GM , and all companies
40:50in the civilized capitalist world : your
40:53wallet .
40:57The internet serves this purpose ,
40:58Guilherme Felipe .
40:59People think the internet is for
41:01conversation , for ... Finding a partner
41:03to go out for drinks . No. It serves to
41:04sell you something you don't need . In
41:132001 , Google hired a guy named Bill
41:15Campbell , a former American football
41:18coach who had become a tech executive ,
41:20was Steve Jobs ' best friend , and was
41:22something of a legend in Silicon Valley
41:25. In practice , his role at the company
41:27would be that of an executive coach .
41:30And as often happens with coaches ,
41:32Campbell decided to gather some of the
41:34company's most prominent employees for
41:36a group dynamic . The idea was to
41:39discuss the company's values . But among
41:42these employees was a guy named Paul
41:45Buite . He had already participated in
41:47similar things at similar companies and
41:49thought that's what employees generally
41:51think of group dynamics . A waste of
41:54time . Years later , the people told
41:57journalist Steven Levy about his
41:59reaction to such a group dynamic . I
42:02suggested something that would make
42:04people uncomfortable , but that would
42:06also be interesting . It came to my mind
42:09that " be evil " would be an interesting
42:11and easy-to-remember statement . And
42:14people laughed , but I said , " No , I " I'm
42:17serious . " " Don't be evil " in
42:19Portuguese means " não seja mal " ( don't
42:22be bad ) . And that didn't sit well with
42:26the coach , but Poight gained the
42:28support of one of Google's top
42:29employees , Amit Patel . And whenever
42:32they saw a whiteboard lying around ,
42:34they would go there and write the same
42:36phrase : " Don't be evil . " And little by
42:38little , that idea got into the heads of
42:40the Googlers and became something of a
42:43secret motto for the company . And then ,
42:45when Google took the decisive step of
42:47starting to sell company shares on the
42:49stock exchange , Larry Page and Sergei
42:51Brin made a decision that would haunt
42:54them forever . When you go public , you
42:57need to send a document called DS1 to
42:59the SEC , which is the American
43:00equivalent of the CVM ( Brazilian
43:01Securities and Exchange Commission ) ,
43:03which is basically a general overview ,
43:04right ? How your business is doing , how
43:06you make money , who the founders are ,
43:08what your history is . And Google put
43:11Beevil's phrase on its DS1 .
43:13A good portion of Google's executives
43:15were obviously terrified that the
43:17company would be seen as naive by the
43:19Wall Street executives . Street . But the
43:21truth is that at the time nobody paid
43:23much attention to it . It was the first
43:25time Google had opened up its finances ,
43:27and with that amount of money coming in
43:29, they could have printed the S1 on
43:31colorful , scented stationery , and
43:33everything would have been fine for the
43:35Wall Street suits .
43:37And that phrase stuck .
43:39When you say it stuck , what do you mean
43:41by that ? When you talk about the
43:43environment , people in the technology
43:45market , in journalism , know that we're
43:47talking about Google , but we don't have
43:49any other big company with any of these
43:51emblematic phrases that stuck . Facebook
43:54doesn't have that , UBA doesn't have
43:57that , Google maybe a little because of
43:59that thing of trying to approach it in
44:01a different way . I think maybe Facebook
44:03approached it , Zuckerberg approached it
44:06in a more , quote-free , cynical way ,
44:08like , this is this , Iris , Iris . But the
44:10fact that the guys at Google put that
44:13in the S1 is also a way of making
44:16people remember . They are They were
44:18criticized for it because they put
44:19themselves back there . If they hadn't
44:21put that Dom Bevil thing back there ,
44:23people would certainly have shrugged it
44:25off . Oh , they were bad , vile , from the
44:27beginning . I think people would have
44:30kind of ignored it . And I suspect Page
44:33and Bren knew they were going to be
44:36criticized for it , as a way to stay on
44:39track .
44:39You think it was kind of a message to
44:42them from the future , you know ? We know
44:45what happens to people in that
44:46situation .
44:46Don't get lost .
44:47Don't get lost along the way .
44:48As the mutants would say , don't get
44:50lost out there . We're going to leave
44:53this bar here for you up ahead , if you
44:56feel tempted to cross it , break this
44:58bar , you're going to suffer the
45:01consequences , because we want to make
45:03sure it's going to be harder . That's it
45:06. I think that was it . Let's go back a
45:09little to that story of Page and Bren
45:11as two researchers with ethical
45:14concerns that academia , especially
45:16academia in places of excellence like
45:19Stanford , instills in us . There was
45:22that The fear is that you'll grow up
45:26and ignore certain ethical issues that
45:29shouldn't be ignored . In a whole story ,
45:33when you have someone who's very
45:36successful , you go from one role to
45:39another , in this capital , let's say ,
45:42and become an executive .
45:44That reference was very , very much 40
45:46years ago . It
45:47was very much 40 years ago . If you're
45:49young , if you still have hangovers that
45:51only last a day , congratulations ,
45:53Google " Nelson da Capitinga " so Google
45:55puts " Nelson da Capitinga " in your
45:57profile . But this arc begins with
45:59something kind of dreamy . And then when
46:02everything works out , you kind of
46:04become a pastiche of the same , right ?
46:06You become everything that , in a way , I
46:08think was hated , which is what
46:10inevitably happened . Then it became a
46:13chirp , it became a strategy , it
46:15became a chirp , but at the same time ,
46:17for a long time , we can assume that
46:20these guys were doing things that
46:22wouldn't fit within that framework . I
46:25mean , they were being bad , believing
46:27for a long time that they were still ...
46:29Being nice . I think a lot of people
46:31still believe they're being nice to
46:32this day . Do
46:33you think so ?
46:34No , I have no doubt . A lot of people
46:35still believe it
46:36within Google , you said
46:37within those companies , and even people
46:39who are in the surrounding areas , you
46:40know ? When your salary depends on you
46:42believing in that , without a doubt ,
46:44there are a lot of people who will dive
46:46in headfirst .
46:48Of course , it's important to consider
46:51that it's always easier to think you're
46:53going to be nice when you're in your
46:55early twenties and immersed in the
46:57technological world of 2004 , not 2023.
47:00Back then , it was also difficult for us
47:02to understand the magnitude of the
47:04problems we would face a decade or two
47:07later . I think people who started
47:09sounding some alarms back then were
47:11seen as alarmists , as annoying . So ,
47:13there was a Belarusian philosopher I
47:15quote quite frequently on my podcast ,
47:18Euggen Morozov , who wrote a book in
47:202008 , 2009 called " The Net Disillusion ,
47:22" in He was talking about how , contrary
47:25to everything that was being propagated
47:27at the time , the internet would be a
47:30great ally of dictatorships and
47:32autocracies and so on . And 15 years
47:34later he was absolutely right . We're
47:37feeling that now . And from that
47:40perspective , when you look in the
47:42rearview mirror , I see these scenes and
47:44it seems like such a silly , cheerful
47:46thing , with so many terrible things
47:48that were going to happen and the
47:50market being super Pollyannaish back
47:52there thinking that only good things
47:54were happening . Anyway , it's easy to
47:57judge when you're looking at things in
47:59the rearview mirror , but I firmly
48:01believe that they believed in Dombo's
48:03story . The main problems that Google
48:07shows today are the fact that you
48:10enable and amplify racist , anti-Semitic
48:14, anti-democratic discourses , public
48:17lynchings , and so on . I think at no
48:20point back then did they imagine that
48:21this could happen , or maybe they
48:23thought about it , but said : " Damn , this
48:24will be solved quickly , the balance is
48:26always better , it always leans towards
48:27the good side . "
48:28On the other hand , there's
48:30a great deal of self-deception in the
48:32tech market . They're clearly convinced
48:36they're not doing anything wrong . " Oh ,
48:39maybe there's one or two things , we can
48:40make a small adjustment , but I'm doing
48:42something good . " And I've seen this
48:44with executives from other tech
48:46companies . It's kind of an echo chamber
48:49. The employees , the directors , the
48:52presidents , everyone's locked in the
48:55same environment , kind of repeating the
48:58same logic . This isn't just a little
49:01ironic , it's very ironic .
49:03One of the main problems introduced by
49:05this social media logic is trapping
49:07people in bubbles
49:09that validate their thoughts .
49:12If you want to know more about this
49:14phenomenon , I suggest you listen to our
49:16trilogy " Depths of the Network , "
49:18episodes 31 , 32 , and 33.
49:20We have a growth of anti-vex , Nazi ,
49:22anti-Semitic people — you can put
49:24whatever crime you want in there —
49:27because of these bubbles . Well ,
49:29ironically , there's also the tech
49:32bubble where you find validation for
49:35any nonsense you do under the guise of
49:38innovation . I've heard excuses that
49:41sound like something a 12 - year-old
49:42would say : " No , they're mad at you
49:44because you're awesome , that you're
49:46innovative and all that . " Besides
49:48appearing on the S1 for the stock
49:50market launch , the phrase " Don't be
49:52Evil " also opened the company's code of
49:55conduct . Until 2018 , trying not to
49:57cause a stir , the phrase disappeared
49:59from there . It ended up at the bottom
50:02of the page , but in an updated version .
50:04In a free translation it became
50:06something like this : And remember ,
50:08don't be evil . And if you see something
50:11you think is wrong , speak up . This
50:13change may seem subtle , it may seem
50:15like an adaptation to our times , a sign
50:17of maturity . At the same time , I keep
50:19thinking about the weight of this
50:21decision . Because okay , the world is
50:23nuanced . It's difficult to talk about
50:25good companies or bad companies without
50:27seeming naive . But once the phrase is
50:30there , that card of naiveté is already
50:32open . And removing the phrase has a
50:35huge impact . Because however silly it
50:37may seem to frame the discussion in
50:39those terms , by doing so , Google shows
50:41that being good isn't so important
50:43anymore . It seems to be publicly
50:46embracing its own status as an evil
50:48company . I asked Guilherme Felite about
50:51this .
50:52IP strategies work up to a certain
50:56point . From the moment your code of
50:58conduct doesn't match the coverage
51:00you're getting , you'll have a reverse
51:03effect where people will basically hold
51:05you accountable for it . Then that trap ,
51:08in quotes , that Page and Bre 2000 set
51:10for themselves falls again . It's stuck
51:13so much to the image that it's going to
51:15have to stay there . And then every news
51:17story that comes out revealing that
51:20Google took questionable action will
51:22bring back that " Okay , fine , don't be
51:25evil , alright " thing . Let's remember
51:28that Google had a researcher , Tim
51:30Jeburé , who was one of the first to
51:33warn about the ethical dangers of large
51:36language models long before launching
51:38GPT chat . She was fired from Google for
51:42claiming that there were serious
51:44problems in the development of this
51:46technology and in her work . She was
51:49responsible for the ethical aspects of
51:52LLM , and this dismissal gave her a
51:54springboard to become known as one of
51:56the leading critical voices , not only
51:58regarding LLM , but also the way
52:00technology companies , represented by
52:02Google , abandoned certain ethical
52:04concepts to jump on the next bandwagon .
52:07And the fact that they originally
52:09removed " Don't Be Evil " in 2018
52:11coincides with the moment when this
52:14wave of optimistic coverage of large
52:16technology companies began to reverse .
52:192018 was already the second year of the
52:22Trump administration , and you already
52:24had a series of reports showing how
52:27social networks had been abused , in
52:29quotes , by the Trump campaign , and the
52:32investigation into Russian election
52:34interference using social media was
52:37beginning to take off . The technology
52:40topic of the year , or one of the topics
52:43of the year , is the instrumentalization
52:46of a social network created in the
52:48United States by foreign agents to
52:51interfere in the American election .
52:54There have been a series of setbacks in
52:56recent years that are beginning to show
52:59society that this Pollyannaish view of
53:01technology only for good is a delusion .
53:16If you open your Google in Brazil on
53:18May 1st and note a different detail , at
53:20the top of the page , right below the
53:22search field , there was a phrase
53:24written in blue letters . There are two
53:27versions of this phrase . The first , "
53:29The fake news bill could worsen your
53:31internet . " And the second : " The fake
53:33news bill could increase the confusion
53:35about what is true or false in Brazil .
53:38" These phrases are the perfect
53:40illustration of why Google had to
53:42abandon " Don't Evil , " because they are
53:45Machiavellianly evil . As you probably
53:48know , they're talking about Bill 2630 ,
53:50which was created in 2020 by Senator
53:52Alessandro Vieira of the PSDB party and
53:55which , at the beginning of May , was to
53:57be analyzed by Congress to assess
53:59whether it would become law or not ,
54:01whether it would need to undergo
54:03changes or not . Bill 2630 became known
54:06as the " fake news bill , " but this name
54:09is bad because it hides the central
54:12objective of the project , which is
54:14simply to regulate digital platforms .
54:18Because today , if you use Google ,
54:21Facebook , WhatsApp , YouTube , etc. , to
54:23commit a crime , these companies have no
54:26legal responsibility . And for me ,
54:29Google's statement has a certain faith
54:30factor , firstly because it puts the
54:31whole thing in a conditional statement .
54:33The bill could worsen your internet
54:35experience , it could increase confusion
54:38. And this is a cynical dissimulation
54:40technique that basically serves to
54:42evade justice , because the bill has not
54:44yet been discussed or voted on . In
54:47theory , it could become anything , but
54:49beyond that , the statement has a faith
54:51factor because it's a piece of
54:52advertising and wasn't presented that
54:54way . This last reason , incidentally , is
54:57what led the Ministry of Justice to
54:58take a precautionary measure that made
55:00Google take the phrase down . How did
55:03Google deal with this ? Google and other
55:05public policy companies dealt with it
55:07by lying and doing what we call on the
55:09internet " FUD , " which is fear , unta
55:11andou . When Google's director of public
55:15policy says that they already moderate
55:18content , he's lying .
55:20On April 28th , in an interview with the
55:22Folha de São Paulo newspaper , the
55:24director of government relations for
55:27Google Brazil , Marcelo Lacerta , said
55:29that PR 2630 was vague and that Google
55:32already moderated content . He's
55:34making the water muddier , so to speak ,
55:38to confuse the debate . These moderation
55:42policies of these platforms , besides
55:45being flawed , are always hidden behind
55:49a veil . Thus , nobody knows which videos
55:53are taken down .
55:54Here , Guilherme Felite is speaking from
55:57experience . Since 2019 , his company ,
55:59Verata , has been monitoring channels ,
56:01mainly far-right ones , on various
56:03social media platforms , but with a
56:06particular focus on YouTube , which is
56:08worth remembering . It's now owned by
56:10Google . They started with just over 50
56:14channels . Today , they analyze more than
56:18500. The main objective is to
56:19understand when and why videos
56:21spreading fake news or inciting crime
56:23were taken down .
56:25They don't disclose the reason , but we
56:27analyze each video manually and
56:29speculate on the reason for their
56:31removal . For example , the now-classic
56:35presentation of Bolsonaro attacking the
56:38UNAS in front of ambassadors . Google
56:41took a long time to remove it ,
56:43resisting until a certain point came .
56:46Nothing specific happened that day , but
56:48one day they woke up and said , " We need
56:50to take this video down . " Meanwhile ,
56:52the video remained there for a long
56:54time , attacking the Brazilian
56:55democratic rule of law and generating
56:57hundreds of thousands of viruses . This
56:59decision , it's important to say , was an
57:01isolated case .
57:0290 % of the time , it's not YouTube that
57:05takes down videos . YouTube does very
57:08targeted cleanups . I think the day with
57:10the most video cleanups , YouTube
57:12deleted 20 videos , but that was an
57:14outlier . Normally it's two or three .
57:16Within that scope of 500 channels . So ,
57:20daily you have dozens of videos that
57:22are deleted even today , and you have at
57:25most one or two that were taken down by
57:28YouTube at most . What we see is that
57:32YouTube has strategies to force
57:34YouTubers to remove videos from their
57:36platform without taking down the
57:39channel . This is something that
57:41happened both with Leda Nagle's channel
57:42in 2021 and with Alexandre Garcia's .
57:45YouTube goes there and takes down a
57:46video from AK . And then they mark
57:50dozens of other videos as problematic
57:53and say : " Look , I've already taken down
57:55one video , you've already received a
57:58strike , I've already marked dozens of
58:00these other videos as problematic . You
58:03have the option to take these videos
58:05down yourself , or if you don't , we can
58:07take them down . But if you accumulate
58:10three strikes in 90 days , you lose your
58:11channel . " And then what happens ? The
58:14YouTube channel is very valuable , both
58:17in terms of exposure and money . So ,
58:19what you see is that these YouTubers go
58:21there and take these videos down . So ,
58:23you have hundreds of videos being
58:25purged , basically forced by YouTube .
58:27It's
58:27also important to say that this effort
58:29by YouTube , an effort whose real
58:31dimension we don't know because it
58:33happens behind closed doors , is not the
58:35only factor that leads to the mass
58:36deletion of videos . The vast majority
58:39of these gigantic purges are done by
58:41the YouTubers themselves when something
58:43happens . Let's go back to February
58:462021. Congressman Daniel Silveira is
58:49arrested after threatening to kill
58:51Supreme Court justices . The next day ,
58:55then also a Bolsonaro-supporting
58:57federal congressman , Paulo Martins ,
59:00deletes , I think , about 10 videos . They
59:03all talked about attacking the Supreme
59:06Court or closing Congress .
59:07Want another example ?
59:08Google has always tolerated content
59:11that lied about early treatment in
59:14Brazil . And then you had a multitude of
59:18doctors who filled their own pockets
59:21with money and gained internet fame
59:24selling a treatment that wasn't
59:27effective . So , you have these people
59:29who gained big fans by lying and also
59:32made money doing this kind of thing .
59:34And for Google , it was always fine . In
59:37April 2021 , Supreme Court Justice Luís
59:39Roberto Barroso orders the Senate
59:41president to establish a Parliamentary
59:43Commission of Inquiry into the pandemic
59:46. The following week , the full Supreme
59:49Court meets , corroborates Minister
59:51Barroso's decision , and two days later
59:54Google announces that it will no longer
59:58tolerate content that speaks or lies
1:00:01about hydroxychloroquine , ivaporin , and
1:00:04so on . So , the moment you understand
1:00:07that your online discourse is
1:00:08disconnected from reality , and remains
1:00:11susceptible to punishment under the
1:00:13Penal Code and actions by the Federal
1:00:15Police or any other branch of the
1:00:17police , then you take two steps back ,
1:00:20pretend you didn't say that to avoid
1:00:22the consequences . In other words , it's
1:00:25a continuing crime being committed in a
1:00:28space provided by these platforms , and
1:00:31it continues until the criminals decide
1:00:34to stop for fear of being arrested .
1:00:37That's basically it . That's
1:00:40the moderation that Google's director
1:00:42of public policy says the platform
1:00:44performs . According to the public
1:00:48agency , between April 1st and May 6th ,
1:00:512023 , Google spent $ 670,000 on
1:00:54Facebook and Instagram ads against Bill
1:00:572630. And everything indicates that
1:01:01this was just the tip of the iceberg of
1:01:03a broad disinformation campaign that
1:01:05united the far-right , parts of
1:01:07evangelical churches , and Big Tech .
1:01:10This movement yielded results , and
1:01:12fearing the project's defeat in the
1:01:14Chamber of Deputies , rapporteur Orlando
1:01:16Silva requested a postponement of the
1:01:18vote , which has not yet been scheduled .
1:01:21On May 12th , Supreme Court Justice
1:01:24Alexandre de Moraes ordered the Federal
1:01:26Police to open an investigation into
1:01:29executives from Google and Telegram .
1:01:32The investigation is the result of a
1:01:34significant union of powers . It was
1:01:36ordered by the Justice based on a
1:01:38request from the Attorney General's
1:01:39Office , which in turn was prompted by
1:01:41the Speaker of the Chamber of Deputies ,
1:01:43Arthur Lira . The goal is to investigate
1:01:46whether Google and Telegram executives
1:01:49participated in an abusive campaign
1:01:51against bill 2630. Before finishing , I
1:02:07need to give you some good news from
1:02:09our partners at Rádio Guarda-chuva .
1:02:13The " Vida de Jornalista " podcast by
1:02:15Rodrigo Alves , which discusses the
1:02:18behind-the-scenes of journalism , is
1:02:20back , and it's back with a special
1:02:22season where Rodrigo will profile big
1:02:24names in the national press . In the
1:02:27most recent episode , he brings
1:02:29flavorful accounts of the adventurous
1:02:31life of the discreet , yet brilliant ,
1:02:33Rit Arazim . Look for " Vida de
1:02:37Jornalista " on your favorite audio
1:02:40platform and listen – you won't regret
1:02:42it . This concludes episode 90 of
1:02:47Escafandro . Thank you for listening ,
1:02:51and until the next dive .
1:02:54Hi , this is Valdemar Fonseca . I'm
1:02:57speaking from Vitória , Espírito Santo
1:03:00, and this episode featured additional
1:03:02narration by Priscila Pastre and
1:03:04editing support from Mateus Marcolino .
1:03:07Sound mixing by Coroa . The theme music
1:03:12is by Paulo Gama . The cover art for the
1:03:16apps and website was created by
1:03:18Cláudia Furnari . Direction , script ,
1:03:22and editing are by Tomás Queerini .
1:03:38This podcast is presented by
1:03:41bov.com.br.