November 5, 2025 • 33 mins
A few weeks ago on Monday, October 20, a major outage in Amazon Web Services took down a bunch of the Internet. How did an error in one data center in Virginia affect everything from Snapchat and Reddit to ‘smartbeds’ to government services? Dexter talks to Dr. Corinne Cath, a cultural anthropologist and technology researcher, about how the “cloud,” and therefore the Internet, became mostly controlled by 3 big tech companies – Amazon, Microsoft, and Google – why that puts our democracy at risk, and what we can do to change it.
Got something you’re curious about? Hit us up killswitch@kaleidoscope.nyc, or @killswitchpod, or @dexdigi on IG or Bluesky.
Read + Watch:
Corinne’s op-ed: https://www.techpolicy.press/amazon-cloud-outage-reveals-democratic-deficit-in-relying-on-big-tech/
Our episode on AI and the environment: https://podcasts.apple.com/us/podcast/kill-switch/id1449757372?i=1000711126940
See omnystudio.com/listener for privacy information.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:10):
We have breaking news for you this morning.
Speaker 2 (00:13):
A widespread Internet outage has caused major disruptions for some
of the biggest web services in the world.
Speaker 3 (00:18):
There's a good chance today that some website, app or
service that you tried to use today was down. That's
following an outage from Amazon Web Services.
Speaker 4 (00:26):
Amazon Web Services, the backbone of the Internet, suffered a
major outage that took hundreds of services offline across the globe.
Monday morning, October twentieth, Americans woke up to news of
a massive Internet outage, and it wasn't just America.
Speaker 1 (00:41):
I think one of the things to really note about
this outage is that it was one of the biggest
in recent history. In my decade in the field, this
definitely like ranks in the top five.
Speaker 4 (00:56):
Doctor Kreen caf is a culture anthropologist and she works
at the Human Rights organization Article nineteen. She researches Internet infrastructure, AI,
cloud computing, and basically how technology affects our day to
day lives.
Speaker 1 (01:08):
I have asked some of my colleagues, like veterans of
the field who've been in the industry since the Internet
was born, and they pretty much said the same thing,
like this was huge a.
Speaker 4 (01:20):
Lot of people didn't necessarily know that there was some
global internet outage going on. They just knew that there
was some website or service that they used that wasn't working,
like their favorite messaging app or even their smart bed.
Speaker 1 (01:32):
Some of these are kind of beIN benign or even funny,
like there were some excellent threads on the Internet about
people who had Internet connected mattresses.
Speaker 4 (01:44):
Oh the beds.
Speaker 2 (01:45):
Yes, yes, the beds. Oh my god, Apods. Shout out
to apods and everyone who thought it was a good
idea to buy an internet connected bed. I sort of
got this is why we don't have a single interconnected
device in our house.
Speaker 4 (01:57):
Krin's talking about these really expensive smart beds by this
company called eight Sleep. Their mattresses are supposed to track
your biometric data and get hotter or cooler, depending on
what your body needs. They're also connected to the Internet,
and apparently they didn't work offline after the outage started.
Some of the mattresses overheated or wouldn't recline. Thinking about
(02:17):
somebody's twenty five hundred dollars internet smartbed going vertical and
trying to cook them alive is kind of dystopian, but
it goes beyond that. This thing took down everything from
Amazon's own products like Ring doorbells and Alexa, to sites
like Reddit, to messaging apps like Signal, to government services.
These outages can impact basically every part of your life.
Speaker 1 (02:39):
This isn't just like me being on Duo Lingo or
me being on Snapchat. It is banking, It is the hospitals,
it is education, like it is things that matter in
my life, that matter in everyone's life.
Speaker 4 (02:57):
The world today relies on basically three major cloud comuting services.
When one of those goes down, so does a huge
chunk of the Internet. So what causes these outages? How
do we get to the point where so much of
the Internet relies on just three big tech companies? And
can that ever change any time? From Kaleidoscope and iHeart Podcast,
(03:26):
this is kill Switch. I'm Dexter Thomas.
Speaker 1 (03:37):
Goodbye.
Speaker 4 (03:43):
What exactly happened on October twentieth, So.
Speaker 1 (03:47):
What essentially happened is that there was an outage within
Amazon web Services. Now, Amazon Web Services is a leading
cloud computing company, and what is super important to know
is that Amazon currently has at least thirty percent of
the global cloud market share.
Speaker 4 (04:05):
We sometimes think of the cloud as something that lives
in this virtual space, not as something tangible. But the
cloud relies on very real and physical data centers that
are located around the world. These things store code and
data and move around that data to make the apps
that you use or the Internet run. And as Caring
(04:25):
just pointed out, about a third of the global Internet
relies on one company for cloud computing, Amazon Web Services.
Speaker 1 (04:33):
This company operates data centers worldwide that a lot of
the Internet and a lot of applications rely on. And
so what happened is within their data center clusters, one
of the oldest and biggest one had a bug in
the automation software that subsequently took down all.
Speaker 3 (04:54):
Of the companies that relied on that data center.
Speaker 4 (04:56):
Basically, the reason this outage was so widespread was because
it was an issue with one specific and heavily relied
upon data center that's located in Virginia.
Speaker 1 (05:06):
I've seen estimates of you know, a billion pins per
day going to this particular data center in Virginia, and
a pin is essentially like when a company needs something
from that database. And because this data center was one
of the biggest, some one of the oldest, about thirty
five to forty percent of all AWS traffic.
Speaker 3 (05:25):
Goes through it.
Speaker 1 (05:27):
Another reason why this impact was so big is if
you are a software developer and like you buy AWS
to develop your software on, this region is the default,
So like, if you don't switch to the default, this
is where your stuff will be housed.
Speaker 4 (05:46):
And so you have a third of the world's Internet
relying on this one company. Within that, there's this one
data center that becomes a default for all kinds of
new apps and websites. And then last month that data
center had a DNS error. The DNS, the domain name system,
translates a written website into an IP address that a
computer can then decode and understand. You might have heard
(06:07):
this described as the phone book of the Internet, but
Carin has a better explanation.
Speaker 3 (06:13):
So I'm going to.
Speaker 1 (06:14):
Try and do a different metaphor to explain this, because
the phone book is really useful in a sense that
what it does it connects names to numbers, right, and
that is what the domain name system does. So when
you type in a websites, you type in riverside dot FM,
not the very long number that is behind it, which
(06:35):
is actually how machines find each other. The problem with
the phone book metaphor is I've held one, I know
what they look like and what they feel like. But
most people younger than me, that is not a helpful
metaphor at all. So the way that I try to
explain what happened with AWS is imagine you're using Google
(06:56):
Maps or Ways or some sort of navigation app because
you want to go to the library because you need
a specific book to do your job. But essentially what
happens is there's a software bug and that causes your app,
your map app, to essentially indicate that the library no
longer exists. Even though the library is right there, it
is fully operational, but you just no longer can see
(07:18):
how to get there, and you're stuck with two problems,
the first one being is you cannot navigate to the library,
and secondly, you cannot access the information that you need,
and so you can't do your work. And that's essentially
what happened with the AWS outage. But the problem is,
like you're not the only one who needs the library.
The writer on your team needs access to it for research,
(07:42):
the accountant needs it for whatever financial records or software,
your lawyer needs it for the database. With case law, right,
But when the map says the library doesn't exist, it
pretty much doesn't show up, and so everyone's work stops simultaneously.
And what AWS figured out is like fixing that map
is not easy. You have to take all of these steps.
(08:05):
You have to identify which version of the mapping software
each person is using, push all of the updates to
all of the systems, wait for everything to restart, and
then verify that everyone can see the library again on
the map. And then you get this really interesting part
where when you fixed it and the library is reachable again,
there's a stampede because everyone realizes, like, oh, it's back up,
(08:26):
so we now all need to go there. And everyone
who's done like Black Friday shopping knows what that leads to.
And that is also why it took so long, right like,
because one thing is fixing the problem, and subsequently you
have to deal with the backlog of everyone trying to
get through the door simultaneously and the system not being
able to deal with that either.
Speaker 4 (08:44):
This isn't the first time the cloud services have had
an error like this, but because of how many applications
rely not only on AWS, but on this one data center,
the effects of the outage were really widespread, and outages
of this scale can cause some very real problem.
Speaker 1 (09:00):
This is not the first outage, not the first and
big one, right. The last big one was on the
nineteenth of July twenty twenty four, And the reason why
I remember that date so precisely is because it happens
to be the day that my second born was born,
and for reasons I had to be rushed to the
hospital and it was like urgent.
Speaker 3 (09:20):
This was no joke.
Speaker 1 (09:22):
And that was another day that there was a massive outage,
but this time it was CrowdStrike Microsoft Cloud.
Speaker 4 (09:27):
In this incident last year in twenty twenty four, what
happened was a faulty update by the security company CrowdStrike
caused a major outage for Microsoft's Azure cloud services. This
one brought down airlines, banks, and even emergency services.
Speaker 1 (09:42):
The massive Microsoft outage impacting businesses around the world, taking
some companies temporarily offline and forcing major American airlines to
ground flights. The outage affected government entities and news organizations, as.
Speaker 4 (09:55):
Well as almost every industry across the globe, including banks, airports,
healthcare providers, and retailers.
Speaker 1 (10:03):
There were a number of hospitals in the Netherlands where
I'm based, where those issues were such that they couldn't
help patients. Now luckily that wasn't in my region. But
when you're in an ambulance being sped to a hospital, you
don't want to show up there on the other end
and then telling you I'm sorry, computer says no. You know,
(10:24):
it's the apps and it's the benign stuff, but it's
also the very serious stuff that we have connected to
these clouds. Healthcare, education, banking, access to your government. These
are all things that are so critical for people's lives.
Like what happens when that is not accessible? I mean,
I am sure that there are people who missed funerals,
(10:45):
who were not able to make really crucial payments that
needed to be made that day, that didn't get whatever
benefits that they really needed to do. This is not
an incident because it keeps happening. When this happened, like
the organization that I work for nineteen like, we had
a blogout within two hours at best, and people were like, wow,
(11:06):
you guys are so fast, like how did you put
that together? And I'm like, well, it's kind of depressing
because this is the fourth time in four years that
I see in outage quite like this, And all I
have to do every time is I switch out the
name of the company and I look into the specifics
of where they feel technically, and then it's the same
up D and so we just you know, here we are.
Speaker 4 (11:29):
That is depressing. That is depressing. It's wild too, you
said it keeps happening. You might already know this, but
I asked you on to talk about aws. Do you
know Minecraft is down right now?
Speaker 2 (11:41):
Oh?
Speaker 3 (11:41):
I did not see that.
Speaker 1 (11:43):
Yeah, preparing for talking to you guys and not.
Speaker 4 (11:46):
Not doing your job, I kept you from doing your job. Yeah, No,
I think Xbox is down right now. Microsoft's Azure was
once again experiencing an outage. At that point in the interview,
all I'd seen was that Xbox and Minecraft were down.
But as always, it's bigger than that. It turns out
banks got hit with a two and so did parts
of the train system, which was a problem because it
(12:08):
also happened to be election day in the Netherlands, where
Krin is, and the disruptions and Dutch railways meant that
some people might not be able to make at home
in time to vote. So the outage was directly interfering
with the democratic process. Microsoft's share of cloud computing isn't
as big as Amazon's, but it's a close second, and
together with Google, these three companies make up almost two
(12:29):
thirds of the entire cloud infrastructure, meaning that two thirds
of the Internet relies on Google, Amazon, and Microsoft keeping
their act together. So how did we get here after
the break the origin story of Amazon Web Services? How
(12:54):
do we get to the point where the bookseller became
the backbone of most of the Internet.
Speaker 1 (12:59):
So there are more multiple origin stories around, like how
AWS became AWS. A lot of the work that people
put out there now like very much follows aws's iteration
of events. You know, we were an e commerce platform.
We needed a lot of compute to be able to
(13:19):
do that. And then at one point they realized, like, oh,
in addition to selling the books, we can sell whatever
compute capacity we don't need ourselves.
Speaker 4 (13:29):
In this version of the story, Amazon Web Services kind
of started as a side hustle with Amazon renting out
the extra computer capacity that they had to other companies.
Fortune magazine did an interview with Amazon's current CEO, Andy Jasse,
who started out as the head of Amazon Web Services.
He said that in the early two thousands, Amazon was
starting to run in the walls when they were developing
(13:50):
features for the site. Remember at this point they were
still mainly thought of as a bookseller, so they started
building out internal features for their own programmers to use,
and they realized that they could sell that infrastructure to
other companies.
Speaker 1 (14:04):
So that is like the long short of it. There's
a lot more nuanced there, and I think some really
great Internet historians are going to be picking through that
to bring the inside story. What is definitely true is
that they were the first.
Speaker 4 (14:15):
In two thousand and six, Amazon Web Services was born,
and it was instantly popular. They were solving for a
problem that companies were already starting to have before this.
Starting an app company could be really risky, partially because
of how unpredictable the traffic demands could be. Let's say
you start up a company and your online service just
goes viral overnight. Well, if you didn't buy enough service
(14:38):
ahead of time, you can't keep up with demand and
you're screwed. But if you did buy a bunch of
servers and your company isn't doing so well, well, now
you're stuck with a building full of servers and you're
wasting money, and now you're really screwed. What AWS did
was let companies rent that computing space and power as
opposed to having to buy it outright themselves. Basically, they
(14:58):
let them do what a lot of us do, pay
Amazon a subscription fee. If your business gets popular, just
rent more space. If you have a slow month or two,
you can scale it back. And on top of that,
AWS had software development tools, which meant that you could
build on top of those tools rather than having to
start to scratch. This was huge for the web two
point zero boom. It allowed for a lot of experimentation
(15:21):
without as much financial risk, which in some ways felt
like a good thing, but it also meant that the
Internet became really reliant on Amazon.
Speaker 1 (15:32):
Companies have gone from largely running their own servers like
physical computers in their own garages, so to speak, to
relying on the servers of AWS. But maybe even more importantly,
like we've also gone from individual companies building software from
scratch to compiling it from various building blocks provided by
(15:53):
other software companies.
Speaker 3 (15:56):
What you then get.
Speaker 1 (15:57):
Is that you create this really complex web of dependencies
and interconnections where like every piece of software relies on
another piece of software which relies on another piece of software.
Speaker 4 (16:10):
However it might have started, AWS is not a side hustle.
Now it brought in almost forty billion dollars in operating
income last year. That's over half of what Amazon made.
Amazon beat Google and Microsoft to the punch here, but
both those companies eventually offered their own competing services, and
now we have this environment of three companies controlling the
(16:30):
cloud that houses the Internet, and what was first just
an extension of basically using someone else's computer for storage
has turned into something much much bigger.
Speaker 1 (16:42):
I think that is the next sort of step that
we need to take, where it's like, in what situation
would you accept, you know, only three companies being responsible
for all of agriculture. We would immediately see how big
of a choke hold that would be. Because we understand
(17:03):
how important food is for our daily sustenance, we need
to start changing our perception of what the digital is,
because it is everything.
Speaker 4 (17:13):
The food comparison is really interesting, Like you want a tomato,
you want a hamburger, you want, I don't know, man,
some spaghetti, all of those. I don't know why I
thought of those three that maybe I'm hungry right now.
But all of those three things are you know, if
(17:34):
one company goes down, if something is one of those
companies has a problem, if one of those companies is
not safe for some reason, starts slacking on safety regulations
or whatever, you just you don't get one of those
three or two of those three, or all of those
three or whatever. You don't eat, man, Yeah, you just
don't eat.
Speaker 1 (17:52):
There is a reason why these three keep getting bigger
because they are also the only ones with like that
amount of money to put into something that loses its
value quite quickly, right like servers infrastructure. It's expensive to build,
it's expensive to maintain, and you have to switch it
out every five to seven years. I've spoken to vcs
(18:14):
for research purposes just get a sense of you know,
when you get a startup and you look at their proposition,
and for instance, they would say like, oh, actually, we
either want to build our own set of servers, our
own mini cloud, or we would want to work with
multiple partners. And these vcs just like burst out laughing
at just the proposition of that, because they were like, well,
(18:37):
if I'm putting my money into this and they're going
to spend it on like something that gives me an
overturn when they could just do it with AWS or
whomever's going to do it cheaper and better, at least
from their perspective, I was just never fund I would
never fund something that would also require my money to
put into machines.
Speaker 3 (18:57):
That are going to die in the next five years.
It's a bad investment right now.
Speaker 4 (19:01):
Smaller companies really can't compete with these giants. So you
could argue here that it's better that we have these
big tech companies monopolizing this. They have more resources, and
they can afford to maintain the data centers and keep
them working, and maybe that is a way to keep
things cheap and reliable. But for Corent, that's not a
good enough reason.
Speaker 1 (19:23):
The thing that I think is important to remember is
that these technical outages have political consequences. So it is
really important to politicize what is happening here, because it's
very easy for people to say like, oh, it's just
a technical mistake, and we should be allowed to make
technical mistakes. But that is not an argument that we
(19:46):
would accept from, you know, the company that operates our
water or our electricity, for them to just be like,
I'm sorry, it's winter, but for the next twelve hours
no heat in your house or no clean water, so
your kids are thirsty. Too bad. And somehow with these companies,
we've decided that they get a pass or like maybe
(20:07):
even better, like they've just lobbied such that there isn't
that much or enough at least from my perspective oversight
on you know, them making these mistakes and that having
really potentially life altering consequences.
Speaker 3 (20:22):
And that I very much see as a democratic deficit.
Speaker 4 (20:26):
So what can we do about this? How do we
take back computing power from the hands of a few
big corporations. That's after the break The same week as
the AWS outage last month, political leaders in Europe met
(20:47):
in Brussels for a summit, and one of the items
on the agenda was this idea of European digital sovereignty,
basically figuring out a way to untangle Europe from being
so reliant on these big US tech companies.
Speaker 1 (21:00):
I'm actually quite critical of people who are very focused
on the European digital sovereignty because I'm like, well, you know,
the world's a big place, and you just you know
what ford Europe is where you put your digital borders,
like that's not good enough. The perspective needs to be.
Speaker 3 (21:14):
One of global solidarity always.
Speaker 1 (21:17):
And I have been in the room with politicians representing
you know, the Greens or left wing parties back home,
who without irony, have said, oh, well, you know, we
need to start looking for alternatives. But the only ones
that are perhaps available to us, or that have experience
(21:38):
running these kind of networks and that kind of skill
are defense contractors. And I would be sitting there picking
my job from the floor, being like, wait, the choice
that we are giving ourselves here is AWS or defense contractors, Right,
that is a rock and a hard place. We have
(22:00):
to think of something else that.
Speaker 4 (22:01):
Is not this Instead of looking to Silicon Valley or
the military for the answer, Corent says, we should be
thinking about nonprofits and universities for ideas here.
Speaker 1 (22:10):
Now, it's not impossible. Wikipedia does it. They specifically have
a setup that has them. I want to say, like
across seven different data centers across numerous continents, and none
of them are the Big Three. But that would be
much harder to do for something like signal right, where
latency and like constant uptime is much more visible when
(22:32):
it doesn't work.
Speaker 3 (22:33):
Then with something like Wikipedia.
Speaker 1 (22:35):
There are alternatives, but it requires us to also change
our expectations, and it requires the people more importantly, like
who put money into this, who will literally have skin
in the game, to change theirs. So it's cultural change
as well. It's not only technical. Locally, we work with
small medium enterprises to see, like, you know, what they
(22:56):
are thinking of the universities in the Netherlands of taking
some really interesting steps where they have all come together
with like local providers to say like, okay, well we
are roughly the same kind of institute. Here are the
needs that we all have. Could you build us something
that is like fit for purpose for universities.
Speaker 4 (23:19):
So there is a kind of precedent for what Kooran's
advocating here. And the day we did this interview was
also an important anniversary.
Speaker 3 (23:26):
Today is the.
Speaker 1 (23:28):
Twenty nine of October, which is Arpennet Launch Day. Arponnet
is the precursor to the Internet, and in October twenty ninth,
sixty nine, the first message was sent over the arpennet.
Speaker 4 (23:41):
So the Arpaennet was basically a pre Internet network that
was developed as a collaboration between universities and the US
Department of Defense. Way back in the sixties.
Speaker 1 (23:50):
And it's interesting because like that version of the Internet
was designed to be resilient. It was designed to write
around problems or failure like the ones that we saw
with WS. It was designed to fundamentally like not have
choke points and concentration points.
Speaker 4 (24:08):
And so you're you're taking us back to ourpenet.
Speaker 1 (24:14):
I can't believe that that is where we Yeah, yeah,
it's full circle.
Speaker 4 (24:17):
No, but I mean, isn't that what's happening is I mean,
we're talking about universities without a profit motive, which is
really important here, without a profit motive, figuring out ways
to connect to each other with no singular choke point,
and then figuring out, okay, we I think we've got
assistant that works, everybody else copy it, but that is.
Speaker 1 (24:38):
Still operating from the assumption that we need to.
Speaker 3 (24:42):
Be moving to a world of more compute.
Speaker 1 (24:45):
And some of the people that I really take inspiration
from in the field are you know, in academics, researchers, colleagues,
but also like artists and organizers who are building technologies
but also ideas and also political worlds that provide actual alternatives.
(25:07):
And two sort of examples that that I think might
be interesting is the Critical Infrastructure Lab at the University
of Amsterdam and the Institute for Technology in the Public Interest,
which is based out of Brussels, and they are doing
super fascinating stuff. They are literally like building compostable data centers.
They are generating energy out of mud just to show
(25:28):
how hard it is to get energy and like how
much of it data centers consume. They are creating like
cloud abolitionists agendas, and that is the kind of thinking
and action that we also need. We need people will
actually question the premises and not just try and improve
what we have already.
Speaker 4 (25:47):
I don't know if this is recency bias, but we
set up this interview I wanted to speak to you
about AWS as we're talking right now, Microsoft's Azure is down.
Is this just recency bias or is this stuff happening
more frequently and are we going to be seeing this
happen more in the future.
Speaker 1 (26:04):
Yes, yes, and yes, I mean we do tend to
see the things more often when we're focused on them,
and this thing happens more often than people would like
to admit. The skill of the impact tends to generate
more or less attention depending on who it specifically hits
and yes, are we likely to see this more in
(26:24):
the future. Absolutely, Will the impact be bigger every time? Yes,
because one of the biggest things that we have on
our radar is that there are a lot of sectors
that are not yet digitized who are immediately going to
make the jump to the cloud. So whatever the next
iteration of this is, just because of that, the impact
(26:47):
will likely be even bigger than the last one.
Speaker 4 (26:49):
Yeah, I mean, I'm just thinking about how excited I
was when finally the trains and buses here in LA,
I was able to not have to carry around that
dumb card and I could use my app. And I
mean this is a real thing because also here in
LA they are introducing police at the entrance points. I
would very much like to not get harassed by the
(27:10):
police because all of a sudden my phone isn't working
to unlock my app so that I can get on
the train, and they think that I'm lying to them
that I have paid for my fare.
Speaker 2 (27:19):
Yep.
Speaker 4 (27:20):
It isn't just that these companies are all at risk
of experiencing another outage. A lot of Korn's work is
looking at all the ways that these companies control how
we live, our lives and what it means to have
so much power. I want to point out something that
AWS did that I think a lot of people maybe
we're a fan of. Actually, in twenty twenty one, AWS
(27:43):
pulled Parlor from being able to use their starvers. Just
for context here, Parlor think of it as it's another
social media app which was pretty explicitly right wing. Let's
just say that a lot of far right personalities have
pretty big followings there. And after January sixth, a little
(28:05):
bit after January sixth, AWS said, hey, listen, we're not
hosting this anymore. And I think they told Parlor that
have found ninety eight posts on the site that encouraged violence.
A lot of people said, good Amazon, doing the right thing.
Could that potentially cut both ways?
Speaker 1 (28:26):
Absolutely? At article nineteen. We have spent a decade documenting, researching,
and intervening in both corporate and government abuse of these
infrastructures to quell the scent, to make it harder to organize,
to make it harder for human rights defenders to speak up.
What this has shown us is the specific power dynamic
(28:48):
of how these cloud companies get to decide who gets
to use their services and by extension, who gets to
exercise their right to online? When so few companies control
the infrastructure, they also effectively control the flow of information,
and that has view implications for our ability to organize,
to find community, to access information, to exercise your freedom
(29:12):
of expression.
Speaker 4 (29:13):
Yeah, I mean so many people rely on signal signal,
relies on aws. Yeah, and what happens if for some
political reason, Amazon decides no more signal? Right now? I
think it's very reasonable to assume, at least at the stage,
(29:33):
they never do that. They never do that, Okay, what
if they do? Right?
Speaker 1 (29:36):
We need to just be prepared for these kind of
scenarios and make sure that these companies have a sense
of accountability for their role. It's the Spider Man quote,
it's a great power, great responsibility.
Speaker 4 (29:48):
So how do we hold that great responsibility? How do
we keep them accountable.
Speaker 1 (29:52):
As individual persons?
Speaker 3 (29:53):
It's very hard.
Speaker 1 (29:55):
Like I know that people said like, oh, you can
start self hosting all of your stuff, and like that's possible,
But then you know, try and self host your own
email right now and then get through the Gmail filters
just forget about it. But I do think there are
other stuffs that we can and should be taking Like
I want to give a shout out to my colleague,
doctor Isa Stasi, who's a lawyer at Article nineteen who
(30:15):
is currently taking Microsoft to court for unfair licensing practices. Right,
there are things that people can do. It's just it's
very hard, and you know, the entirety of the weight
of that corporate power comes sparing down on you. So
you need to have the kind of you know, backbone
to take on that backbone.
Speaker 4 (30:35):
So unless you're the head of some massive tech company
that's also using AWS or Microsoft Azure, Google Cloud, the
decision on de monopolizing the Internet probably isn't directly in
your hands. This is a big problem and it's not
something that any one of us as an individual is
going to fix. But at the very least, this is
another instance where we should be able to have some transparency.
(30:58):
Companies don't have to answer all of our questions, but
governments are a little different. We should be able to
ask our local officials what the hospital is depending on
to keep people alive, and what the backup plan is
when not if, but when that service goes offline next.
It's a small thing, but maybe that's one step toward
(31:19):
thinking about this stuff a little differently. Thank you so
much for listening to another episode of kill Switch. If
you want to talk to us, you could email us
at kill Switch at Kaleidoscope dot NYC, or we're also
on Instagram at kill Switch pod. And while you got
your phone out, you know, maybe leave us a review.
(31:39):
It helps other people find the show, which helps us
keep doing our thing. Kill Switch is hosted by Me
Dexter Thomas. It's produced by Shena Ozaki, Dar Luck Potts,
and Julia Nutter. Our theme song is by me and
Kyle Murdoch, and Kyle also mixes the show Kaleidoscope. Our
executive producers are Osra Lashin, Mangesh Hajikadur and k Osbourne
(32:01):
from iHeart. Our executive producers our Katrina Norville and Nikki
etoor Oh. And by the way, I asked Karin if
she had anything she wanted to add, and she said
she did.
Speaker 1 (32:12):
People in the past have asked me, like, you know,
what kind of power or force does it take to
push back on the sort of self propelling force of
these data centers? And I do think that, like coming
to the realization that we cannot keep growing our computing
industry like this without us becoming the victims of it
(32:34):
is really important. One of the big things that we
know about data centers is that it takes huge amounts
of energy, huge amounts of water, land, et cetera. So like,
even for the most sort of hard nos, like I'm
all about competition, and like I want to whatever make
money out of this, Like, yes, I'm sure you do.
(32:55):
Would you also like to be alive because those two
things in the long run not possible.
Speaker 4 (33:02):
If you want to hear more on that, we do
have an episode on the time that I tried to
figure out whether or not my embarrassingly prolific use of
chat GPT is better or worse than driving a gas
guzzling car. And we'll leave that as always in the
show notes. Anyway, catch all the next one.
We have breaking news for you this morning.
Speaker 2 (00:13):
A widespread Internet outage has caused major disruptions for some
of the biggest web services in the world.
Speaker 3 (00:18):
There's a good chance today that some website, app or
service that you tried to use today was down. That's
following an outage from Amazon Web Services.
Speaker 4 (00:26):
Amazon Web Services, the backbone of the Internet, suffered a
major outage that took hundreds of services offline across the globe.
Monday morning, October twentieth, Americans woke up to news of
a massive Internet outage, and it wasn't just America.
Speaker 1 (00:41):
I think one of the things to really note about
this outage is that it was one of the biggest
in recent history. In my decade in the field, this
definitely like ranks in the top five.
Speaker 4 (00:56):
Doctor Kreen caf is a culture anthropologist and she works
at the Human Rights organization Article nineteen. She researches Internet infrastructure, AI,
cloud computing, and basically how technology affects our day to
day lives.
Speaker 1 (01:08):
I have asked some of my colleagues, like veterans of
the field who've been in the industry since the Internet
was born, and they pretty much said the same thing,
like this was huge a.
Speaker 4 (01:20):
Lot of people didn't necessarily know that there was some
global internet outage going on. They just knew that there
was some website or service that they used that wasn't working,
like their favorite messaging app or even their smart bed.
Speaker 1 (01:32):
Some of these are kind of beIN benign or even funny,
like there were some excellent threads on the Internet about
people who had Internet connected mattresses.
Speaker 4 (01:44):
Oh the beds.
Speaker 2 (01:45):
Yes, yes, the beds. Oh my god, Apods. Shout out
to apods and everyone who thought it was a good
idea to buy an internet connected bed. I sort of
got this is why we don't have a single interconnected
device in our house.
Speaker 4 (01:57):
Krin's talking about these really expensive smart beds by this
company called eight Sleep. Their mattresses are supposed to track
your biometric data and get hotter or cooler, depending on
what your body needs. They're also connected to the Internet,
and apparently they didn't work offline after the outage started.
Some of the mattresses overheated or wouldn't recline. Thinking about
(02:17):
somebody's twenty five hundred dollars internet smartbed going vertical and
trying to cook them alive is kind of dystopian, but
it goes beyond that. This thing took down everything from
Amazon's own products like Ring doorbells and Alexa, to sites
like Reddit, to messaging apps like Signal, to government services.
These outages can impact basically every part of your life.
Speaker 1 (02:39):
This isn't just like me being on Duo Lingo or
me being on Snapchat. It is banking, It is the hospitals,
it is education, like it is things that matter in
my life, that matter in everyone's life.
Speaker 4 (02:57):
The world today relies on basically three major cloud comuting services.
When one of those goes down, so does a huge
chunk of the Internet. So what causes these outages? How
do we get to the point where so much of
the Internet relies on just three big tech companies? And
can that ever change any time? From Kaleidoscope and iHeart Podcast,
(03:26):
this is kill Switch. I'm Dexter Thomas.
Speaker 1 (03:37):
Goodbye.
Speaker 4 (03:43):
What exactly happened on October twentieth, So.
Speaker 1 (03:47):
What essentially happened is that there was an outage within
Amazon web Services. Now, Amazon Web Services is a leading
cloud computing company, and what is super important to know
is that Amazon currently has at least thirty percent of
the global cloud market share.
Speaker 4 (04:05):
We sometimes think of the cloud as something that lives
in this virtual space, not as something tangible. But the
cloud relies on very real and physical data centers that
are located around the world. These things store code and
data and move around that data to make the apps
that you use or the Internet run. And as Caring
(04:25):
just pointed out, about a third of the global Internet
relies on one company for cloud computing, Amazon Web Services.
Speaker 1 (04:33):
This company operates data centers worldwide that a lot of
the Internet and a lot of applications rely on. And
so what happened is within their data center clusters, one
of the oldest and biggest one had a bug in
the automation software that subsequently took down all.
Speaker 3 (04:54):
Of the companies that relied on that data center.
Speaker 4 (04:56):
Basically, the reason this outage was so widespread was because
it was an issue with one specific and heavily relied
upon data center that's located in Virginia.
Speaker 1 (05:06):
I've seen estimates of you know, a billion pins per
day going to this particular data center in Virginia, and
a pin is essentially like when a company needs something
from that database. And because this data center was one
of the biggest, some one of the oldest, about thirty
five to forty percent of all AWS traffic.
Speaker 3 (05:25):
Goes through it.
Speaker 1 (05:27):
Another reason why this impact was so big is if
you are a software developer and like you buy AWS
to develop your software on, this region is the default,
So like, if you don't switch to the default, this
is where your stuff will be housed.
Speaker 4 (05:46):
And so you have a third of the world's Internet
relying on this one company. Within that, there's this one
data center that becomes a default for all kinds of
new apps and websites. And then last month that data
center had a DNS error. The DNS, the domain name system,
translates a written website into an IP address that a
computer can then decode and understand. You might have heard
(06:07):
this described as the phone book of the Internet, but
Carin has a better explanation.
Speaker 3 (06:13):
So I'm going to.
Speaker 1 (06:14):
Try and do a different metaphor to explain this, because
the phone book is really useful in a sense that
what it does it connects names to numbers, right, and
that is what the domain name system does. So when
you type in a websites, you type in riverside dot FM,
not the very long number that is behind it, which
(06:35):
is actually how machines find each other. The problem with
the phone book metaphor is I've held one, I know
what they look like and what they feel like. But
most people younger than me, that is not a helpful
metaphor at all. So the way that I try to
explain what happened with AWS is imagine you're using Google
(06:56):
Maps or Ways or some sort of navigation app because
you want to go to the library because you need
a specific book to do your job. But essentially what
happens is there's a software bug and that causes your app,
your map app, to essentially indicate that the library no
longer exists. Even though the library is right there, it
is fully operational, but you just no longer can see
(07:18):
how to get there, and you're stuck with two problems,
the first one being is you cannot navigate to the library,
and secondly, you cannot access the information that you need,
and so you can't do your work. And that's essentially
what happened with the AWS outage. But the problem is,
like you're not the only one who needs the library.
The writer on your team needs access to it for research,
(07:42):
the accountant needs it for whatever financial records or software,
your lawyer needs it for the database. With case law, right,
But when the map says the library doesn't exist, it
pretty much doesn't show up, and so everyone's work stops simultaneously.
And what AWS figured out is like fixing that map
is not easy. You have to take all of these steps.
(08:05):
You have to identify which version of the mapping software
each person is using, push all of the updates to
all of the systems, wait for everything to restart, and
then verify that everyone can see the library again on
the map. And then you get this really interesting part
where when you fixed it and the library is reachable again,
there's a stampede because everyone realizes, like, oh, it's back up,
(08:26):
so we now all need to go there. And everyone
who's done like Black Friday shopping knows what that leads to.
And that is also why it took so long, right like,
because one thing is fixing the problem, and subsequently you
have to deal with the backlog of everyone trying to
get through the door simultaneously and the system not being
able to deal with that either.
Speaker 4 (08:44):
This isn't the first time the cloud services have had
an error like this, but because of how many applications
rely not only on AWS, but on this one data center,
the effects of the outage were really widespread, and outages
of this scale can cause some very real problem.
Speaker 1 (09:00):
This is not the first outage, not the first and
big one, right. The last big one was on the
nineteenth of July twenty twenty four, And the reason why
I remember that date so precisely is because it happens
to be the day that my second born was born,
and for reasons I had to be rushed to the
hospital and it was like urgent.
Speaker 3 (09:20):
This was no joke.
Speaker 1 (09:22):
And that was another day that there was a massive outage,
but this time it was CrowdStrike Microsoft Cloud.
Speaker 4 (09:27):
In this incident last year in twenty twenty four, what
happened was a faulty update by the security company CrowdStrike
caused a major outage for Microsoft's Azure cloud services. This
one brought down airlines, banks, and even emergency services.
Speaker 1 (09:42):
The massive Microsoft outage impacting businesses around the world, taking
some companies temporarily offline and forcing major American airlines to
ground flights. The outage affected government entities and news organizations, as.
Speaker 4 (09:55):
Well as almost every industry across the globe, including banks, airports,
healthcare providers, and retailers.
Speaker 1 (10:03):
There were a number of hospitals in the Netherlands where
I'm based, where those issues were such that they couldn't
help patients. Now luckily that wasn't in my region. But
when you're in an ambulance being sped to a hospital, you
don't want to show up there on the other end
and then telling you I'm sorry, computer says no. You know,
(10:24):
it's the apps and it's the benign stuff, but it's
also the very serious stuff that we have connected to
these clouds. Healthcare, education, banking, access to your government. These
are all things that are so critical for people's lives.
Like what happens when that is not accessible? I mean,
I am sure that there are people who missed funerals,
(10:45):
who were not able to make really crucial payments that
needed to be made that day, that didn't get whatever
benefits that they really needed to do. This is not
an incident because it keeps happening. When this happened, like
the organization that I work for nineteen like, we had
a blogout within two hours at best, and people were like, wow,
(11:06):
you guys are so fast, like how did you put
that together? And I'm like, well, it's kind of depressing
because this is the fourth time in four years that
I see in outage quite like this, And all I
have to do every time is I switch out the
name of the company and I look into the specifics
of where they feel technically, and then it's the same
up D and so we just you know, here we are.
Speaker 4 (11:29):
That is depressing. That is depressing. It's wild too, you
said it keeps happening. You might already know this, but
I asked you on to talk about aws. Do you
know Minecraft is down right now?
Speaker 2 (11:41):
Oh?
Speaker 3 (11:41):
I did not see that.
Speaker 1 (11:43):
Yeah, preparing for talking to you guys and not.
Speaker 4 (11:46):
Not doing your job, I kept you from doing your job. Yeah, No,
I think Xbox is down right now. Microsoft's Azure was
once again experiencing an outage. At that point in the interview,
all I'd seen was that Xbox and Minecraft were down.
But as always, it's bigger than that. It turns out
banks got hit with a two and so did parts
of the train system, which was a problem because it
(12:08):
also happened to be election day in the Netherlands, where
Krin is, and the disruptions and Dutch railways meant that
some people might not be able to make at home
in time to vote. So the outage was directly interfering
with the democratic process. Microsoft's share of cloud computing isn't
as big as Amazon's, but it's a close second, and
together with Google, these three companies make up almost two
(12:29):
thirds of the entire cloud infrastructure, meaning that two thirds
of the Internet relies on Google, Amazon, and Microsoft keeping
their act together. So how did we get here after
the break the origin story of Amazon Web Services? How
(12:54):
do we get to the point where the bookseller became
the backbone of most of the Internet.
Speaker 1 (12:59):
So there are more multiple origin stories around, like how
AWS became AWS. A lot of the work that people
put out there now like very much follows aws's iteration
of events. You know, we were an e commerce platform.
We needed a lot of compute to be able to
(13:19):
do that. And then at one point they realized, like, oh,
in addition to selling the books, we can sell whatever
compute capacity we don't need ourselves.
Speaker 4 (13:29):
In this version of the story, Amazon Web Services kind
of started as a side hustle with Amazon renting out
the extra computer capacity that they had to other companies.
Fortune magazine did an interview with Amazon's current CEO, Andy Jasse,
who started out as the head of Amazon Web Services.
He said that in the early two thousands, Amazon was
starting to run in the walls when they were developing
(13:50):
features for the site. Remember at this point they were
still mainly thought of as a bookseller, so they started
building out internal features for their own programmers to use,
and they realized that they could sell that infrastructure to
other companies.
Speaker 1 (14:04):
So that is like the long short of it. There's
a lot more nuanced there, and I think some really
great Internet historians are going to be picking through that
to bring the inside story. What is definitely true is
that they were the first.
Speaker 4 (14:15):
In two thousand and six, Amazon Web Services was born,
and it was instantly popular. They were solving for a
problem that companies were already starting to have before this.
Starting an app company could be really risky, partially because
of how unpredictable the traffic demands could be. Let's say
you start up a company and your online service just
goes viral overnight. Well, if you didn't buy enough service
(14:38):
ahead of time, you can't keep up with demand and
you're screwed. But if you did buy a bunch of
servers and your company isn't doing so well, well, now
you're stuck with a building full of servers and you're
wasting money, and now you're really screwed. What AWS did
was let companies rent that computing space and power as
opposed to having to buy it outright themselves. Basically, they
(14:58):
let them do what a lot of us do, pay
Amazon a subscription fee. If your business gets popular, just
rent more space. If you have a slow month or two,
you can scale it back. And on top of that,
AWS had software development tools, which meant that you could
build on top of those tools rather than having to
start to scratch. This was huge for the web two
point zero boom. It allowed for a lot of experimentation
(15:21):
without as much financial risk, which in some ways felt
like a good thing, but it also meant that the
Internet became really reliant on Amazon.
Speaker 1 (15:32):
Companies have gone from largely running their own servers like
physical computers in their own garages, so to speak, to
relying on the servers of AWS. But maybe even more importantly,
like we've also gone from individual companies building software from
scratch to compiling it from various building blocks provided by
(15:53):
other software companies.
Speaker 3 (15:56):
What you then get.
Speaker 1 (15:57):
Is that you create this really complex web of dependencies
and interconnections where like every piece of software relies on
another piece of software which relies on another piece of software.
Speaker 4 (16:10):
However it might have started, AWS is not a side hustle.
Now it brought in almost forty billion dollars in operating
income last year. That's over half of what Amazon made.
Amazon beat Google and Microsoft to the punch here, but
both those companies eventually offered their own competing services, and
now we have this environment of three companies controlling the
(16:30):
cloud that houses the Internet, and what was first just
an extension of basically using someone else's computer for storage
has turned into something much much bigger.
Speaker 1 (16:42):
I think that is the next sort of step that
we need to take, where it's like, in what situation
would you accept, you know, only three companies being responsible
for all of agriculture. We would immediately see how big
of a choke hold that would be. Because we understand
(17:03):
how important food is for our daily sustenance, we need
to start changing our perception of what the digital is,
because it is everything.
Speaker 4 (17:13):
The food comparison is really interesting, Like you want a tomato,
you want a hamburger, you want, I don't know, man,
some spaghetti, all of those. I don't know why I
thought of those three that maybe I'm hungry right now.
But all of those three things are you know, if
(17:34):
one company goes down, if something is one of those
companies has a problem, if one of those companies is
not safe for some reason, starts slacking on safety regulations
or whatever, you just you don't get one of those
three or two of those three, or all of those
three or whatever. You don't eat, man, Yeah, you just
don't eat.
Speaker 1 (17:52):
There is a reason why these three keep getting bigger
because they are also the only ones with like that
amount of money to put into something that loses its
value quite quickly, right like servers infrastructure. It's expensive to build,
it's expensive to maintain, and you have to switch it
out every five to seven years. I've spoken to vcs
(18:14):
for research purposes just get a sense of you know,
when you get a startup and you look at their proposition,
and for instance, they would say like, oh, actually, we
either want to build our own set of servers, our
own mini cloud, or we would want to work with
multiple partners. And these vcs just like burst out laughing
at just the proposition of that, because they were like, well,
(18:37):
if I'm putting my money into this and they're going
to spend it on like something that gives me an
overturn when they could just do it with AWS or
whomever's going to do it cheaper and better, at least
from their perspective, I was just never fund I would
never fund something that would also require my money to
put into machines.
Speaker 3 (18:57):
That are going to die in the next five years.
It's a bad investment right now.
Speaker 4 (19:01):
Smaller companies really can't compete with these giants. So you
could argue here that it's better that we have these
big tech companies monopolizing this. They have more resources, and
they can afford to maintain the data centers and keep
them working, and maybe that is a way to keep
things cheap and reliable. But for Corent, that's not a
good enough reason.
Speaker 1 (19:23):
The thing that I think is important to remember is
that these technical outages have political consequences. So it is
really important to politicize what is happening here, because it's
very easy for people to say like, oh, it's just
a technical mistake, and we should be allowed to make
technical mistakes. But that is not an argument that we
(19:46):
would accept from, you know, the company that operates our
water or our electricity, for them to just be like,
I'm sorry, it's winter, but for the next twelve hours
no heat in your house or no clean water, so
your kids are thirsty. Too bad. And somehow with these companies,
we've decided that they get a pass or like maybe
(20:07):
even better, like they've just lobbied such that there isn't
that much or enough at least from my perspective oversight
on you know, them making these mistakes and that having
really potentially life altering consequences.
Speaker 3 (20:22):
And that I very much see as a democratic deficit.
Speaker 4 (20:26):
So what can we do about this? How do we
take back computing power from the hands of a few
big corporations. That's after the break The same week as
the AWS outage last month, political leaders in Europe met
(20:47):
in Brussels for a summit, and one of the items
on the agenda was this idea of European digital sovereignty,
basically figuring out a way to untangle Europe from being
so reliant on these big US tech companies.
Speaker 1 (21:00):
I'm actually quite critical of people who are very focused
on the European digital sovereignty because I'm like, well, you know,
the world's a big place, and you just you know
what ford Europe is where you put your digital borders,
like that's not good enough. The perspective needs to be.
Speaker 3 (21:14):
One of global solidarity always.
Speaker 1 (21:17):
And I have been in the room with politicians representing
you know, the Greens or left wing parties back home,
who without irony, have said, oh, well, you know, we
need to start looking for alternatives. But the only ones
that are perhaps available to us, or that have experience
(21:38):
running these kind of networks and that kind of skill
are defense contractors. And I would be sitting there picking
my job from the floor, being like, wait, the choice
that we are giving ourselves here is AWS or defense contractors, Right,
that is a rock and a hard place. We have
(22:00):
to think of something else that.
Speaker 4 (22:01):
Is not this Instead of looking to Silicon Valley or
the military for the answer, Corent says, we should be
thinking about nonprofits and universities for ideas here.
Speaker 1 (22:10):
Now, it's not impossible. Wikipedia does it. They specifically have
a setup that has them. I want to say, like
across seven different data centers across numerous continents, and none
of them are the Big Three. But that would be
much harder to do for something like signal right, where
latency and like constant uptime is much more visible when
(22:32):
it doesn't work.
Speaker 3 (22:33):
Then with something like Wikipedia.
Speaker 1 (22:35):
There are alternatives, but it requires us to also change
our expectations, and it requires the people more importantly, like
who put money into this, who will literally have skin
in the game, to change theirs. So it's cultural change
as well. It's not only technical. Locally, we work with
small medium enterprises to see, like, you know, what they
(22:56):
are thinking of the universities in the Netherlands of taking
some really interesting steps where they have all come together
with like local providers to say like, okay, well we
are roughly the same kind of institute. Here are the
needs that we all have. Could you build us something
that is like fit for purpose for universities.
Speaker 4 (23:19):
So there is a kind of precedent for what Kooran's
advocating here. And the day we did this interview was
also an important anniversary.
Speaker 3 (23:26):
Today is the.
Speaker 1 (23:28):
Twenty nine of October, which is Arpennet Launch Day. Arponnet
is the precursor to the Internet, and in October twenty ninth,
sixty nine, the first message was sent over the arpennet.
Speaker 4 (23:41):
So the Arpaennet was basically a pre Internet network that
was developed as a collaboration between universities and the US
Department of Defense. Way back in the sixties.
Speaker 1 (23:50):
And it's interesting because like that version of the Internet
was designed to be resilient. It was designed to write
around problems or failure like the ones that we saw
with WS. It was designed to fundamentally like not have
choke points and concentration points.
Speaker 4 (24:08):
And so you're you're taking us back to ourpenet.
Speaker 1 (24:14):
I can't believe that that is where we Yeah, yeah,
it's full circle.
Speaker 4 (24:17):
No, but I mean, isn't that what's happening is I mean,
we're talking about universities without a profit motive, which is
really important here, without a profit motive, figuring out ways
to connect to each other with no singular choke point,
and then figuring out, okay, we I think we've got
assistant that works, everybody else copy it, but that is.
Speaker 1 (24:38):
Still operating from the assumption that we need to.
Speaker 3 (24:42):
Be moving to a world of more compute.
Speaker 1 (24:45):
And some of the people that I really take inspiration
from in the field are you know, in academics, researchers, colleagues,
but also like artists and organizers who are building technologies
but also ideas and also political worlds that provide actual alternatives.
(25:07):
And two sort of examples that that I think might
be interesting is the Critical Infrastructure Lab at the University
of Amsterdam and the Institute for Technology in the Public Interest,
which is based out of Brussels, and they are doing
super fascinating stuff. They are literally like building compostable data centers.
They are generating energy out of mud just to show
(25:28):
how hard it is to get energy and like how
much of it data centers consume. They are creating like
cloud abolitionists agendas, and that is the kind of thinking
and action that we also need. We need people will
actually question the premises and not just try and improve
what we have already.
Speaker 4 (25:47):
I don't know if this is recency bias, but we
set up this interview I wanted to speak to you
about AWS as we're talking right now, Microsoft's Azure is down.
Is this just recency bias or is this stuff happening
more frequently and are we going to be seeing this
happen more in the future.
Speaker 1 (26:04):
Yes, yes, and yes, I mean we do tend to
see the things more often when we're focused on them,
and this thing happens more often than people would like
to admit. The skill of the impact tends to generate
more or less attention depending on who it specifically hits
and yes, are we likely to see this more in
(26:24):
the future. Absolutely, Will the impact be bigger every time? Yes,
because one of the biggest things that we have on
our radar is that there are a lot of sectors
that are not yet digitized who are immediately going to
make the jump to the cloud. So whatever the next
iteration of this is, just because of that, the impact
(26:47):
will likely be even bigger than the last one.
Speaker 4 (26:49):
Yeah, I mean, I'm just thinking about how excited I
was when finally the trains and buses here in LA,
I was able to not have to carry around that
dumb card and I could use my app. And I
mean this is a real thing because also here in
LA they are introducing police at the entrance points. I
would very much like to not get harassed by the
(27:10):
police because all of a sudden my phone isn't working
to unlock my app so that I can get on
the train, and they think that I'm lying to them
that I have paid for my fare.
Speaker 2 (27:19):
Yep.
Speaker 4 (27:20):
It isn't just that these companies are all at risk
of experiencing another outage. A lot of Korn's work is
looking at all the ways that these companies control how
we live, our lives and what it means to have
so much power. I want to point out something that
AWS did that I think a lot of people maybe
we're a fan of. Actually, in twenty twenty one, AWS
(27:43):
pulled Parlor from being able to use their starvers. Just
for context here, Parlor think of it as it's another
social media app which was pretty explicitly right wing. Let's
just say that a lot of far right personalities have
pretty big followings there. And after January sixth, a little
(28:05):
bit after January sixth, AWS said, hey, listen, we're not
hosting this anymore. And I think they told Parlor that
have found ninety eight posts on the site that encouraged violence.
A lot of people said, good Amazon, doing the right thing.
Could that potentially cut both ways?
Speaker 1 (28:26):
Absolutely? At article nineteen. We have spent a decade documenting, researching,
and intervening in both corporate and government abuse of these
infrastructures to quell the scent, to make it harder to organize,
to make it harder for human rights defenders to speak up.
What this has shown us is the specific power dynamic
(28:48):
of how these cloud companies get to decide who gets
to use their services and by extension, who gets to
exercise their right to online? When so few companies control
the infrastructure, they also effectively control the flow of information,
and that has view implications for our ability to organize,
to find community, to access information, to exercise your freedom
(29:12):
of expression.
Speaker 4 (29:13):
Yeah, I mean so many people rely on signal signal,
relies on aws. Yeah, and what happens if for some
political reason, Amazon decides no more signal? Right now? I
think it's very reasonable to assume, at least at the stage,
(29:33):
they never do that. They never do that, Okay, what
if they do? Right?
Speaker 1 (29:36):
We need to just be prepared for these kind of
scenarios and make sure that these companies have a sense
of accountability for their role. It's the Spider Man quote,
it's a great power, great responsibility.
Speaker 4 (29:48):
So how do we hold that great responsibility? How do
we keep them accountable.
Speaker 1 (29:52):
As individual persons?
Speaker 3 (29:53):
It's very hard.
Speaker 1 (29:55):
Like I know that people said like, oh, you can
start self hosting all of your stuff, and like that's possible,
But then you know, try and self host your own
email right now and then get through the Gmail filters
just forget about it. But I do think there are
other stuffs that we can and should be taking Like
I want to give a shout out to my colleague,
doctor Isa Stasi, who's a lawyer at Article nineteen who
(30:15):
is currently taking Microsoft to court for unfair licensing practices. Right,
there are things that people can do. It's just it's
very hard, and you know, the entirety of the weight
of that corporate power comes sparing down on you. So
you need to have the kind of you know, backbone
to take on that backbone.
Speaker 4 (30:35):
So unless you're the head of some massive tech company
that's also using AWS or Microsoft Azure, Google Cloud, the
decision on de monopolizing the Internet probably isn't directly in
your hands. This is a big problem and it's not
something that any one of us as an individual is
going to fix. But at the very least, this is
another instance where we should be able to have some transparency.
(30:58):
Companies don't have to answer all of our questions, but
governments are a little different. We should be able to
ask our local officials what the hospital is depending on
to keep people alive, and what the backup plan is
when not if, but when that service goes offline next.
It's a small thing, but maybe that's one step toward
(31:19):
thinking about this stuff a little differently. Thank you so
much for listening to another episode of kill Switch. If
you want to talk to us, you could email us
at kill Switch at Kaleidoscope dot NYC, or we're also
on Instagram at kill Switch pod. And while you got
your phone out, you know, maybe leave us a review.
(31:39):
It helps other people find the show, which helps us
keep doing our thing. Kill Switch is hosted by Me
Dexter Thomas. It's produced by Shena Ozaki, Dar Luck Potts,
and Julia Nutter. Our theme song is by me and
Kyle Murdoch, and Kyle also mixes the show Kaleidoscope. Our
executive producers are Osra Lashin, Mangesh Hajikadur and k Osbourne
(32:01):
from iHeart. Our executive producers our Katrina Norville and Nikki
etoor Oh. And by the way, I asked Karin if
she had anything she wanted to add, and she said
she did.
Speaker 1 (32:12):
People in the past have asked me, like, you know,
what kind of power or force does it take to
push back on the sort of self propelling force of
these data centers? And I do think that, like coming
to the realization that we cannot keep growing our computing
industry like this without us becoming the victims of it
(32:34):
is really important. One of the big things that we
know about data centers is that it takes huge amounts
of energy, huge amounts of water, land, et cetera. So like,
even for the most sort of hard nos, like I'm
all about competition, and like I want to whatever make
money out of this, Like, yes, I'm sure you do.
(32:55):
Would you also like to be alive because those two
things in the long run not possible.
Speaker 4 (33:02):
If you want to hear more on that, we do
have an episode on the time that I tried to
figure out whether or not my embarrassingly prolific use of
chat GPT is better or worse than driving a gas
guzzling car. And we'll leave that as always in the
show notes. Anyway, catch all the next one.
kill switch News
Hosts And Creators
Oz Woloshyn
Karah Preiss
Show Links
AboutPopular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
Las Culturistas with Matt Rogers and Bowen Yang
Ding dong! Join your culture consultants, Matt Rogers and Bowen Yang, on an unforgettable journey into the beating heart of CULTURE. Alongside sizzling special guests, they GET INTO the hottest pop-culture moments of the day and the formative cultural experiences that turned them into Culturistas. Produced by the Big Money Players Network and iHeartRadio.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com