0:00

Hello, and welcome to this podcast

0:02

from the BBC World Service. Please

0:04

let us know what you think and tell

0:06

other people about us on social

0:08

media. Podcasts from the BBC

0:10

World service are supported by advertising.

0:15

This podcast is supported

0:17

by athletic Greens. With one scoop of

0:19

AG1 in cold water, you're covering your day's

0:21

nutritional bases and supporting your long

0:23

term gut health. With seventy five high

0:26

vitamins, minerals, probiotics, and adaptogens,

0:28

AG1 supports your immune system energy

0:31

aging and focus, and it's less than three dollars

0:33

a day. Get the nutritional insurance your body

0:35

craves while starting a daily habit you'll

0:37

actually enjoy. Visit athleticgreens dot

0:39

com slash BFC for free one year supply

0:41

of immune supporting vitamin d and five

0:43

free travel packs with your first purchase.

0:49

Hello. It's Tuesday the seventeenth of

0:51

January twenty twenty three.

0:53

I'm Gareth

0:54

Mitchell. And the listeners are all

0:56

thinking, I wonder who the studio expert

0:58

is going to be today.

1:01

Oh, drumroll. You've given it away now. Angelica

1:03

is here. Bill nice to have you back on

1:05

the program. You sound just like Angelica. And

1:08

is that Gilen? No. Angelica, lovely

1:10

to have you. Angelica Murray in in

1:13

in Sao Paulo. And and,

1:15

Teddy, I wonder if we can just take the listeners

1:17

a little bit behind the scenes here. Incidentally,

1:20

the way, folks, we're going to be talking about the unrest

1:22

in Brazil, and we feel that Angelica is

1:24

well positioned to be talking

1:26

about that. I'm not saying it or anything to do with

1:29

the unrest, but you've been reporting it.

1:31

And we're very interested in the

1:33

social media aspects of it as well. So that's what

1:35

we're leading to. But to maybe

1:37

just ease ourselves

1:40

in by going a bit behind the scenes.

1:42

We had a a slight technical issue because in

1:44

you're in a studio there, aren't you in the Sao

1:46

Paulo

1:47

bureau? Yeah. I'm at I'm at yeah.

1:50

I'm at BBC Sao

1:52

Paulo's bureau. And

1:55

I'm in a nice little

1:57

studio and it's all

1:59

really tidy, but it's surrounded

2:02

by cables. Wow. And just before

2:05

we

2:05

started, there was some

2:07

echo from your side. Wasn't

2:09

there? Yeah. I think it's what we call

2:11

bleed from the headphones. Mhmm. And we didn't

2:14

have any plasters or any bandages or

2:16

compression, you know, dressings or anything

2:18

like that. No. But bleed just by the the sound.

2:20

Just comes out of the headphones and ends up with

2:22

a bit of an echo on the

2:23

line. Tell me. Had to unplug it,

2:26

but I couldn't find where to unplug

2:28

it. So just had to put headphones

2:30

inside my

2:31

bag. That's So they inside my backpack

2:33

now. They're safe. So

2:35

what we really expect from all our guests is

2:37

that they turn up of the studio with soundproofed

2:40

bags. Just in case we do

2:42

have a headphone bleed issue. Well, I'm glad

2:44

you've done that. I I'm more comfortable with you doing that

2:46

than unplugging cables in studio you

2:48

know, the lights could go out and

2:50

sell power. You don't know what's gonna happen if you

2:52

start messing with cables.

2:54

So that's

2:55

good on that. So we still got tiny

2:57

bit of bleed, but I think we've we've

2:59

launched the bleed quite well. It's not as bad as

3:01

it was. So thank you because you've just got remembered

3:03

to take out of your bag on the way out of the studio.

3:05

That's all I'm saying.

3:06

Okay.

3:07

It's not gonna look good if I leave

3:10

with the pair of

3:11

BBC headphones in my bag

3:13

not a good look. So you'll be in

3:15

trouble with BBC security. So

3:17

that so think that's easy to

3:19

send. But apart from that, I'm forgetting all my manners.

3:21

But are you well in all that, sir?

3:23

Yeah. I'm good. Yeah. I'm just gonna make

3:25

you jealous. There's this blistering heart out

3:27

there. Whoa. It's,

3:29

yeah, the height of summer, and so paleo, the

3:32

steps have been sitting here. I'm wearing

3:34

three layers even here in the studio and

3:36

I'm still quite cold from my walkover here.

3:38

So on that fascinating chat about the

3:40

weather, what's the weather like with you folks?

3:43

We'll we'll do the

3:44

program, but I hope that you've got Aircon in if you,

3:47

you're not willing to be good. Alright? That's

3:50

Ready to go. Alright. That's good to

3:52

know. Alright. So let's jump in. This is what the

3:54

program sounded like on air.

4:06

Hello. Hope you're well. I'm Garik Mitchell,

4:08

and this is digital planet. Today,

4:11

why autonomous vehicles are being

4:13

described as data centers on

4:15

wheels. And also this week, South

4:17

Africa has ambitious plans for

4:20

free community WiFi, but haven't

4:22

we been here before? And Jennifer,

4:24

Mary, joins us. We've been here before, definitely.

4:26

And Jennifer, marry you're in. That's how Paolo

4:28

joining us from Brazil. It's always a pleasure.

4:31

What else might we have in the show?

4:33

Hi, Gareth. Also up for discussion today,

4:35

I'm looking at the new social media trends

4:37

that are driving unresting

4:39

Brazil, following Luiz inacio

4:41

Lula, the silver coming to power.

4:44

Alright. But you just got leading us

4:46

up to that. Let's talk about autonomous

4:49

vehicles because in a future where

4:51

all cars are autonomous, the

4:53

combined power consumption of

4:55

all the onboard computers, that's

4:57

just the computers combined, that

5:00

could be equivalent to

5:02

all of today's data centers.

5:04

And I said that slowly because it's quite a thing to get

5:06

your head around, isn't it? Now some are describing

5:09

autonomous vehicles as effectively supercomputers

5:12

or indeed data centers on wheels.

5:14

The cars won't, of course, directly give

5:16

off the emissions of today's petrol and

5:19

diesel vehicles, and there'll be all kinds

5:21

of benefits than being self driving.

5:23

But the combined computing

5:25

power consumption is set to

5:27

be a very big deal. Now

5:29

short data processing, it does get more

5:32

efficient as time goes on, but will

5:34

need to exceed the current rate

5:36

of power saving to achieve leave

5:38

a global autonomous vehicle fleet

5:40

that consumes less electricity and

5:43

therefore carbon as the sum total

5:45

of all of today's data centers.

5:48

So says new research from MIT just

5:50

published in the journal I triple e

5:52

micro. And I've been hearing more

5:54

from Lead Arthur some Sidharkar,

5:57

who's a graduate student in aeronautics

5:59

and

6:00

astronautics, and she also has a

6:02

focus on robotics. Computers

6:05

consume electricity. And that electricity is

6:07

generated somewhere.

6:09

If it's from a gasoline car,

6:12

it's gonna be a much higher carbon intensity.

6:14

And if it's from the mix of renewable

6:16

and fossil fuel resources in our current electric

6:19

grid, it'll be lower carbon intensity.

6:21

But regardless, some amount of emissions are made

6:23

to produce that electric on that computer.

6:26

What is the computing that's

6:28

actually going on then onboard one

6:30

of these cars? So to make

6:32

a car fully autonomous, it needs

6:35

to take in a lot of data about its surroundings

6:37

and compute on that data,

6:39

make inferences, and decide

6:42

how to move and how to move without

6:44

colliding into any obstacles to keep everyone

6:46

safe. All of that computing

6:49

is pretty significant. And it's

6:51

often done with algorithms called

6:53

deep neural networks or DNNs.

6:56

And one thing surprising about

6:59

this study was that looking

7:01

at just the sheer magnitude of the

7:03

number of times these kind of algorithms

7:05

would need to be run. For example, across

7:08

all of Facebook's data centers, there

7:10

was on the order of trillions of

7:12

deep neural network or DNN inferences

7:16

or forward passes day.

7:18

And if you look at one autonomous

7:20

vehicle with ten cameras looking

7:22

at a three hundred sixty degree view of the world,

7:25

and it has ten DNNs that

7:27

it runs at sixty hertz or sixty

7:30

times a second, that would be twenty-one point

7:32

six million inferences per hour

7:34

driven. And if you scale that up

7:36

to a billion vehicles, that would

7:38

be twenty one point six quadrillion inferences

7:41

per hour So it's

7:43

an issue where there's a lot of data coming

7:46

in and a lot of computing needs to be done on it

7:48

to complete the

7:48

task. And we could potentially have

7:50

many of these vehicles running around.

7:53

So

7:53

having done all these calculations, have

7:55

you been able to come up with

7:58

a desired power level where

8:00

you say that if we can just get the

8:02

consumption of each one of these

8:04

autonomous vehicles down to this level,

8:06

then we stand a chance of an

8:08

overall energy consumption that

8:11

is less than all the data centers

8:13

in the world at the moment.

8:14

Yes. What we do to model

8:17

and find this number is that we

8:19

model the emissions from computing onboard

8:21

autonomous vehicles as a

8:23

function of the number of vehicles in

8:25

this potential global fleet the

8:27

computer power of the computers

8:29

onboard these autonomous vehicles, the

8:32

number of hours driven per autonomous vehicle,

8:35

and the carbon intensity of electricity

8:37

generation. While this might look like a

8:39

deceptively simple equation, it's

8:41

quite complicated because there's a lot of uncertainty

8:43

in each of these variables since

8:46

we do not today have a global fleet of autonomous

8:48

vehicles, and we're looking at an emerging

8:50

technology. For example, the

8:52

number of hours driven. There are some studies

8:55

that say that the total time

8:57

driven will go down because of improvements

8:59

in efficiency through precision routing. And

9:02

other eco driving strategies. And there's

9:04

other studies that say driving time might overall

9:06

go up because now we can multitask

9:09

while we're sitting in cars, we might be okay

9:11

commuting longer. Currently underserved populations

9:14

may now be able to drive when they

9:16

couldn't do so before. So what

9:18

we do to model that is that we probabilistically

9:21

model it. And then we ask the question

9:24

if we want emissions to stay constrained to

9:26

a certain baseline, For example, all

9:29

the missions from current data centers today.

9:31

What would the computer power have to be

9:34

in these scenarios? We

9:36

sample million scenarios, and

9:39

what we found is using our

9:41

probabilistic model, using current estimates,

9:44

the computer power would have to stay under one

9:46

point two kilowatts

9:48

in over ninety percent of the scenarios.

9:50

So okay. So one point two kilowatts watts,

9:53

you're saying, okay, if we can get down to computing

9:57

power of around one point two

9:59

kilowatts, then that might

10:01

be a desired or at least a

10:02

maximum. We we ideally don't go any higher

10:04

than that per vehicle. Yeah. And over

10:06

ninety percent of the scenarios that we it.

10:09

Now there are a lot of future trends

10:11

that can happen that would change that number,

10:13

of course. So there are trends like decarbonization

10:16

of the electricity generated that

10:18

will power that computer. So the more

10:20

the electricity gets decarbonized, the

10:22

less emissions that are gonna be produced, and

10:25

we might have more leeway on that power

10:27

number for the computer. And the

10:29

other big trend is hardware energy

10:31

efficiency increasing. Where

10:33

over time we see computers getting more and more

10:35

efficient. And based on that rate of

10:38

improvement, we might be able to

10:40

not have as big of a problem as we

10:42

might expect

10:43

today. Okay. So we have more efficiencies

10:46

all the time in computing. But

10:48

if we extrapolate from

10:50

where we are at the moment, are

10:52

we likely to achieve the kind

10:55

of power saving that we're hoping for

10:57

if we keep going at the present rate of making

10:59

machines more efficient.

11:01

There's a lot of exciting things we can

11:03

do with specialized hardware to

11:05

improve the hardware energy efficiency onboard

11:07

these vehicles. Specialized hardware

11:09

is where we design hardware

11:12

to do specific tasks. And

11:14

what we're doing is we're trading off some flexibility

11:17

and what kind of tasks that hardware can be used

11:19

for. In return for efficiency.

11:21

And autonomous vehicles might be

11:23

a really good application for this because we

11:25

know the tasks ahead of time For example,

11:27

we know how many cameras are being computed. We know

11:30

the types of perception and planning

11:32

tasks that need to be done. We know

11:34

at what rate everything needs to be done. It's

11:36

very known ahead of time. So we

11:38

potentially could trade off that flexibility in

11:41

return for getting efficiency boost by designing

11:43

specialized hardware. Of course, there's

11:45

a caveat that autonomous vehicles and

11:47

the computers onboard them will likely have

11:50

long lifetimes between

11:52

ten to twenty years So we wanna make sure

11:54

we can still be able to run future workloads

11:57

and still be flexible enough to do that.

11:59

That's Sonya seat Tawke, and

12:02

Angelica Murray is listening to that.

12:04

So what do you think might be some of the

12:06

solutions to all

12:07

this? I mean, obviously technological solutions,

12:09

but must go deeper than that?

12:12

Yeah. Complexity is

12:14

an issue that doesn't look like is going

12:16

to go any anytime soon in the industry.

12:19

And players one

12:21

way of looking at it is that players need

12:23

to define their role in the ecosystem more

12:25

clearly and evolve partnerships

12:29

to to they need to

12:31

build future proof hardware, for

12:33

example, to handle their computing

12:35

demand to create better horizons

12:37

and so on. I I think the

12:40

autonomous vehicles and the the rising

12:42

adoption calls for not

12:44

just a move away from traditional technology,

12:47

but also reengineering value

12:49

chains. You need the sort of coordinate

12:51

response from private and public

12:53

sectors to ease that transition

12:56

to autonomous vehicles, but that will happen

12:58

differently different speeds,

13:01

different approaches globally. I'm

13:04

sure. And shared standards and

13:06

shared suppose research

13:08

efforts, but

13:09

Thanks on infrastructure as well.

13:11

Yeah. We talked about this before here.

13:13

There is that. But, of course, it's going to be

13:15

a competitive game out there in the autonomous

13:18

a little industry. So how realistic is it

13:20

that competing companies

13:23

will want to work with each other and

13:25

share best practice and knowledge

13:28

in bringing down power consumption in

13:30

there onboard computing.

13:32

Yeah. But as consumers

13:35

call for that sort of stuff, then they

13:39

they become increasingly aware, then

13:42

there will be some kind

13:44

of coordinate response sneakers.

13:47

What about other sort of carbon free

13:49

mobility? Can you think of other options here?

13:53

Yeah. You have hydrogen vehicles,

13:55

don't you? Then you

13:58

have battery powered AVs

14:00

that that they easily absorb

14:03

many, many megawatts. So electricity to

14:05

cover the peak load. Hydrogen

14:08

vehicles are more flexible in

14:10

terms of sources to power the

14:12

fuel cells, but And they also

14:14

don't impact the environment as much

14:17

as using components like

14:19

Nickel, lithium and so on. And

14:22

Hydrogen is a very common element. The problem

14:25

is that fuel cell vehicles are very expensive,

14:27

but technology is evolving, costs are

14:29

low in the next decade. And further

14:31

down the line, you probably see hydrogen

14:34

and battery powered vehicles being

14:36

offered in a complementary way.

14:38

Alright. Okay. Now, Angelica, while we're

14:40

chatting, The digital planet autonomous

14:43

vehicle has now funny if it's brought us onto another

14:45

topic. And it brings us

14:47

to a a very serious issue. Obviously,

14:49

it's going on with you in Brazil.

14:52

Huge violence, protests on

14:54

the streets, key government buildings

14:56

trashed, and this of course, is the

14:59

violence following Luis Inacio

15:01

Lula de Silva coming to power as your

15:04

new president. And, I

15:06

mean, what is driving?

15:09

What's the motivation behind all these

15:11

riots and violence?

15:13

In a very sad thing the very

15:15

sad day for for democracy here

15:17

in Brazil. Basically,

15:20

the the motivation for riots is

15:23

is the result of the elections. Post

15:26

and Harris supporters were not happy about

15:28

it, and it's worth mentioning that

15:30

the voting process in Brazil is

15:32

very modern, very secure, very

15:35

reliable, one of the most modern

15:37

in the world and delivers the results

15:39

minutes after the polls are closed.

15:41

But the former president has spent

15:43

his entire mandate casting doubts

15:46

over the election chronic voting

15:48

process. And so people thought the

15:50

election was rigged. And

15:52

that's one thing.

15:53

And what about social media then?

15:56

Well, what

15:59

what people have

16:02

been telling me, I've been writing

16:04

about this, is that getting

16:07

people to that point of trashing government

16:10

buildings is the result of a process

16:12

that has been going on for nearly a

16:14

decade. So Bolsonaro was

16:16

the first president to dominate this

16:19

narrative online. Mind you,

16:21

Brazil is only behind the Philippines South

16:23

African in terms of Internet use, so he could

16:26

connect with people that way

16:28

very efficiently and consistently over

16:30

the years. So you

16:33

you use that technique, the dog

16:35

whistle. And when you whistle,

16:38

the pack comes. Right. And we

16:40

we have certainly seen that now. Also, with

16:42

us on the program is a billionaire

16:45

Martin Stoss Santos, visiting researcher

16:47

in the politics of digitization at

16:50

the Bernd Elin Social Science Center.

16:52

Thanks very much for joining us at

16:54

Bruner. And I just wondered what your

16:57

analysis is in the situation ratio

16:59

in Brazil. And the regulation,

17:02

is there any around to tackle disinformation?

17:05

Was I gonna be talking about regulating this

17:07

space better?

17:09

He so much for the invitation, and

17:11

it's a pleasure to be here. Indeed,

17:13

I agree with a lot of the points Angelica

17:15

just brought. We have

17:17

gone through almost six

17:20

years, I would even say that it dates a little

17:22

back from Duma who sets

17:24

impeachment of attempts

17:26

of massive manipulation of

17:29

of the Brazilian society in general. Things

17:31

such as how much we could have

17:33

we could trace back how much more

17:36

conservative. The country has

17:38

gone ever since around so many things.

17:40

And and social media has a key

17:43

role in all of these. Because we

17:45

know that a lot of the groups that were

17:47

involved into the

17:49

rioters in Brazil, when

17:52

they stormed the buildings, the

17:54

democratic institutions in in Brazil.

17:56

We know a lot of those things were coordinated through

17:58

social media, through private

18:01

messaging apps. And and for a long time, Brazil

18:04

has been in a lot of these

18:06

companies kind of alerts or even reports

18:10

about radicalization of speech in this

18:12

course here, but they have failed to

18:14

act upon all of this.

18:16

So -- Mhmm. -- I would say that while

18:19

we do face kind of an internal issue

18:21

of our society, not

18:23

just questioning, but also trying to

18:25

provide measures against the

18:27

false allegation that the elections

18:30

were also rigged. We do like

18:32

some more control or even some more

18:34

compromises from from from social

18:37

media

18:37

plus. Okay. Well, so as it is then, somebody

18:40

in Brazil, they they can get on to telegraph

18:42

or a platform their choice and say

18:44

the election was rigged.

18:46

We're going to storm government

18:48

buildings. We don't like this. Lula

18:51

is

18:51

corrupt. There's nothing to stop them doing that.

18:53

No. It it doesn't.

18:56

And and it's been pretty much how they have organizing

18:58

in the past years. But also, I would

19:01

also say that some other factors,

19:03

they do help as well. Right? Bolsonaro

19:05

will portrained himself in every

19:08

single of his social media profiles

19:10

until yesterday as the president

19:12

of Brazil. Uh-huh. And all of those private

19:15

actors, they to even moderate.

19:17

That's one little piece of information that

19:19

he's no longer the Brazilian president. So

19:22

it's not just one thing or not just

19:24

one so Shoomita. It's this very continuing

19:27

speech that goes on from just beyond

19:30

telegram, but also kind of it goes

19:32

through Twitter and WhatsApp in a lot of other

19:34

spaces but we do need some more

19:36

consistent moderation when

19:38

we get the speech that's

19:40

escalating. So the anti Democratic and

19:42

and things like that. Right. So Angelica is

19:44

like one platform, amplifies another platform,

19:47

but that is such an interesting point about Bolsonaro

19:49

on his social media profile still describing

19:51

himself as president, which suggests

19:54

a a president or former president who

19:56

really understands and he gets social

19:58

media, he understands information, a player,

20:00

and winning hearts and minds clearly,

20:03

adds to that the huge divisions

20:05

in Brazil and then

20:08

all is being amplified by social

20:10

media. What about, Luca, on his

20:12

side, because he's not really into tech then? Is he necessarily

20:15

so how are his team gonna do anything about

20:17

this if they don't understand it?

20:19

Yeah. You you

20:21

need to dominate the

20:24

the the use of these platforms during year.

20:26

Lula didn't have a

20:29

mobile phone even. He didn't

20:31

own a smartphone until very recently.

20:33

Wow. It's true

20:35

that Bolsonaro doesn't do

20:38

things. He he doesn't actually goes

20:41

and go to the the to Twitter

20:43

and post the tweet. His

20:45

son does it and his staff does it.

20:48

But you need to be minimally

20:51

acquainted to

20:53

these platforms to be able to catch

20:56

of people's imagination. But he's very

20:58

much a face to face leader.

21:01

He describes himself as

21:03

such. But he will have to

21:05

digitize himself a bit more

21:08

if he wants to minimally

21:11

get up to speed to what has been done

21:13

over the last four years. You've got some catching

21:15

up to do. And so just a final one from

21:17

you then, Bruno. I mean, really just in ten seconds,

21:19

where do you see this gig over the next four

21:21

years, for instance, is it just going get worse

21:23

or are there signs of some improvement? I

21:26

do hope that new government is able

21:29

to provide us with a consistent framework

21:31

for regulating social media

21:33

because I don't see the conspirators stopping

21:36

anytime soon. We're still under the

21:38

threat of new protests

21:41

of new riots in the country. So it

21:43

would be interesting to have both the Congress

21:45

and the executive like, powers

21:47

discussing how to control and tackle

21:50

all of these protests from being organized online

21:53

because

21:54

they're threatening our democracy. Okay.

21:56

Thank you. Bruna, Martin's, Dolce

21:58

Santos. Thanks very much. And we'll hear more from

22:00

you in a few moments, Angelica Mary. Well,

22:02

finally, let's continue a theme that has been

22:05

emerging in recent weeks, there's been a

22:07

lively discussion in the podcast version

22:09

of this program available by the way on all good

22:11

podcatchers and by BBC sounds. This

22:13

discussion about broadband connectivity and

22:17

Woes. Last week, we heard from our listener

22:19

Vladimir in Ukraine about the

22:21

problems and some workarounds in his country.

22:23

Well, let's turn our attention this week to South Africa

22:26

where frequent power cuts are messing

22:28

up. Connectivity pretty badly. Could community

22:30

WiFi hotspots be the answer?

22:33

South Africa's government certainly hopes so, and

22:35

it's investing in thirty three thousand community

22:37

WiFi hotspots. Rolling out the public broadband

22:40

has been a bit slow in the

22:41

past, so what might be different this time,

22:43

our reporter, Ronnie Singh, has been investigating.

22:47

The South African government has big plans

22:49

for its IT industry. This

22:51

is communications minister Kombucha

22:54

and Gervini

22:55

speaking at the International Telecommunications Union

22:58

in the autumn. In South Africa, we

23:00

have commenced the implementation of our broadband

23:03

connectivity program known as South Africa

23:05

or SA

23:06

Connect. To ensure that all South Africans

23:08

have access to the Internet. Mobile telecom

23:11

operating companies are to spend US1.3

23:13

billion dollars connecting

23:15

schools, health facilities, public libraries,

23:18

and government centers. But

23:20

getting people online is only part of

23:22

this initiative. Increasing

23:24

IT skills in the population is

23:26

also a high priority. At

23:28

present, this is being left in large

23:30

part to non governmental and nonprofit

23:32

organizations, NGOs and

23:35

NPOs, who are trying to plug

23:37

the digital skills gap. I

23:39

spoke with Terence Saunders, digital

23:42

content manager at TriLog, a

23:44

corporate responsibility consultancy in

23:46

South Africa. Taren lives

23:48

in Cape Town and creates digital information

23:51

and e learning products for NGOs

23:53

and

23:53

NPOs. She told me

23:56

how they stepped up their IT support

23:58

during the pandemic. During the pandemic,

24:00

the need for NPO services was

24:03

increased. And it also forced most

24:05

of them to move their operations online. And

24:07

while digital transformation did help

24:10

many NPOs to serve their communities

24:12

when service fricked, it also

24:14

brought the stark digital inequality to

24:16

light. So some NPOs were

24:18

able to harness online resources such

24:20

as crowd broadening platforms and

24:22

online training, but others found

24:24

greater challenges. So for example,

24:27

South Africa's mobile data is among

24:29

the most expensive in Africa. There are

24:31

constant power cuts to beating the economy,

24:34

and infrastructure weakness makes Internet

24:36

access

24:36

unreliable, especially in lower income

24:39

communities.

24:40

What is one solution you can think of,

24:42

Taren? One solution that's going

24:44

forward is that NBOs are receiving

24:46

funding from full profit companies.

24:49

According to Trilogue's research in twenty

24:51

twenty two, seventy six percent of

24:53

NPOs are receiving funding from

24:55

for profit companies. But of course,

24:58

the more fragmented and unstable an NPO

25:00

is, the more risky it is for a company

25:02

to partner with them, which means that

25:04

it's in a company's interest to help build

25:07

a resilient NPO

25:08

sector. So to hear what's happening

25:10

on the ground, I try to get hold of

25:14

who works directly with NPOs in

25:16

low income communities and

25:18

a random power cut or outage known

25:20

as load shedding hit just as we started

25:23

to record. You can hear some of the hiss

25:25

as we try to connect, showing the challenges

25:27

of this kind of recording, but we did manage

25:29

to get hold of Zincile eventually.

25:34

I know he's there. Is it a power cut

25:37

since he left?

25:38

Okay. Is that good? Yes.

25:41

Is this okay? Okay. Yes.

25:43

Yes. Yes. Because of that, because

25:46

we're, like, load saving three

25:49

times a day.

25:49

Hello, Zelle. Good to hear you.

25:52

Zenzele, you're a community mobilizer

25:55

and liaison officer with the Ujama collective

25:57

and the warehouse

25:58

trust. Tell me about the Kailitra

26:01

population where you're based. There's

26:03

this dark history in South Africa of special

26:05

injustice. For example, in Kailika,

26:07

there are two point four million people living there.

26:10

It's a fact that if you put two point

26:12

four million people in a two

26:14

thousand square meter land, then

26:17

violence is inevitable. High

26:19

rates of unemployment poverty,

26:22

they are inevitable. And doing

26:24

what there creates a level

26:26

of hope and hopelessness where

26:28

sometimes you feel like you are

26:31

putting air conditioners in hell

26:33

because this the the context is

26:35

very

26:36

dire. And do

26:37

there challenges for an NPO when

26:39

you move from working on the ground, working

26:41

online. For me, technological

26:44

infrastructure follows this

26:47

legacy of spatial injustice. This

26:50

comment is fifty percent of the people do not

26:52

have electricity. That means you

26:54

cannot reach these people. And

26:56

the technological advancement, it's

26:58

not an art historical phenomena.

27:01

It is deeply entrenched in social fabric.

27:04

And the social fabric of South

27:05

Africa, it has stains.

27:08

So in this tough environment, what digital

27:10

technology were you able to introduce

27:12

sincerely? We have to connect to community

27:15

of people, asking them if you have a smartphone

27:17

that you're not using, give it to us. We'll give

27:19

it to our people because they need

27:21

more of smartphones because now

27:23

we are moving from international call

27:25

communication into this online

27:28

communication. And we we did

27:30

get cell phones, but the problem

27:32

was data. We had to go and

27:34

seek data allowance for people.

27:37

After doing that, we installed WhatsApp, Zoom,

27:39

and Google Docs, Tempods,

27:42

and Salesforce to do the

27:44

work.

27:44

And how did it go? Has the technology made

27:47

a difference?

27:48

For the first year, it was difficult because

27:51

people not even used to the technologies.

27:53

But I think twenty twenty two, people

27:56

were used to them and

27:58

what started to flow and people

28:00

were engaging with technology and

28:02

the changes and the new normal people

28:05

were like, this is not going anywhere. Now

28:07

let's adapt to this new reality of

28:09

using this smartphone in order to to interact.

28:12

Even though our work is more on the ground,

28:15

but technology has helped us in keeping

28:17

a community of people together.

28:19

So while load shedding and expensive data

28:22

gives South Africa a challenge, the

28:24

skills and information gap in NGOs

28:26

and NPOs is being addressed by digital

28:28

interventions. This enables

28:31

NGOs and NPOs to become better

28:33

problem solvers and As

28:35

connectivity across the wider population

28:38

improves, these programs should

28:40

roll out more easily.

28:42

Report from Ronnie Singh. So

28:44

final few words from you, Angelica, Marion.

28:46

think what I'm picking up from that is it's not as easy

28:49

as the government just Right. There we go. Thirty

28:51

three thousand community WiFi hotspots. There are

28:53

a lot of social issues and and many

28:55

more to consider.

28:57

Yeah. Just like most

28:59

economies around the world that this titration

29:01

accelerated in South Africa after

29:03

COVID, and the government there wants

29:05

to by the universal Internet

29:07

access by next year, but it looks unlikely.

29:11

But the reality is, like, just like in

29:13

Brazil, the government has a lot on their plate.

29:15

First of all, hunger, rising poverty,

29:18

violence, inequality. So what

29:20

you really need is a strong partnership with

29:22

the private sector in the government with

29:24

incentives in place and clear

29:26

goals to move the needle. We've seen that in

29:29

Brazil around the 5G auction.

29:31

By the start,

29:32

how ever the results are yet to be seen?

29:35

Yeah. And of course, yeah, so selling

29:39

bandwidth in the radio spectrum

29:41

is a way of financing it and

29:44

that happens. So but there's

29:46

a lovely idea there from Zenilei

29:48

about getting people to day innate the

29:50

unused mobile phones?

29:53

Yeah. That you need to create

29:56

that culture, don't you, of not

29:59

starting old phones and

30:00

drawers. And we talked about

30:02

this a lot here in the program. Oh,

30:05

yeah. That's gonna be a big one for this year.

30:07

Well. So, you know, giving up mobile

30:09

phones to people. But then he

30:10

said, then you got the problem of data and data

30:12

plans. Exactly. Yeah. But

30:15

mind you inter

30:18

mobile connectivity is a lot cheaper

30:20

in South Africa than in

30:21

Brazil, so they could build on

30:23

that. Sure. And, Jessica,

30:26

thank you. There you go. We need to be out of here.

30:28

Our studio manager is Michael Mullen,

30:30

and the producer's annual letter over little

30:32

deeper this week. See you next week. Bye bye.

30:35

Alright. Well, let's carry on. That's the program,

30:37

but we can relax a little bit because it's only

30:39

the podcast outflow and well,

30:42

that's still an important part of the program really.

30:44

But we still have Angelica who

30:46

I thought you're still

30:47

there. Angelica with these headphones Yeah. I'm still

30:49

here. I'm still here with two headphones,

30:52

one in my my head and another

30:54

in my

30:54

bag. Right. And Bruno

30:57

Martin's dos Santos is still

30:59

with us as well if we didn't put you off too much in

31:01

the program and it's hopefully,

31:04

okay for you to stay back and tell us little bit

31:06

more now that you're here or

31:08

still here. So Well,

31:11

I've got so many things I want to go through in this out

31:13

chair, actually. But, Bruno, I just wonder if I

31:15

can start with you because I know that

31:17

one of your roles that we didn't get

31:19

to talk about in the program is with

31:21

ICANN, which is

31:24

would we say it's a regulatory body

31:26

for the Internet assigned names

31:28

and

31:28

numbers. So tell us what I can is

31:30

and what your involvement with I can is.

31:34

Nice. Thanks for the question. For the beat

31:36

invitation again. I think Aiken is

31:38

one of these to the broader

31:40

public and look as yet another

31:42

mysterious internet space

31:45

where we discuss policies that might

31:47

look as if they don't affect a lot of the

31:50

our day to day or access to the Internet, but

31:52

it turns out that it's rather relevant. I

31:54

can is this space, this multicycle space

31:57

that discuses policies for IP

31:59

numbers and DNS, so domain

32:02

name system. So whether your

32:04

computer is able to connect to another

32:06

computer in your region or even

32:09

if your connection is redirected through

32:11

kind of this whole network of computers within

32:13

regions or If you wanted to discuss

32:16

whether the dot com should be used just

32:18

for commercial purposes or educational

32:20

purposes, this is kind of what Aiken does

32:23

this day. So it's it looks a little mysterious,

32:26

but it does talk about this very little

32:28

relevant niche around the Internet.

32:30

Yeah. Exactly. That's what I tend think of related

32:33

to I can. It's the the top level domains.

32:35

And some people, if they have heard of it, might be

32:37

about that. The organized that said

32:39

quite some time ago that we can go beyond

32:42

just the dot nets, the dot orgs and

32:44

the dot coms. And now, you know,

32:46

we can have dot f m dot radio

32:48

and and all these other top level domains. And

32:51

some people might have thought, well, it's

32:53

lovely that they've done that, but I've got a good

32:55

watch. I don't go on about what's the big deal, but

32:57

they they do. That's just a small part

32:59

of what I can does and a small part of why

33:01

they matter to all of

33:02

us. On the internet, it doesn't

33:05

much more fundamental than that in the internet society.

33:07

It matters. Exactly.

33:11

So no. Of course. And so what do you come

33:13

into it then? Because you feed into

33:15

it, like, a or of some regional

33:17

level as it were. So here's

33:19

a blunt question. And how how is

33:21

my it's net life? And how is how

33:24

are the listeners' internet lives enriched

33:26

by what you do with

33:27

iCAN? There's a question. That's

33:29

a good one. I think I mean,

33:31

my my part of work and and participation

33:34

in item is more related to the

33:36

domain name, so the the generic top

33:38

level domain. So which means dot

33:40

com dot e d you dot a lot

33:42

of the things we know dot live and and things

33:45

that are more easy to to see and

33:47

identify address on the

33:49

Internet. So I think when we

33:51

look at the kind of like the the Internet

33:53

users' lives and how it does relate

33:56

is it's when we're able to read

33:58

an address and for that to be the site

34:00

is comprehensible. So when we

34:02

see not, like, addresses

34:05

such as icon dot org or

34:07

something dot governments or things like

34:09

that. So it's the part of this discussion

34:12

that helps us address debates

34:15

like trends and even discussions

34:18

based on their top level domain

34:20

and and how can it be more trustworthy

34:23

for the Internet user when you'd go to a

34:25

website that has dot gov

34:27

dot the country code dot b r

34:29

in my case. And instead of just,

34:31

like, just go me through a very

34:33

generic or strange space with

34:36

very strange domain. So I do

34:38

think that the domain name discussion does translate

34:40

into Internet, trust and

34:42

trust readiness and how we can

34:44

read these things better and and educate

34:47

ourselves around the use of the Internet

34:49

in the end of the day. Yeah. And that's all

34:52

part of your because why the role of

34:54

around governance isn't it in internet

34:56

governance? And talk

34:58

us through what that was some people might wonder what

35:00

Internet governance even means because they

35:02

might be thinking, well, the whole point about the internet

35:04

is that it it doesn't have any governance.

35:07

That's the thing. It has these technical

35:09

standards of TCP IP and the

35:11

bedrock on which the internet is based,

35:13

and then the web is, if you like, a an

35:15

application that sits on top of

35:18

the internet. The whole

35:20

point of it is it's it's only an infrastructure.

35:23

Where does governments come

35:24

in? So help people out who might just be a

35:27

bit confused about that. Bauchner:

35:29

Exactly. Although the name might be a

35:31

little misleading. Right? Because if we think about

35:33

governance, we go directly to governments, the

35:36

whole discussion about Internet governance is

35:38

it's basically being able to put every

35:40

single stakeholder. So governments, private

35:42

sector, civil society, academia

35:45

and activism and a lot of people in

35:47

the same space just so they're able

35:49

to discuss within their respective roles,

35:52

things such as chart principles, norms,

35:55

rules, and even sets decision

35:57

making procedures around the use of

35:59

the Internet. So where

36:01

this this is all relevant because once

36:04

governments are in this kind of, like, rather

36:06

in global space of Internet governance,

36:08

they're allowed to discuss rules

36:10

and even like, try to understand

36:12

the transfer for for what what should

36:14

they be regulating and then bring

36:17

these discussions to their kind of, like, local

36:19

realities and and discussions as

36:21

well. So it's interesting to kind

36:23

of understand that the Internet, like, one

36:25

of the goals and maybe one of the compromises

36:27

about regulating

36:29

the Internet nowadays is that we

36:32

very often tend or try to do

36:34

this in a multi stakeholder way. That

36:36

should be democratic, should be participatory, and

36:39

involving a lot of the stakeholders that

36:42

have some interest in this discussion.

36:44

So if you're gonna saying, if you're

36:46

gonna discuss something related to social media

36:48

regulation, then you might as well have

36:50

Facebook, but not just Facebook, but also

36:52

the users on the table. So

36:55

it just means a democratic approach to

36:57

regulation and norm procedure and norm

36:59

settings. So I'll translate it kind

37:01

of like that. Okay. And just in a

37:03

few words, and it will bring Antarctica back in then.

37:05

But just to put like, is it

37:08

is it still US

37:10

centric because the internet started with

37:12

Dharpenet, obviously, in the United States, the

37:14

ICAN came from that as well.

37:16

And some people around world might be

37:18

thinking, it's it's all still very

37:20

US and English language centric.

37:23

Is that changing?

37:25

It tears a little bit still, but

37:27

I do think that the discussion has evolved.

37:30

If you think about first the players or

37:32

who are the countries or spaces that are

37:34

setting the discussion for it. It's undeniable

37:37

that places such as the US or the European

37:39

Union, they still have a lot to

37:41

talk about, but countries such as

37:43

mines such as Brazil, they do have some

37:45

antagonism in this discussion as well. Whether

37:47

it's setting the the the,

37:50

like, the patterns or the norms or suggesting

37:52

things or even hosting discussions. So

37:54

if we're talking about the players

37:57

and where this is good discussion is leading

37:59

I do think that it's slightly north,

38:02

like global north centric, but

38:04

this is definitely improving with the participation

38:06

of countries such as India, Brazil

38:08

and many other places in the world. And

38:11

we just came back from an Internet

38:13

governance forum, that's the EUN forum

38:15

for the Internet that was hosted in

38:17

a few in allowed us to discuss

38:20

a little more in-depth questions that were more

38:22

relevant to the African continent such as

38:24

meaningful access to the Internet? How do

38:26

we translate? Just

38:28

from the English, but to French and to many other

38:30

local languages and things like

38:32

that. So I do think this is evolving.

38:35

And and just maybe to to mention one last

38:37

thing about and relating

38:39

to the eye can work as well. Eye can is it's

38:41

nowadays discussing what we call

38:43

universal access, so universal

38:46

translation of of domain names, which

38:48

means that we want to be

38:50

able to to translate domain

38:52

names from just the the latching transcripts

38:55

and and to allow for other types of

38:57

languages and expressions to also

38:59

be present on the Internet when you type

39:01

the dot selfie now. So,

39:04

yeah, it's it's a lot of a lot of things

39:06

ongoing, but it's definitely changing

39:08

outside. Oh, that yeah. That isn't testing.

39:10

Yeah. Alright. Good. Well, Angelica,

39:13

Marriott, it's bringing you back in and we'll substeer

39:16

seamlessly back kind of towards the

39:18

discussion that we were having in

39:20

the program about social

39:23

media and the role

39:25

that various actors have played relating

39:28

to the unrest. And

39:30

in looking into all this, I came across

39:32

a bit of a kind of slang term that seems to

39:35

to be going around in Brazil, which

39:37

literary translates into English is like Selfdriving

39:39

people as zap uncles or zap aren't

39:42

these. And it's a bit derogatory.

39:44

I think what So what's all about?

39:47

Yeah. It's a it's a really popular term.

39:50

It's used to describe specifically

39:52

all the people who are supposedly more

39:55

likely to receive and

39:57

spread this informational line. It's

40:00

a rather controversial area of studying

40:02

cyberpsychology, which is my field.

40:05

And part of the general argument here

40:07

is that all the people tend to prefer

40:10

consistency and that grows with

40:12

age. And that behavioral pattern

40:14

is particularly useful for

40:17

actors targeting all

40:19

the demographics with fake news

40:22

because when information is

40:24

consistently repeated and

40:26

people tend to believe it to be

40:29

more valid than when it's presented

40:31

just once. If that makes sense,

40:34

and all the adults are particularly

40:37

susceptible to this truth effect.

40:39

But on the other hand, you have some really

40:41

exciting studies that challenge those

40:44

assumptions. There is a researcher,

40:47

did then Pili Vano Glu

40:50

and colleagues. They published a study

40:52

last year. They found the ability

40:55

to detect this set of news is actually

40:57

comparable between younger and

40:59

older adults. So age

41:02

related vulnerabilities to this

41:04

information like analytical reasoning.

41:08

So our ability to assess people's

41:10

intentions is only a parent at very

41:12

old age after seventy years old.

41:14

So it's something

41:17

that we we should discuss

41:20

and break that so that

41:22

paradigm, you know.

41:23

Right. You

41:24

already eating that only old people

41:27

spread fake news. Yeah. Exactly.

41:29

You know, blaming the old people, you know, and they

41:31

just spread all this fake news and then bad things in the

41:33

world happen. Yeah. That would be deeply controversial.

41:36

And, you know, many of the listeners to this program, I

41:38

I would say are older people. They'll

41:40

be sitting brisling away at any

41:42

suggestion that their generation somehow

41:45

just are zap uncles or zap aren't

41:47

these. So that that of terms that just

41:49

come from, say, a stereotype

41:51

of your old uncle sitting there just on

41:53

WhatsApp, just zapping out all these different messages.

41:56

Is that where it comes from? Yeah. Exactly. That's

41:58

where it comes from. It's derogatory. Not

42:00

very nice to say. Because and,

42:02

yeah, there's so many other conversations that go over.

42:04

People blame it. On the young people, they're always

42:06

on the screens all the time. They never bothered to

42:08

read the news properly, and they just keep you

42:10

know, I've been two seconds there sharing

42:12

something. So it cuts both ways. Doesn't

42:15

it? You know? It's it's --

42:15

Yeah. -- young people

42:16

have been blamed on me now. Than that.

42:18

Patronizing and and

42:20

just ignorant really.

42:22

So so if we've between

42:25

us in our little seminar here, ruled

42:27

out the age connection. We've

42:30

we've critiqued it quite strongly

42:33

there. In cyber psychology, what

42:35

might help lead us towards the

42:37

kind of character or the trait

42:40

that might make people just consistently

42:42

share harmful discs misinformation?

42:47

Well, it's one of the things

42:49

is consistency that

42:52

I just talked about. And the

42:54

there is authority. So that comes

42:56

from Robert Kildini's seven

42:59

principles of persuasion. So

43:02

how you persuaded online So

43:05

another principle is authority. So

43:08

the argument is that all the people

43:11

they look for more for

43:15

they they value information that comes

43:17

from people in positions of authority.

43:20

So in Brazil, it would be, for

43:22

example, the president. So

43:24

if the president posts something,

43:27

then that is supposed

43:29

to be true. And that's

43:31

quite dangerous. When you have an

43:34

authoritarian

43:35

regime, their likes

43:37

to spread this information. Yeah.

43:41

It makes us wonder where that takes us in a sense.

43:43

I mean, Bruno, I don't know if if you can come in

43:45

here even from Maybe it's not

43:48

exactly a governance question, but you'd be

43:50

bound to be interested yourself in

43:52

tackling this problem of missing disinformation

43:55

and a big part of that is understanding

43:57

the kinds of traits behind

44:00

the people who might be sharing all this stuff.

44:02

So whether it can be then

44:05

that that we can build or

44:07

policies at least can be built to try

44:09

and ameliorate some of

44:10

that. I I don't know. Yeah.

44:13

I I first of all, would just add I

44:15

completely agree with everything Angelica just

44:17

said, but I would just add as well that

44:19

maybe in the Brazilian scenario what

44:21

some researchers have seen as well, it's

44:23

so level of proximity. So

44:26

as if Brazilians would like to be

44:28

closer to the person that's telling

44:31

them whatever they're listening to whether it's

44:33

kind of a hopes about

44:35

government or the fact that the

44:37

the the Brazilian elections might have

44:39

been rigged. There is a lot of dispute information

44:42

networks that it's built on the

44:44

proximity between the people who are like

44:46

actually talking about those subjects.

44:49

And think like in Brazil, in the recent

44:51

years, in the last two to three years,

44:54

we have been discussing very consistent draft

44:57

bill about with

44:59

some ideas for tackling the information. But

45:02

at first, like, we decided to take a step back

45:04

and and kind of discuss what could

45:06

be the baseline for understanding the problem

45:08

and the the the size of the problem

45:10

of this information in Brazil. So what

45:13

places like the coalition and part of

45:16

in my country they try to set

45:18

forward is for, like, increased rules

45:21

around transparency of these

45:23

platforms and things like such as

45:25

transpires reports and the publication

45:27

of the numbers of users just so we can understand

45:30

unless a bit of how many contents is

45:32

tackled annually that's actually

45:35

legal or how many of those contents

45:37

is just, like, taken down based

45:39

on violation of the rule. For these

45:41

platforms. So these are some of the

45:43

ideas. And I think, like, Brazil nowadays, it's

45:46

I would say it's rather fine tuned with the

45:48

discussions that happened around the DSA.

45:50

I just hope we have the time

45:52

and dedication from our policy

45:55

makers to actually approve such

45:57

such a legislation for the country that would

45:59

allow us to have more

46:01

transparency, do diligence, into process

46:04

around social media.

46:05

Yeah. Sure. If I may come in here,

46:09

it seems that is really promising, isn't

46:12

it, Bruno? Because the

46:14

the group that supported the transition of

46:17

government

46:18

has been discussing that

46:22

sort of topic hasn't it?

46:24

Mhmm. Definitely. A

46:27

lot of the the I would say experts

46:29

and and colleagues as well that went to the transition

46:31

governments and are now joining

46:33

the the Lula's third term.

46:36

There are people who have been on these discussions

46:38

for a lot of years now. So people

46:40

who have participated in the debates of

46:42

the fake news graph view, but also the

46:44

construction of the civil rights framework for the

46:46

Internet. That's the most relevant

46:48

legislation in Brazil nowadays for

46:50

social media, intermediary liabilities and

46:53

things like that. So I do think that

46:55

that there's a lot of things related

46:57

to that. I do remember that around

46:59

the go the transition government days,

47:02

they were even discussing, like, having digital

47:05

services coordinators such as the

47:07

one that's mandated on the

47:09

DSA, just so the Brazil the

47:11

government would have someone educated

47:13

together. Sure. And and just for people, we've made we've missed

47:15

it. What's the DSA AI again?

47:17

Digital Services Act. Yes. Is this

47:19

regulation from the European Union

47:21

rights?

47:22

Yeah. Okay. I've I've got that. And Brazil

47:24

has it equivalent then? Brazil

47:27

is discussing is yet to discuss any equivalent

47:29

to that. What we have not is

47:31

the civil rights work for the Internet. That

47:33

is kind of a human rights based regulation

47:37

for for just the the use

47:39

of the Internet in the country. And we have

47:41

one one, like,

47:43

specific intermediary rule, but

47:45

it's not really some some critics.

47:48

They say that's not really expendable for

47:50

the Internet we have nowadays. With

47:52

so many illegal and disinformation contents.

47:55

And when you because I

47:57

know that you are as you mentioned

47:59

there, that you've been involved in

48:02

the discussions leading to the approval of the

48:04

Brazilian civil rights framework,

48:06

probably have to close on this thought. So trust

48:08

me to come up with massive question to finish with.

48:10

This might be your homework. But

48:13

this must be all about balance saying,

48:16

often competing rights, say people's

48:19

rights to freedom of expression versus

48:21

other people's rights, not to be victimized,

48:23

not to have their buildings smashed up by

48:26

a mob or what have you. Is that part

48:28

of it balancing different rights like that if

48:30

I massively, as I often do, oversimplified

48:33

it.

48:33

No. It it is you're very much correct.

48:37

The discussion about regulating the Internet

48:39

or even regulating social media. It's all about

48:41

balancing rights. Right? Because at

48:43

the same time, we want to provide appropriate

48:46

tools and or even like good enough tools

48:49

to badness spread of misinformation. And

48:51

also be respectful of, like, freedom of

48:53

expression or even privacy line. So

48:56

when we were discussing this Raspio in Brazil,

48:59

like a lot of the things they they they

49:01

were into, like, on the table, like,

49:03

such as, are we gonna allow for

49:05

more in increase protections to privacy

49:07

of Internet users. At the same time,

49:09

we wanna provide tools for investigation

49:11

of this information. And there was even kind

49:13

of a discussion about follow the money. That

49:16

was kind of finding out who are the sponsors

49:18

of those networks. And the

49:20

point is that our discussion kind of got

49:23

stuck, and we're still waiting for

49:25

the the congressional work to come back

49:27

with the parliament in on the second of February.

49:29

So there's still, like, a few next

49:31

chapters to this discussion and into balancing

49:34

those rights as

49:34

well. Sure. So well, there, I think

49:36

we'd better leave it. This has been quite an out shows it.

49:39

I love the podcast out there. We can have

49:41

these discussions, but not

49:43

forever. But

49:45

thank you both. Seriously. That is that is lovely.

49:48

And I'd love to go on a bit. We'd better just

49:50

let everybody get on with things as well and the listeners

49:53

might want to move on to their next podcast and I can't.

49:55

Well, I can't blame them a bit, but you know what I mean?

49:57

So Bruno Martin Stos Santos has been lapsed

49:59

with pleasure to meet you for the first

50:01

time. Thank

50:02

you so much. Thanks for your attention. We'll do it

50:04

again. And and, yeah, Elica as ever.

50:06

And, yeah, you can take those headphones out

50:08

of your bag before you

50:10

leave. And then

50:11

Thanks for the reminder. Breathe in

50:13

that fresh air conditioned air

50:15

and make the most of it before you head out into

50:17

the boiling streets of sour palate.

50:20

But they will leave it wherever you are folks,

50:22

whatever the weather. Yeah.

50:24

I didn't know how to say goodbye really apart from

50:27

take care and goodbye. That was it. Bye.