WEBVTT

99:59:59.999 --> 99:59:59.999
Chris Anderson:
What worries you right now?

99:59:59.999 --> 99:59:59.999
You've been very open
about lots of issues on Twitter.

99:59:59.999 --> 99:59:59.999
What would be your top worry

99:59:59.999 --> 99:59:59.999
about where things are right now?

99:59:59.999 --> 99:59:59.999
Jack Dorsey: Right now,
the health of the conversation.

99:59:59.999 --> 99:59:59.999
So our purpose is to serve
the public conversation,

99:59:59.999 --> 99:59:59.999
and we have seen
a number of attacks on it.

99:59:59.999 --> 99:59:59.999
We have seen abuse, we have harassment,

99:59:59.999 --> 99:59:59.999
we've seen manipulation,

99:59:59.999 --> 99:59:59.999
automation, human coordination,

99:59:59.999 --> 99:59:59.999
misinformation.

99:59:59.999 --> 99:59:59.999
So these are all dynamics
that we were not expecting

99:59:59.999 --> 99:59:59.999
13 years ago when we
were starting the company,

99:59:59.999 --> 99:59:59.999
but we do now see them at scale,

99:59:59.999 --> 99:59:59.999
and what worries me most

99:59:59.999 --> 99:59:59.999
is just our ability to address it
in a systemic way that is scalable,

99:59:59.999 --> 99:59:59.999
that has a rigorous understanding
of how we're taking action,

99:59:59.999 --> 99:59:59.999
a transparent understanding
of how we're taking action,

99:59:59.999 --> 99:59:59.999
and a rigorous appeals process
for when we're wrong,

99:59:59.999 --> 99:59:59.999
because we will be wrong.

99:59:59.999 --> 99:59:59.999
Whitney Pennington Rodgers:
And I'm really glad to hear

99:59:59.999 --> 99:59:59.999
that that's something that concerns you,

99:59:59.999 --> 99:59:59.999
because I think there's been
a lot written about people

99:59:59.999 --> 99:59:59.999
who feel they've been abused
and harassed on Twitter,

99:59:59.999 --> 99:59:59.999
and I think no one more so
than women and women of color

99:59:59.999 --> 99:59:59.999
and black women

99:59:59.999 --> 99:59:59.999
and there's been data that's come out,

99:59:59.999 --> 99:59:59.999
Amnesty International put out
a report a few months ago

99:59:59.999 --> 99:59:59.999
where they showed that a subset
of active black female Twitter users

99:59:59.999 --> 99:59:59.999
were receiving on average
one in 10 of their tweets

99:59:59.999 --> 99:59:59.999
were some form of harassment.

99:59:59.999 --> 99:59:59.999
And so when you think about health
for the community on Twitter,

99:59:59.999 --> 99:59:59.999
I'm interested to hear,
health for everyone,

99:59:59.999 --> 99:59:59.999
but specifically how are you looking
to make Twitter a safe space

99:59:59.999 --> 99:59:59.999
for that subset, for women,
for women of color and black women.

99:59:59.999 --> 99:59:59.999
JD: Yeah.

99:59:59.999 --> 99:59:59.999
So it's a pretty terrible situation

99:59:59.999 --> 99:59:59.999
when you're coming to a service

99:59:59.999 --> 99:59:59.999
that ideally you want to learn
something about the world

99:59:59.999 --> 99:59:59.999
and you spend the majority of your time

99:59:59.999 --> 99:59:59.999
reporting abuse, receiving abuse,

99:59:59.999 --> 99:59:59.999
receiving harassment.

99:59:59.999 --> 99:59:59.999
So what we're looking most deeply at
is just the incentives

99:59:59.999 --> 99:59:59.999
that the platform naturally provides
and the service provides.

99:59:59.999 --> 99:59:59.999
Right now the dynamic of the system
makes it super-easy to harass

99:59:59.999 --> 99:59:59.999
and to abuse others through the service,

99:59:59.999 --> 99:59:59.999
and unfortunately the majority
of our system in the past

99:59:59.999 --> 99:59:59.999
worked entirely based on people
reporting harassment and abuse.

99:59:59.999 --> 99:59:59.999
So about midway last year we decided
that we were going to apply

99:59:59.999 --> 99:59:59.999
a lot more machine learning,
a lot more deep learning to the problem

99:59:59.999 --> 99:59:59.999
and try to be a lot more proactive
around where abuse is happening

99:59:59.999 --> 99:59:59.999
so that we can take the burden
off the victim completely.

99:59:59.999 --> 99:59:59.999
And we've made some progress recently.

99:59:59.999 --> 99:59:59.999
About 38 percent of abusive tweets

99:59:59.999 --> 99:59:59.999
are now proactively identified
by machine learning algorithms

99:59:59.999 --> 99:59:59.999
so that people don't actually
have to report them,

99:59:59.999 --> 99:59:59.999
but those that are identified
are still reviewed by humans,

99:59:59.999 --> 99:59:59.999
so we do not take down content or accounts
without a human actually reviewing it.

99:59:59.999 --> 99:59:59.999
But that was from zero percent
just a year ago.

99:59:59.999 --> 99:59:59.999
So that meant, at that zero percent,

99:59:59.999 --> 99:59:59.999
every single person who received abuse
had to actually report it,

99:59:59.999 --> 99:59:59.999
which was a lot of work for them,
a lot of work for us,

99:59:59.999 --> 99:59:59.999
and just ultimately unfair.

99:59:59.999 --> 99:59:59.999
The other thing that we're doing
is making sure that we, as a company,

99:59:59.999 --> 99:59:59.999
have representation of all the
communities that we're trying to serve.

99:59:59.999 --> 99:59:59.999
We can't build a business
that is successful

99:59:59.999 --> 99:59:59.999
unless we have a diversity
of perspective inside of our walls

99:59:59.999 --> 99:59:59.999
that actually feel these issues
every single day,

99:59:59.999 --> 99:59:59.999
and that's not just with the team
that's doing the work,

99:59:59.999 --> 99:59:59.999
it's also within our leadership as well.

99:59:59.999 --> 99:59:59.999
So we need to continue to build empathy
for what people are experiencing

99:59:59.999 --> 99:59:59.999
and give them better tools to act on it

99:59:59.999 --> 99:59:59.999
and also give our customers
a much better and easier approach

99:59:59.999 --> 99:59:59.999
to handle some of the things
that they're seeing.

99:59:59.999 --> 99:59:59.999
So a lot of what we're doing
is around technology,

99:59:59.999 --> 99:59:59.999
but we're also looking at
the incentives on the service.

99:59:59.999 --> 99:59:59.999
What does Twitter incentivize you to do
when you first open it up?

99:59:59.999 --> 99:59:59.999
And in the past, it's incentivized
a lot of mob behavior,

99:59:59.999 --> 99:59:59.999
it's incentivized a lot
of group harassment,

99:59:59.999 --> 99:59:59.999
and we have to look a lot deeper
at some of the fundamentals

99:59:59.999 --> 99:59:59.999
of what the service is doing
to make the bigger shifts.

99:59:59.999 --> 99:59:59.999
We can make a bunch of small shifts
around technology, as I just described,

99:59:59.999 --> 99:59:59.999
but ultimately we have to look deeply
at the dynamics in the network itself,

99:59:59.999 --> 99:59:59.999
and that's what we're doing right now.

99:59:59.999 --> 99:59:59.999
CA: And what's your sense?

99:59:59.999 --> 99:59:59.999
What is the kind of thing
that you might be able to change

99:59:59.999 --> 99:59:59.999
that would actually
fundamentally shift behavior?

99:59:59.999 --> 99:59:59.999
JD: Well, one of the things,
we started the service

99:59:59.999 --> 99:59:59.999
with this concept of following an account,

99:59:59.999 --> 99:59:59.999
as an example,

99:59:59.999 --> 99:59:59.999
and I don't believe that's why
people actually come to Twitter.

99:59:59.999 --> 99:59:59.999
I believe Twitter is best
as an interest-based network.

99:59:59.999 --> 99:59:59.999
People come with a particular interest.

99:59:59.999 --> 99:59:59.999
They have to do a ton of work to find
and follow the related accounts

99:59:59.999 --> 99:59:59.999
around those interests.

99:59:59.999 --> 99:59:59.999
What we could do instead
is allow you to follow an interest,

99:59:59.999 --> 99:59:59.999
follow a hashtag, follow a trend,

99:59:59.999 --> 99:59:59.999
follow a community,

99:59:59.999 --> 99:59:59.999
which gives us the opportunity
to show all of the accounts,

99:59:59.999 --> 99:59:59.999
all the topics, all the moments,
all the hashtags

99:59:59.999 --> 99:59:59.999
that are associated with that
particular topic and interest,

99:59:59.999 --> 99:59:59.999
which really opens up
the perspective that you see.

99:59:59.999 --> 99:59:59.999
But that is a huge fundamental shift

99:59:59.999 --> 99:59:59.999
to bias the entire network
away from just an account bias

99:59:59.999 --> 99:59:59.999
towards a topics and interest bias.

99:59:59.999 --> 99:59:59.999
CA: Because isn't it the case

99:59:59.999 --> 99:59:59.999
that one reason why you have
so much content on there

99:59:59.999 --> 99:59:59.999
is a result of putting millions
of people around the world

99:59:59.999 --> 99:59:59.999
in this kind of gladiatorial
contest with each other

99:59:59.999 --> 99:59:59.999
for followers, for attention?

99:59:59.999 --> 99:59:59.999
Like, from the point of view
of people who just read Twitter,

99:59:59.999 --> 99:59:59.999
that's not an issue, but for
the people who actually create it,

99:59:59.999 --> 99:59:59.999
everyone's out there saying,

99:59:59.999 --> 99:59:59.999
"You know, I wish I had
a few more likes, followers, retweets."

99:59:59.999 --> 99:59:59.999
And so they're constantly experiment,

99:59:59.999 --> 99:59:59.999
trying to find the path to do that,

99:59:59.999 --> 99:59:59.999
and what we've all discovered
is that the number one path to do that

99:59:59.999 --> 99:59:59.999
is to be some form of provocative,

99:59:59.999 --> 99:59:59.999
obnoxious, eloquently obnoxious,

99:59:59.999 --> 99:59:59.999
like eloquent insults
are a dream on Twitter,

99:59:59.999 --> 99:59:59.999
where you rapidly pile up,

99:59:59.999 --> 99:59:59.999
and it becomes this self-fueling
process of driving outrage.

99:59:59.999 --> 99:59:59.999
How do you defuse that?

99:59:59.999 --> 99:59:59.999
JD: Yeah, I mean, I think you're spot on,

99:59:59.999 --> 99:59:59.999
but that goes back to the incentives.

99:59:59.999 --> 99:59:59.999
Like, one of the choices
we made in the early days

99:59:59.999 --> 99:59:59.999
was we had this number

99:59:59.999 --> 99:59:59.999
that showed how many people follow you.

99:59:59.999 --> 99:59:59.999
We decided that number
should be big and bold

99:59:59.999 --> 99:59:59.999
and anything that's on the page
that's big and bold has importance,

99:59:59.999 --> 99:59:59.999
and those are the things
that you want to drive.

99:59:59.999 --> 99:59:59.999
Was that the right decision at the time?

99:59:59.999 --> 99:59:59.999
Probably not.

99:59:59.999 --> 99:59:59.999
If I had to start the service again,

99:59:59.999 --> 99:59:59.999
I would not emphasize
the follower account as much.

99:59:59.999 --> 99:59:59.999
I would not emphasize
the like count as much.

99:59:59.999 --> 99:59:59.999
I don't think I would even
create like in the first place,

99:59:59.999 --> 99:59:59.999
because it doesn't actually push

99:59:59.999 --> 99:59:59.999
what we believe now
to be the most important thing,

99:59:59.999 --> 99:59:59.999
which is health contribution
back to the network,

99:59:59.999 --> 99:59:59.999
and conversation to the network,

99:59:59.999 --> 99:59:59.999
participation within conversation,

99:59:59.999 --> 99:59:59.999
learning something from the conversation.

99:59:59.999 --> 99:59:59.999
Those are not things
that we thought of 13 years ago

99:59:59.999 --> 99:59:59.999
and we believe are extremely
important right now.

99:59:59.999 --> 99:59:59.999
So we have to look at
how we display the follower account,

99:59:59.999 --> 99:59:59.999
how we display retweet count,

99:59:59.999 --> 99:59:59.999
how we display likes,

99:59:59.999 --> 99:59:59.999
and just ask the deep question:

99:59:59.999 --> 99:59:59.999
is this really the number
that we want people to drive up?

99:59:59.999 --> 99:59:59.999
Is this the thing that,
when you open Twitter,

99:59:59.999 --> 99:59:59.999
you see, "That's the thing
I need to increase?"

99:59:59.999 --> 99:59:59.999
And I don't believe
that's the case right now.

99:59:59.999 --> 99:59:59.999
(Applause)

99:59:59.999 --> 99:59:59.999
WPR: I think look at some of the tweets

99:59:59.999 --> 99:59:59.999
that are coming in
from the audience as well.

99:59:59.999 --> 99:59:59.999
CA: Let's see what you guys are asking.

99:59:59.999 --> 99:59:59.999
I mean, this is generally one of
the amazing things about Twitter

99:59:59.999 --> 99:59:59.999
is how you can use it for crowd wisdom.

99:59:59.999 --> 99:59:59.999
You know,

99:59:59.999 --> 99:59:59.999
that more knowledge, more questions,
more points of view than you can imagine,

99:59:59.999 --> 99:59:59.999
and sometimes many of them really healthy.

99:59:59.999 --> 99:59:59.999
WPR: And I think one that I saw
that passed already quick down here,

99:59:59.999 --> 99:59:59.999
"What's Twitter's plan to combat
foreign meddling in the 2020 US election?"

99:59:59.999 --> 99:59:59.999
And I think that's something
that's an issue we're seeing

99:59:59.999 --> 99:59:59.999
on the internet in general,

99:59:59.999 --> 99:59:59.999
that we have a lot of malicious
automated activity happening.

99:59:59.999 --> 99:59:59.999
And on Twitter, for example,

99:59:59.999 --> 99:59:59.999
we have in fact some work

99:59:59.999 --> 99:59:59.999
that's come from our friends
at Zignal Labs, rather,

99:59:59.999 --> 99:59:59.999
and maybe we can even see that
to give us an example

99:59:59.999 --> 99:59:59.999
of what exactly I'm talking about,

99:59:59.999 --> 99:59:59.999
where you have these bots, if you will,

99:59:59.999 --> 99:59:59.999
or coordinated automated
malicious account activity,

99:59:59.999 --> 99:59:59.999
that is being used to influence
things like elections.

99:59:59.999 --> 99:59:59.999
And in this example we have
from Zignal which they've shared with us

99:59:59.999 --> 99:59:59.999
using the data that
they have from Twitter,

99:59:59.999 --> 99:59:59.999
you actually see that in this case,

99:59:59.999 --> 99:59:59.999
white represents the humans,

99:59:59.999 --> 99:59:59.999
human accounts,
each dot is an account.

99:59:59.999 --> 99:59:59.999
The pinker it is,

99:59:59.999 --> 99:59:59.999
the more automated the activity is,

99:59:59.999 --> 99:59:59.999
and you can see how you have
a few humans interacting with bots,

99:59:59.999 --> 99:59:59.999
and in this case it's related
to the election in Israel,

99:59:59.999 --> 99:59:59.999
and spreading misinformation
about Benny Gantz,

99:59:59.999 --> 99:59:59.999
and as we know in the end,
that was an election

99:59:59.999 --> 99:59:59.999
that Netanyahu won by a slim margin,

99:59:59.999 --> 99:59:59.999
and that may have been
in some case influenced by this,

99:59:59.999 --> 99:59:59.999
and when you think about
that happening on Twitter,

99:59:59.999 --> 99:59:59.999
what are the things that you
are doing specifically to ensure

99:59:59.999 --> 99:59:59.999
that you don't have misinformation
like this spreading in this way,

99:59:59.999 --> 99:59:59.999
influencing people in ways
that could affect democracy?

99:59:59.999 --> 99:59:59.999
JD: Just to back up a bit,

99:59:59.999 --> 99:59:59.999
we asked ourselves a question:

99:59:59.999 --> 99:59:59.999
can we actually measure
the health of a conversation,

99:59:59.999 --> 99:59:59.999
and what does that mean?

99:59:59.999 --> 99:59:59.999
And in the same way
that you have indicators

99:59:59.999 --> 99:59:59.999
and we have indicators as humans
in terms of are we healthy or not

99:59:59.999 --> 99:59:59.999
such as temperature,
the flushness of your face,

99:59:59.999 --> 99:59:59.999
we believe that we could find
the indicators of conversational health,

99:59:59.999 --> 99:59:59.999
and we worked with a lab
called Cortico at MIT

99:59:59.999 --> 99:59:59.999
to propose four starter indicators

99:59:59.999 --> 99:59:59.999
that we believe that we could ultimately
measure on the system.

99:59:59.999 --> 99:59:59.999
And the first one is
what we're calling shared attention.

99:59:59.999 --> 99:59:59.999
It's a measure of how much
of the conversation is attentive

99:59:59.999 --> 99:59:59.999
on the same topic versus disparate.

99:59:59.999 --> 99:59:59.999
The second one is called shared reality,

99:59:59.999 --> 99:59:59.999
and this is what percentage
of the conversation

99:59:59.999 --> 99:59:59.999
shares the same facts,

99:59:59.999 --> 99:59:59.999
not whether those facts
are truthful or not

99:59:59.999 --> 99:59:59.999
but are we sharing
the same facts as we converse.

99:59:59.999 --> 99:59:59.999
The third is receptivity.

99:59:59.999 --> 99:59:59.999
How much of the conversation
is receptive or civil

99:59:59.999 --> 99:59:59.999
or the inverse, toxic?

99:59:59.999 --> 99:59:59.999
And then the fourth
is variety of perspective.

99:59:59.999 --> 99:59:59.999
So are we seeing filter bubbles
or echo chambers

99:59:59.999 --> 99:59:59.999
or are we actually getting
a variety of opinions

99:59:59.999 --> 99:59:59.999
within the conversation?

99:59:59.999 --> 99:59:59.999
And implicit in all four of these
is the understand that as they increase,

99:59:59.999 --> 99:59:59.999
the conversation gets
healthier and healthier.

99:59:59.999 --> 99:59:59.999
So our first step is to see
if we can measure these online,

99:59:59.999 --> 99:59:59.999
which we believe we can.

99:59:59.999 --> 99:59:59.999
We have the most momentum
around receptivity.

99:59:59.999 --> 99:59:59.999
We have a toxicity score,
a toxicity model,

99:59:59.999 --> 99:59:59.999
on our system that can actually measure

99:59:59.999 --> 99:59:59.999
whether you are likely to walk away
from a conversation

99:59:59.999 --> 99:59:59.999
that you're having on Twitter
because you feel it's toxic

99:59:59.999 --> 99:59:59.999
with some pretty high degree.

99:59:59.999 --> 99:59:59.999
We're working to measure the rest,

99:59:59.999 --> 99:59:59.999
and the next step is,

99:59:59.999 --> 99:59:59.999
as we build up solutions,

99:59:59.999 --> 99:59:59.999
to watch how these measurements

99:59:59.999 --> 99:59:59.999
trend over time

99:59:59.999 --> 99:59:59.999
and continue to experiment.

99:59:59.999 --> 99:59:59.999
And our goal is to make sure
that these are balanced,

99:59:59.999 --> 99:59:59.999
because if you increase one,
you might decrease another.

99:59:59.999 --> 99:59:59.999
If you increase variety of perspective,

99:59:59.999 --> 99:59:59.999
you might actually decrease
shared reality.

99:59:59.999 --> 99:59:59.999
CA: Just picking up on some of
the questions flooding in here.

99:59:59.999 --> 99:59:59.999
A lot of people are puzzled why,

99:59:59.999 --> 99:59:59.999
how hard is it to get rid
of Nazis from Twitter?

99:59:59.999 --> 99:59:59.999
JD: So we have policies
around violent extremist groups,

99:59:59.999 --> 99:59:59.999
and the majority of our work
and our terms of service

99:59:59.999 --> 99:59:59.999
works on conduct, not content,

99:59:59.999 --> 99:59:59.999
so we're actually looking for conduct,

99:59:59.999 --> 99:59:59.999
so conduct being

99:59:59.999 --> 99:59:59.999
using the service to repeatedly
or episodically harass someone,

99:59:59.999 --> 99:59:59.999
using hateful imagery

99:59:59.999 --> 99:59:59.999
that might be associated with the KKK

99:59:59.999 --> 99:59:59.999
or the American Nazi Party.

99:59:59.999 --> 99:59:59.999
Those are all things that we did,
that we act on immediately.

99:59:59.999 --> 99:59:59.999
We're in a situation right now
where that term is used fairly loosely,

99:59:59.999 --> 99:59:59.999
and we just cannot take
any one mention of that word

99:59:59.999 --> 99:59:59.999
accusing someone else

99:59:59.999 --> 99:59:59.999
as a factual indication that they
should be removed from the platform.

99:59:59.999 --> 99:59:59.999
So a lot of our models are based around,

99:59:59.999 --> 99:59:59.999
number one, is this account associated
with a violent extremist group,

99:59:59.999 --> 99:59:59.999
and if so we can take action,

99:59:59.999 --> 99:59:59.999
and we have done so on the KKK
and the American Nazi Party and others,

99:59:59.999 --> 99:59:59.999
and number two, are they using imagery
or conduct that would associated them

99:59:59.999 --> 99:59:59.999
as such as well.

99:59:59.999 --> 99:59:59.999
CA: How many people do you have
working on content moderation

99:59:59.999 --> 99:59:59.999
to look at this?

99:59:59.999 --> 99:59:59.999
JD: It varies.

99:59:59.999 --> 99:59:59.999
We want to be flexible on this,

99:59:59.999 --> 99:59:59.999
because we want to make sure
that we're, number one,

99:59:59.999 --> 99:59:59.999
building algorithms instead of just
hiring massive amounts of people,

99:59:59.999 --> 99:59:59.999
because we need to make sure
that this is scalable,

99:59:59.999 --> 99:59:59.999
and there are no amount of people
that can actually scale this.

99:59:59.999 --> 99:59:59.999
So this is why we've done so much work

99:59:59.999 --> 99:59:59.999
around proactive detection of abuse

99:59:59.999 --> 99:59:59.999
that humans can then review.

99:59:59.999 --> 99:59:59.999
So we want to have a situation

99:59:59.999 --> 99:59:59.999
where algorithms are constantly
scouring every single tweet

99:59:59.999 --> 99:59:59.999
and bringing the most
interesting ones to the top

99:59:59.999 --> 99:59:59.999
so that humans can bring their judgment
to whether we should take action or not

99:59:59.999 --> 99:59:59.999
based on our terms of service.

99:59:59.999 --> 99:59:59.999
WPR: But there's not an amount
of people that are scalable,

99:59:59.999 --> 99:59:59.999
but how many people do you
currently have monitoring these accounts

99:59:59.999 --> 99:59:59.999
and how do you figure out what's enough?

99:59:59.999 --> 99:59:59.999
JD: They're completely flexible.

99:59:59.999 --> 99:59:59.999
Sometimes we associate folks with spam.

99:59:59.999 --> 99:59:59.999
Sometimes we associate folks
with abuse and harassment.

99:59:59.999 --> 99:59:59.999
We're going to make sure that
we have flexibility in our people

99:59:59.999 --> 99:59:59.999
so that we can direct them
at what is most needed.

99:59:59.999 --> 99:59:59.999
Sometimes, the elections,

99:59:59.999 --> 99:59:59.999
so we've had a string of elections

99:59:59.999 --> 99:59:59.999
in Mexico,

99:59:59.999 --> 99:59:59.999
one coming up in India,

99:59:59.999 --> 99:59:59.999
obviously the election last year,
the midterm election,

99:59:59.999 --> 99:59:59.999
so we just want to be flexible
with our resources.

99:59:59.999 --> 99:59:59.999
So when people, just as an example,

99:59:59.999 --> 99:59:59.999
if you go to our current terms of service

99:59:59.999 --> 99:59:59.999
and you bring the page up

99:59:59.999 --> 99:59:59.999
and you're wondering about

99:59:59.999 --> 99:59:59.999
abuse and harassment
that you just received

99:59:59.999 --> 99:59:59.999
and whether it was against
our terms of service to report it,

99:59:59.999 --> 99:59:59.999
the first thing you see
when you open that page

99:59:59.999 --> 99:59:59.999
is around intellectual
property protection.

99:59:59.999 --> 99:59:59.999
You scroll down and you get to
abuse, harassment,

99:59:59.999 --> 99:59:59.999
and everything else
that you might be experiencing.

99:59:59.999 --> 99:59:59.999
So I don't know how that happened
over the company's history,

99:59:59.999 --> 99:59:59.999
but we put that above
the thing that people want

99:59:59.999 --> 99:59:59.999
the most information on
and to actually act on.

99:59:59.999 --> 99:59:59.999
And just our ordering, like, shows
the world what we believed was important.

99:59:59.999 --> 99:59:59.999
So we're changing all that.

99:59:59.999 --> 99:59:59.999
We're ordering it the right way,
but we're also simplifying the rules

99:59:59.999 --> 99:59:59.999
so that they are human readable

99:59:59.999 --> 99:59:59.999
so that people can actually
understand themselves

99:59:59.999 --> 99:59:59.999
when something is against our terms
and when something is not,

99:59:59.999 --> 99:59:59.999
and then we're making, again,

99:59:59.999 --> 99:59:59.999
like, our big focus is on removing
the burden of work from the victims.

99:59:59.999 --> 99:59:59.999
So that means push more towards technology

99:59:59.999 --> 99:59:59.999
rather than humans doing the work.

99:59:59.999 --> 99:59:59.999
That means the humans receiving the abuse

99:59:59.999 --> 99:59:59.999
and also the humans
having to review that work.

99:59:59.999 --> 99:59:59.999
So we want to make sure

99:59:59.999 --> 99:59:59.999
that we're not just encouraging more work

99:59:59.999 --> 99:59:59.999
around something
that's super, super-negative

99:59:59.999 --> 99:59:59.999
and we want to have a good balance
between the technology

99:59:59.999 --> 99:59:59.999
and where humans can actually be creative,

99:59:59.999 --> 99:59:59.999
which is the judgment of the rules

99:59:59.999 --> 99:59:59.999
and not just all the mechanical stuff
of finding them and reporting them.

99:59:59.999 --> 99:59:59.999
So that's how we think about it.

99:59:59.999 --> 99:59:59.999
CA: I'm curious to dig in
more about what you said.

99:59:59.999 --> 99:59:59.999
I mean, I loved that you said
you are looking for ways

99:59:59.999 --> 99:59:59.999
to re-tweak the fundamental
design of the system

99:59:59.999 --> 99:59:59.999
to discourage some of
the reactive behavior,

99:59:59.999 --> 99:59:59.999
and perhaps,

99:59:59.999 --> 99:59:59.999
to use Tristan Harris-type language,

99:59:59.999 --> 99:59:59.999
engage people's more reflective thinking.

99:59:59.999 --> 99:59:59.999
How far advanced is that?

99:59:59.999 --> 99:59:59.999
What would alternatives
to that "like" button be?

99:59:59.999 --> 99:59:59.999
JD: Well, first and foremost,

99:59:59.999 --> 99:59:59.999
my personal goal with the service

99:59:59.999 --> 99:59:59.999
is that I believe fundamentally
that public conversation is critical.

99:59:59.999 --> 99:59:59.999
There are existential problems
facing the world

99:59:59.999 --> 99:59:59.999
that are facing the entire world,
not any one particular nation-state,

99:59:59.999 --> 99:59:59.999
that global public conversation benefits,

99:59:59.999 --> 99:59:59.999
and that is one of the unique
dynamics of Twitter

99:59:59.999 --> 99:59:59.999
is it is completely open,

99:59:59.999 --> 99:59:59.999
it is completely public,

99:59:59.999 --> 99:59:59.999
it is completely fluid,

99:59:59.999 --> 99:59:59.999
and anyone can see any other conversation
and participate in it.

99:59:59.999 --> 99:59:59.999
So there are conversation
like climate change.

99:59:59.999 --> 99:59:59.999
There are conversations
like the displacement in the work

99:59:59.999 --> 99:59:59.999
through artificial intelligence.

99:59:59.999 --> 99:59:59.999
There are conversations
like economic disparity.

99:59:59.999 --> 99:59:59.999
No matter what any one nation-state does,

99:59:59.999 --> 99:59:59.999
they will not be able to solve
the problem alone.

99:59:59.999 --> 99:59:59.999
It takes coordination around the world,

99:59:59.999 --> 99:59:59.999
and that's where I think
Twitter can play a part.

99:59:59.999 --> 99:59:59.999
The second thing is that

99:59:59.999 --> 99:59:59.999
Twitter right now, when you go to it,

99:59:59.999 --> 99:59:59.999
you don't necessarily walk away
feeling like you learned something.

99:59:59.999 --> 99:59:59.999
Some people do.

99:59:59.999 --> 99:59:59.999
Some people have
a very, very rich network,

99:59:59.999 --> 99:59:59.999
a very rich community
that they learn from every single day.

99:59:59.999 --> 99:59:59.999
But it takes a lot of work
and a lot of time to build up to that.

99:59:59.999 --> 99:59:59.999
So we want to get people
to those topics and those interests

99:59:59.999 --> 99:59:59.999
much, much faster

99:59:59.999 --> 99:59:59.999
and make sure that
they're finding something that,

99:59:59.999 --> 99:59:59.999
no matter how much time
they spend on Twitter,

99:59:59.999 --> 99:59:59.999
and I don't want to maximize
the time on Twitter,

99:59:59.999 --> 99:59:59.999
I want to maximize

99:59:59.999 --> 99:59:59.999
what they actually take away from
and what they learn from it --

99:59:59.999 --> 99:59:59.999
CA: Well, do you, though,

99:59:59.999 --> 99:59:59.999
because that's the core question
that a lot of people want to know,

99:59:59.999 --> 99:59:59.999
is surely, Jack, you're constrained
to a huge extent by the fact

99:59:59.999 --> 99:59:59.999
that you're a public company,
you've got investors pressing on you,

99:59:59.999 --> 99:59:59.999
the number one way you make your money

99:59:59.999 --> 99:59:59.999
is from advertising.

99:59:59.999 --> 99:59:59.999
That depends on user engagement.

99:59:59.999 --> 99:59:59.999
Are you willing to sacrifice
user time if need be

99:59:59.999 --> 99:59:59.999
to go for a more reflective conversation?

99:59:59.999 --> 99:59:59.999
JD: Yeah. More relevance
means less time on the service,

99:59:59.999 --> 99:59:59.999
and that's perfectly fine,

99:59:59.999 --> 99:59:59.999
because we want to make sure
that, like, you're coming to Twitter

99:59:59.999 --> 99:59:59.999
and you see something immediately
that you learn from and that you push.

99:59:59.999 --> 99:59:59.999
We can still serve an ad against that.

99:59:59.999 --> 99:59:59.999
That doesn't mean you need to spend
any more time to see more.

99:59:59.999 --> 99:59:59.999
The second thing that we're looking at --

99:59:59.999 --> 99:59:59.999
CA: But just, on that goal,

99:59:59.999 --> 99:59:59.999
daily active usage,

99:59:59.999 --> 99:59:59.999
if you're measuring that,
that doesn't necessarily mean things

99:59:59.999 --> 99:59:59.999
that people value every day.

99:59:59.999 --> 99:59:59.999
It may well mean things that people

99:59:59.999 --> 99:59:59.999
are drawn to like a moth
to the flame every day.

99:59:59.999 --> 99:59:59.999
We are addicted because we see
something that pisses us off,

99:59:59.999 --> 99:59:59.999
so we go in and add fuel to the fire,

99:59:59.999 --> 99:59:59.999
and the daily active usage goes up,

99:59:59.999 --> 99:59:59.999
and there's more ad revenue there,

99:59:59.999 --> 99:59:59.999
but we all get angrier with each other.

99:59:59.999 --> 99:59:59.999
How do you define -- ?

99:59:59.999 --> 99:59:59.999
Daily active usage seems like a really
dangerous term to be optimizing.

99:59:59.999 --> 99:59:59.999
(Applause)

99:59:59.999 --> 99:59:59.999
JD: Taken alone, it is,

99:59:59.999 --> 99:59:59.999
but let me finish the other metric,

99:59:59.999 --> 99:59:59.999
which is, we're watching for conversations

99:59:59.999 --> 99:59:59.999
and conversation chains.

99:59:59.999 --> 99:59:59.999
So we want to incentivize
health contribution back to the network,

99:59:59.999 --> 99:59:59.999
and what we believe that is
is actually participating in conversation

99:59:59.999 --> 99:59:59.999
that is healthy,

99:59:59.999 --> 99:59:59.999
as defined by those four indicators
I articulated earlier.

99:59:59.999 --> 99:59:59.999
So you can't just optimize
around one metric.

99:59:59.999 --> 99:59:59.999
You have to balance and look constantly

99:59:59.999 --> 99:59:59.999
at what is actually going to create
a healthy contribution to the network

99:59:59.999 --> 99:59:59.999
and a healthy experience for people.

99:59:59.999 --> 99:59:59.999
Ultimately, we want to get to a metric

99:59:59.999 --> 99:59:59.999
where people can tell us that,
hey, I learned something from Twitter

99:59:59.999 --> 99:59:59.999
and I'm walking away
with something valuable.

99:59:59.999 --> 99:59:59.999
That is our goal ultimately over time,

99:59:59.999 --> 99:59:59.999
but that's going to take some time.

99:59:59.999 --> 99:59:59.999
CA: