clock menu more-arrow no yes

Twitter CEO Jack Dorsey talked to NYU’s Jay Rosen for an hour, on the record. Read and listen to the full interview here.

Dorsey says “we have definitely been gamed” by bad-faith actors and doesn’t expect that Twitter will ever build a “perfect antidote.”

Twitter CEO Jack Dorsey
Twitter CEO Jack Dorsey
Fairfax Media / Getty

Jack Dorsey runs one of the most valuable, maddening, delightful and frustrating technology companies in the world.

Now Dorsey is trying to make Twitter better, and he wants your help.

And if you’re very lucky, he’ll spend an hour talking to you, asking you what you like about the service, what you’d like to change about it, and telling you what he thinks.

That’s what happened this week when Dorsey sat down for an hour with Jay Rosen, a Twitter power user who is also a professor of journalism at New York University. Rosen recorded the conversation on his iPhone (5!), and you can listen to it now, on a special episode of Recode Media with Peter Kafka. We’ve also included a transcript of their conversation below.

Dorsey has done a lot of talking over the past few weeks, and if you are a Dorsey completist you may recognize components of this discussion from other interviews. But this is more of a conversation than a conventional Q&A, and if you have deep interest in the way Dorsey thinks about his powerful/fraught platform/messaging service (don’t call it a microblog!) you’ll get a lot out of this.

For instance, he’s clearly interested in the idea of letting users tell the world that they’re actively on the service, in real time, a notion he has noticed bubbling up from Black Twitter. And while he’s very interested in presenting different slices of Twitter to different users, he’s also adamantly opposed to closing off subsets of the service to specific groups, like Facebook is encouraging its users to do.

Pro tip: If you’d like to spend an hour talking to Jack Dorsey about Twitter, one way to go about it is to tweet about Dorsey’s conversations about Twitter, after spending years building a following as a thoughtful observer of journalism, media and technology.

That’s what Rosen did last month, which led to this response from Dorsey:

Which led to the conversation below.

You can find more of Jay Rosen’s work at his blog PressThink and on Twitter; you can also listen to him on two other episodes of Recode Media. Special thinks to Ben Brandstein at NYU for his audio production help with this recording.

You can listen to Recode Media wherever you get your podcasts — including Apple Podcasts, Spotify, Google Podcasts, Pocket Casts and Overcast.

Below, we’ve shared a full transcript of Jay’s conversation with Jack.


Jack Dorsey: You’re sticking with the [iPhone] 5?

Jay Rosen: Yeah. I don’t like to change much, and that’s not so great for technology.

Yeah.

I’m one of those users who always screams when you change something, and then half the time I hate it, and half the time I get to like it.

What did you think of the 280?

Exactly that. At first, I said, “I’m not gonna do this. I’m gonna voluntarily stick to 140,” because I thought that was a great constraint and I thought it was unwise to change it. But now, it became a little ridiculous to hold out because everybody else was using 280 when they needed to. So now I’m comfortable with it. It didn’t actually create any huge problems.

Yeah. Well, we found that when people are just tweeting more as a broadcast, it on average tends to stay below 140.

Yeah, that’s very interesting.

Which has been great, but where it’s really been helpful is when people have a conversation. So, replies do go above 140 and allows a little bit more nuance and a little bit more space to have a discussion and discourse.

Yeah. It’s actually been very successful for me in that sense because I often have views that need qualification, and knowing Twitter the way I do and the kinds of reactions that you sometimes get, if what you’re saying sounds like something people have heard before, they react to that. In order to say, “I’m not saying this. I am saying that,” 280 actually makes a big difference.

Yeah.

NYU professor Jay Rosen
NYU professor Jay Rosen
Bosch Foundation

What has been ... I have several questions I wanted to ask you. What has been your experience in your listening tour, especially with conservatives? What have you learned?

Well, first and foremost, at least I personally have not tended to have conversations with many people in a more conservative end of the spectrum or right end of the spectrum, so goal number one was to say that we’re here, be present, and see the folks who I personally haven’t talked to, and as an organization, we tend not to naturally lean towards, and I don’t know if there are any fundamentally different learnings that are different from the conversations that we have with folks who are more on the left end of the spectrum, more of the liberal end of the spectrum or libertarian end of the spectrum, wherever that lies.

So, there’s a lot of questions as to how — well, why we make decisions the way we make them, how the algorithms work, a lot of questions around timeline ranking, and that changed three years ago. It was the first time we really applied machine learning to where people spent the majority of their time within the service, and some confusion about that. I think there was a question as to why there haven’t been a lot of conversations with people on the more conservative end of the spectrum in the past. There was a desire to have more. There’s a desire to be able to have an open feedback channel, and I don’t know, always good conversations. We had some folks from the media in every one of these. We had some folks in politics, some folks in academia, some technology folks, and pretty good.

Mm-hmm. You say they’re similar to conversations, people on the left side of the spectrum, but people on the left side of the spectrum aren’t accusing Twitter of shadow banning their voices, are they?

The people we talked to didn’t really accuse us of shadow banning their voices.

They didn’t? They didn’t accuse you of bias against conservatives?

They asked questions as to whether a bias within the company would translate into the service and into actions, but it was all questions rooted in, “I follow this person. Why am I not seeing their tweets in my timeline?” The majority of the questions I got about shadow banning and bias were either on Twitter or within the Congressional hearing.

So, public performances, where there was some external audience to perform for. Yeah. Why did you go public with this generalization that people in Twitter are generally liberal or lean left, and that’s our culture, which I have no trouble believing is true, but why did you start talking about that?

I think it’s more and more important to at least clarify what our own bias leans towards, and just express it. I’d rather know what someone biases to rather than try to interpret through their actions. So, if we can say that, and also have the freedom to evolve and change, then at least people know it, and I think it allows us to remove that a little bit more from the work, but it has to be proven out in our actions as well, so ... I mean, we have a lot of conservative-leaning folks in the company as well, and to be honest, they don’t feel safe to express their opinions at the company.

They do feel silenced by just the general swirl of what they perceive to be the broader percentage of leanings within the company, and I don’t think that’s fair or right. We should make sure that everyone feels safe to express themselves within the company, no matter where they come from and what their background is. I mean, my dad was a Republican. When I was growing up, was on the radio all the time with Rush Limbaugh and Sean Hannity, so my mom was on the opposite end of the spectrum and ...

Right. I read about that in another interview you did. Yeah.

Yeah, and I appreciate that, and I always felt safe to challenge both of them, especially my dad, and so it was definitely a privilege, but if we’re creating a culture that doesn’t enable people or empower people to speak up or not, we’re gonna be able to do that for our service.

I’m of two minds on that. On the one hand, it’s absolutely true that if you don’t feel safe in speaking out, then in effect, you don’t have free speech. I think that’s true, and that’s a tremendously important thing for a company like Twitter, but it’s also true that public life and participating in public debate involves risk. The most basic risk is that someone will criticize you, and do you ever feel like saying to your conservative employees, “Look. Speak up. You might get criticized, but you have to have the courage to do that. We’re not gonna penalize you, but you are, to some degree, when you speak up in public or in a public culture of a company, you are, yes, vulnerable to criticism, vulnerable to reaction. That’s part of public life. That’s part of being a mature citizen?” Do you ever say that back to them?

I mean, yeah. It’s easier said than done. I know this for myself. I was a kid that was very shy. I grew up with a speech impediment. It taught me not to speak at all, and I eventually got over that, and speaking up in a collection of 3,000 people where you make an assumption that they potentially think differently than you or believe differently is hard. It does require sacrificing a lot of your ego and your intellect in being vulnerable for a minute.

What are the best bridges to enable people to do that? I don’t know if entering into a political debate is always the right first step, but maybe. But yeah, I definitely encourage speaking up and having the courage to do so, but one has to feel it maybe in a different context before they get more of that, and I think it just takes time. But we have people who will be courageous and speak up, but I don’t know. It’s hard to do as any individual.

When I interviewed Adam Sharp, who at the time was the head of news for Twitter, he said something very interesting to me, which was that we may, from the outside, underestimate how passionate people on Twitter are about news, how much they love news and how obsessed they are with the news. Is that true? Is that part of the culture at the company? Where does that come from?

Oh, what do you mean? In terms of us being obsessed with news?

Yeah, the people who work there are lovers of news. They are obsessed with it. They follow it, obviously. They contribute to it through the company, the platform, and they are especially tuned to the problems in the news system. That’s what he was saying.

Yeah. I think so. I mean, when we started the company, we built something that we wanted to use, and more and more, people taught us what they wanted to use it for, and then what they wanted to use it for was to see what’s happening, which was ...

News.

... news and newsworthy, and whether it be a conversation between two people as being newsworthy to someone, or actual breaking news that is more global of nature, it is certainly now in our DNA, but I do think it’s a byproduct of what our real fundamental is, which is public conversation. Entertainment is a byproduct of that. If we lose sight of the fundamental that we’re serving, which is enabling people to converse in public and discuss in public ...

Why do you see that as more fundamental than news? Because I don’t see it that way.

Because we never know what is going to be newsworthy or not, and oftentimes we’ve seen simple conversations turn into something really meaningful. The most prominent that’s top of mind immediately in this moment is there was an IT professional in Pakistan who was tweeting at 3:40 in the morning about helicopters over his apartment, and five hours later, we learned that that was the raid on Osama bin Laden.

Bin Laden. I remember that. Yep.

If we purely focus on news as just the fundamental atomic unit, I think we become dependent upon news sources rather than seeing potentially everything as newsworthy to someone, and I also think if we were to focus entirely just on news as the atomic unit, we wouldn’t see as much discussion and debate and dialogue around what’s happening as well. So, it’s not that I don’t think Twitter is about news and people don’t use it to see what’s happening and get their news, but there’s gotta be something that feeds that, and to me it’s that public conversation aspect.

See, I see it almost as the reverse. When I say news, though, I don’t mean the news industry or the major providers of news.

Anything new or noteworthy.

Yeah. I mean news in the more fundamental sense of “what’s happening?” It could be what’s happening in my life. It could be what’s happening in my neighborhood. It could be what’s happening in the world, what happened today in Congressional hearings. And the way I look at Twitter is that it’s a part of the news system. It’s a fundamental part of the news system. It’s become part of the infrastructure of the news system, which is why anybody who pays attention to the news closely knows that news usually breaks on Twitter.

Very often, the first notice you have of it is on Twitter, and I first became aware of that with the plane crash in the Hudson, where I kind of knew it as an abstract thing, but seeing it happen, especially because I live in New York and I wasn’t that far from the event itself, and so Twitter is fundamentally ... It’s part of the news system. You go there to find out what’s happening, and what people naturally do, what they need to do to have a healthy democracy is, of course, discuss the news, and so I see them as complementary ...

Oh, absolutely.

... and equally fundamental.

And they cycle. Sometimes those discussions become what’s happening and become news as well. And to your point around ... Twitter was described in the past, I’m not sure who said it, as where the news gets its news.

Yeah, that’s true, too. Yeah.

I think a lot of that has to do with the fact that people do feel free to discuss what’s happening. What I appreciate about Twitter is, unlike in Instagram, it doesn’t feel like a post. It feels like the start of a — it feels like a message.

So, you want to increase that sense of flow rather than posting, right?

Fluidity. Yeah. We don’t want to post and then stop. One of the descriptions and labels that we had in the past, which I always despised, was microblogging, because you ...

Yeah. I remember that.

It encourages behavior of post, comment, comment, comment, and it just feels such to be a dead end versus message, message, message, message, which is just as fluid reply, and it could go anywhere, and I think that speaks to the fact that we don’t have a social graph. We have a graph around interest. We have people following you not because of the fact that you’re in their address book, but because they’re interested in what you have to say, and the spectrum of what you have to say changes over time, and that fluidity is really important and isn’t really captured in, “I’m composing a post.” It’s like, “this is what I think right now, and this is what I think right now, and let me clarify, and let me clarify, and this is what I think right now.”

Yeah. I use it that way, and I try to benefit from that feature of Twitter and the thread mechanism, which ...

Has that been helpful?

Yeah. I like the fact that it started as a user convention, and then ...

Everything on Twitter.

Yeah, everything starts that way. Right.

The @ symbol, the hashtag, the retweet.

Totally. I enjoy that part of Twitter, and I like the thread because it’s a form of writing that is very demanding. So, for me, it is what you say. It’s intended to be part of conversation. It lends itself to conversation. I can always add to my thread if I want to, which is really interesting. I could actually come back three weeks later and add something to it if I want to, but because of the risks that anybody with a public platform on Twitter has whenever they post — meaning if I say something stupid, it could create a viral attack on me, if it’s really out of bounds. I’m always aware of that. And so even though it’s supposed to be a flow and supposed to be a natural conversation, that’s an artifice for me because I have to think super carefully about everything I post.

How do we fix that? How do we allow more space for verification?

An edit button would be one. I know that’s hard. I know there’s all kinds of issues, but that’s ...

It’s not that it’s hard. It’s that if you asked a hundred different people what they intend by “edit,” you’ll get a hundred different answers.

Well, I just mean fixing mistakes within a few minutes.

Within a five-minute window?

Yeah, within a few minutes of something that you post.

What about longer? But that doesn’t address the clarification issue. You may tweet something stupid and not realize it after an uproar or outrage for a day.

Yeah. Well, there, I don’t know how hard this is from an engineering point of view, but it would be interesting and I would use it if I could clarify something that I posted.

Do you ever do that with your own tweets through quote-tweet?

I do. Yeah. Mm-hmm. Yeah. I correct things that way.

Does that work?

It does, except that very often the follow-up doesn’t spread the way that the original did. So I can use it that way, but somehow, to indicate that something that is spreading has a further explanation to it or qualification, I would probably use that, even though it’s only a handful of times when I need it.

You need it.

But when I need it, I need it a lot.

Yeah.

But as I said, for me, I don’t really mind having to think through what I say, because I’m a college professor. I’m supposed to be careful and precise, and actually, I consider it a good discipline, the fact that one wrong tweet could blow up your life, which is a fact. It’s happened to many people. It does make it risky. It means never tweet drunk, all these basic things, but it’s also a very good discipline for me as a writer. I like it.

Yeah.

Let me ask you about something else. When I read through your testimony, which was a series of tweets, and some of the interviews ...

Which one?

The one that you did recently.

No, I did two hearings, and there are two different series of tweets. Those were just the openings, but they were different. One was focused on election interference and then the other was focused on bias.

I think it was the second one.

Congress.

I felt like there’s a tension in what you’re trying to do. The tension is, on the one hand you’re saying, “We are impartial. We’re an impartial platform. When we write our rules, when we enforce our rules, we’re not trying to discriminate against this group or that group. Even though we, as individuals at Twitter and as a company with a culture, do lean this way.” Right there, those are two different motions. One is impartial, we are neutral, we’re objective.

Not neutral.

Okay. You didn’t use the word neutral. You don’t want the word neutral.

No.

All right. That’s good. I wanted to figure that out. All right. I’ll ask you about that in a second. So, “We’re impartial, but we’re not saying that we don’t have any points of view, or that we don’t have our own culture. We do.”

Correct.

Okay. That’s one system. Another system that seemed to be coming through in that tweet thread is, “We are a company that contributes to the public square and to conversation in the public sphere.”

That’s how we’re used.

That’s how we’re used. Okay.

We believe that many people see us as a public square and use us as a public square. I don’t have expectations accordingly. They expect, because we’re used as a public square, they expect to have the same sort of, well, they have the same sort of expectations they would have of a public square, like Bryant Park.

Okay. So my question is, why aren’t the values of Twitter what it takes to have a healthy public sphere?

I guess it depends on what you mean by value. We value health in dialogue so that is our singular objective right now, is to increase the health of how people participate in our public square. I think it does benefit to get into neutrality versus impression.

Yeah, let’s do that.

A lot of people have asked us, “Are you a public utility?” and, “Are you a neutral communications platform?”

Like the electric company or something like that.

Or AT&T. Neutral communications platform, presumably, you pick up the phone, you call your friend, you call your father, you call your mother.

Right. It doesn’t care who you’re calling or what you’re using.

Presumably, they do not care about the content whatsoever because the other person picked up the phone — or multiple, you have a group conference call, they all decided to participate in that thing. And they should be completely neutral to whatever happens on that pipe and through that pipe. And a lot of people have taken that neutrality and then said, “Well, we should also be that because we are like the internet.”

Yeah, pipes.

Pipes. But we, upon further reflection, we are being used more like what you would find in Washington Square Park. You walk into Washington Square Park and there’s a bunch of people who, when I walk in, there’s a bunch of people there who are not expecting me to walk in and aren’t expecting me to do the things that I intend to do and might see it out of the corner of their eye and might come over and listen or interact or whatnot. In that public square, there’s all these things that happen and some are amazing, and some are stupid, and some are silly, and some are really terrible. There’s a guy in the corner with a megaphone broadcasting his thoughts and then he recognizes you and he says, “Jay, get the hell over here. You’re a terrible person and I hate you,” and all these other things. And it’s completely directed at you.

And at that point, people recognize it and they tell him to stop, or the park stewards or police come over and say, “Here’s a warning and if you keep attacking this one person who doesn’t want it and is not even paying attention to you, then you’re out.” So that action right there was not neutrality, it was being impartial to the conduct and with an eye towards more of the collective, with an eye towards like, “We need to make Washington Square Park something that people actually want to be at and recognize that there’s going to be people who choose unhealthy behaviors and we’re going to at least demonstrate what is not healthy and what could be healthier.”

I do believe health is a value that we’ve chosen to make a singular objective, and we value health in public conversation, but in order to do it correctly, we need to do it with a principle of impartiality, which means that we’re not going to do on the basis of bias or prejudice or favoring one account over another for improper reasons. Where we have failed in that is to be transparent around how we write our rules and how we enforce them.

And where we’ve also failed in the past is being open with our mistakes and then correcting them. Because we’re not always going to see outcomes of impartiality even though that is our intent. We will see outcomes that are not impartial and that may favor one account over another or inject bias, whether it be through the training set that was used to train our algorithm or a human operator that had to make a judgment call and made it incorrectly.

So I think it’s really important we clarify, we do value health in this thing that people see as a public square. We are going to make it our objective to increase the collective health. We realize not everyone is going to choose health in the short term, but we want to demonstrate by choosing healthy conduct. We can further amplify your reach, but if you don’t, it’s only your earned audience and if you’re harassing people, then we’re going to ask you to leave.

What do you think this principle of health that you just articulated replaces in your old thinking, if this is the new thinking? What did you have in that place before?

I think we were just addressing this in terms of what we’re seeing on the surface.

Just going from problem to problem without really having a coherent point of view?

A coherent, cohesive framework to link them altogether?

Yeah.

We saw abuse, we went after abuse. We saw misinformation, we went a little bit after that.

Right. It’s fighting fires.

But they’re all tied together. This is what the Cortico folks and Deb Roy asked us about a year-and-a-half ago now, which is like, “What if you can measure health of conversation?” I said, “How would you do that?” He said, “Well, first you need to figure out what the indicators of health are.” I said, “What do you mean?” He said, “Well, your body has an indicator of health and system imbalance which is your temperature, 98.6 means more or less your system’s in balance. If it’s above or below then something’s out of whack and something’s out of balance.”

We recognize that’s an indicator of health, but we also need to measure it so we needed to invent thermometers to measure it. Then we get a metric out and then we can see how that metric trends over time. And as we deploy solutions like hot water with lemon or when you take wine, what happens with that trend? Then based on those experiences, we can say that, “If you want to be healthy, then you drink more hot water with lemon and less wine during that time of imbalance.”

So he came up with four, which are shared attention — and implicit in all these indicators is that more of them is healthier. Shared attention, like what percentage of the conversation is focused on the same things rather than disparate things. Shared reality, which is what percentage of conversation is sharing the same facts, not that they’re true or not, but this percentage of the conversation or accounts or people believe the world is round and this percentage believe it’s flat. The implied assumption there is that the more of us that are sharing the same facts, the healthier the conversation will be.

Yeah, but the world is round.

Yes, and the predominant percentage believes that that is factual. There are other people who share different facts. And that’s just one crazy but illustrative example.

That’s a rabbit hole there.

It’s just an example, but there are ... religion. And I believe this fact to be fact and I believe this to be factual and those aren’t necessarily shared. And then the third one being receptivity. Participants, are they receptive or are they creating toxic action? And then fourth and finally is variety of perspective. Are we feeding a filter bubble or echo chamber or are we actually seeing divergent — it’s not diverse, but divergent abuse.

That’s the beginning of the direction that I would like to see you go in. It’s fundamentally compatible with what you’re saying. But the way I would say it is, Twitter contributes to, is part of, wants to be a key part of a healthy public sphere.

Which we’ve committed to.

Okay, you committed to that already. And, in order to have a healthy public sphere, certain things have to be true, certain things have to be had.

Or increased.

Yes. And we are continually in the process of learning what the requirements of a healthy public sphere are. We know what some of them are and we are designing our systems and enforcing our rules accordingly. And we are continually learning what it really takes to have a healthy public sphere in a global connected society and a platform that reaches everywhere in the world. And therefore, we don’t want to apologize for changing our policies or evolving them because nobody knows everything about this problem.

And furthermore, because we’re always learning, we have to learn in a very public way and therefore we are committed to, yes, transparency as an abstraction, but in practice, it means we have to share a lot more data. We have to share our thinking. We have to share the kind of criticism we’re getting and we may have to build systems to get better criticism. That’s the way I would state it.

And that is — it’s important because it means that you’re not saying anything goes, you’re not saying you’re indifferent to what happens on the platform. You’re saying, “We are standing up for these values and we’re continually learning what they require of us and if you violate the requirements of a healthy public sphere, we will take action against you.”

And one of the things that obviously you have to recognize — and I’m sure you’re recognizing it more and more — is that there are bad actors who don’t actually want there to be a healthy public sphere. They want it to be distorted. They want it to break down. They even might want to destroy the institutions of a public sphere because, in the wreckage of those institutions, a lot of energy is released, a lot of fury, controversy. And you can power a political movement with them.

I agree with all that. Did you read my health thread?

I think I did.

Because we said, we committed to all the things you said.

I know you committed to it, right.

But it is, like we do say a lot and we need to prove it through the work, but ultimately ... and we’ll have updates to that, obviously, as well. But we believe those health indicators shouldn’t be owned by us. They should be designed and defined by a third party. We should publish to them and we should be held accountable to them.

Third party meaning ...

Third party being like Cortico, MIT, social apps. For example, the researchers that we’re working with through our feed process to define what these indicators are. I don’t know, but we recognize we’re not the only ones serving a public square, a public conversation, that there are multiple folks including journalists and publications like the New York Times and the National Review or whoever. And maybe they could also utilize these indicators to determine health of the discussion around their work, or even measure the efficacy of their work as a contribution, as an impact to the public conversation.

What about having public editor or public editors, plural.

Internally?

No, like a Margaret Sullivan, somebody who’s empowered to listen to complaints that users are having, get answers from the company. They are not beholden to the company, but they are like ombudsmen. And that’s a way of showing that there’s somebody that you can talk to when you think something is amiss and they push for more transparency. They push for answers on behalf of the users.

I like the idea of it.

Yeah. Who is empowered to work on behalf of the users at Twitter? Everybody?

Everyone, in terms of direct, the most direct contribution, it’s engineers and product and design folks. We have a research team that is constantly having conversations with people all around the world that use our service and people who don’t use it, asking them why. We’re about to hire a social scientist as well to help us understand more of our impact, not within the service alone, but also off platform.

This is another dimension that we’re adding is, the ramifications of what people do on Twitter and how they translate off, and vice versa. So it really depends on what you mean, but we are most frontline, to actual people that we serve, is our research team, our customer support team and our sales team. They offer conversations on the databases, live.

I guess what I meant is, I see people sometimes explode in frustration, and I’m sure you’ve seen this too. They say, “Hey, @Jack, would you fix ...” and they’re trying to talk to the top of the company, which isn’t really a very practical way of getting their concerns registered and acted upon. So you need a system for that, that’s immediate flare-ups.

Then it seems to me you have another listening project, which would be medium-term, things like, “How’s the 280 going?” that’s a medium-term concern. And then you have these longer-term issues like the ones you’re talking about, of how do people follow conversation, how do we make the platform more topic-focused, things like that. But I don’t know who, if I have something I think is wrong with Twitter, I don’t know where to go and I’m a pretty experienced user.

Yeah, tweeting it, usually. The amount of tweets we send internally as, “Look at what people are saying about this,” or this feature, or, “This is really topical,” or, “This is a fair critique,” is pretty immense. Our leadership team alone, we’ve have a private DM that we coordinate almost everything that we do around, and we’re constantly sending tweets that we find that are critiques of us or ... like there was a really good critique sent yesterday around the rank timeline and how Sarah ... I forget how to pronounce her last name ...

Kendzior, from St. Louis?

Same as I’m from.

Yeah, you’re from there, right.

She sent a tweet about, she turns the switch off, the ranked timeline ...

I saw that.

... and she saw a bunch of things from people that she hadn’t heard from in weeks. So we passed that around and we need to look into why that was happening because that’s not the intention of the ranking. It should be focused on like, “If this person is extremely relevant within the moment and relevant to the viewer, it should be at the top of the timeline.” So these are all signals that we use.

So who’s responsible for explaining that?

Our engineering team ultimately designing the home timeline experience and the algorithms behind it.

But who’s in charge of explaining it to users?

It really depends. I would like more of our employees to be more conversational on our service on a regular basis. I want to be a really open company where we’re just having conversations nonstop about our work. We tried another set with this when we were ... we’d been playing with the idea of what presence means on Twitter and also playing with threading on Twitter as well, both with the desire to incentivize more conversation and, ultimately, healthier conversation.

So if, you know, people get a lot ... like Zeynep is someone who says like, “Twitter’s all about conversation,” and like, “The amount of work I have to do to follow replies and understand who’s replying to what and this tweet goes here and this tweet goes here is ridiculous. You’re taking a bunch of time away from me.” So it’s really under that initiative of like, “Let’s better organize how people are utilizing this so that we make Twitter feel more conversational.”

What does presence have to do with it? Why do you use the term “presence?”

Presence is like, it’s, you know, it’s basically, “I’m online, I’m here.” There’s been a really amazing, kind of organic thing that’s come out of Black Twitter, which is #onhere, to describe just experiences that are happening right now or contextualize the experiences happening on Twitter itself. We’ve been playing internally, and we shared some screenshots of what if — of course if you give permission — what if you see a green dot next to my name? Seeing that, like ...

That means that I’m on ...

I’m active on Twitter right now and you could reach out and talk with me right now or tweet at me right now. Or if you tweet, I might be more likely to see it, but just that sense of like —

I might use that. Yeah, I might use that sometimes.

Like, it’s who’s up.

Yeah. Who’s up and who’s open to being communicated with.

Yeah. It kind of goes back to our early days of like, a lot of Twitter came from like the AOL Instant Messenger status and just extracting that into an entire product. But the thing that we lost was the “available” signal.

Right.

It’s interesting, the other thing that we’re noticing is that ...

Which is, messaging systems have that.

Yeah.

Yeah.

The other thing we’re noticing is that, I don’t know if you’ve noticed this, but we have the @ name and then we have the display name.

Yes.

And people have been changing their display name because we increased the size.

Yeah, I saw that.

They do it for a lot of reasons. We see a lot of “resist” in it, we see Xs.

I would never do that. I noticed that. I would never in a zillion years do that.

A lot of people who think we’re shadow banning them put an X in it, and you see all sorts of things. But what’s really interesting is people have started to put their status in it, like “Jack in NYC” or “NYC Jack” and that’s another, it’s another opportunity. People are using this thing that wasn’t designed for it, but they’re telling us that they want to have some omnipresent status independent of the tweets that might be more ephemeral. So it’s just interesting to look at these organic behaviors that emerge and whether they should be live features. But a lot of people have been utilizing this #onhere of like, “I’m up.”

Right.

Like, “Who’s around?”

Right.

Black Twitter is extremely conversational and very much used like we are a text messaging app but the whole world can join in. I just think that’s so fascinating. That is why I say like news and entertainment are byproducts of conversation, and vice versa. Sometimes something is happening in the world and I see it outside, like a plane landing in the Hudson, and I have a conversation about it and sometimes I have a conversation, and that becomes the news.

Yeah, I think it works both ways.

Yeah, and they feed each other.

Right. And I think, conversation, when it’s good, also can cause people to look for news.

Yeah. Yeah. That’s a good point.

Which is really important.

That’s a good point.

I want to ask you about one other thing and then any questions you have for me, I’d be happy to answer. It would seem to me, if I were a CEO of Twitter, that the nightmare scenario for me, as a executive, would be something like what’s been happening in Myanmar and other places where false information gets communicated over a public platform and it leads to essentially genocide or attacks on people. You don’t necessarily know it because it’s really hard to keep track of everything that’s happening on a global platform. Does that keep you up at night? How do you handle the very real possibility that your machine could be involved in something like that?

It definitely keeps me up, and something I’ve been reflecting a lot more on, in terms of just in the context of the hearings and the election interference. Like what I think is happening is much bigger than through elections. There are, as you said, bad-faith actors who have an agenda to distract and divide and to confuse. The election just happened to be a hook.

Yeah, and they may not be evenly distributed across the political spectrum.

Absolutely, or ...

Bad-faith actors, they may not be.

Or they may not be on the spectrum at all.

Yeah.

They may not be on the spectrum at all other than having an agenda just to ...

To destroy public conversation.

To divide.

Yeah. To divide.

To divide.

And to, yeah, and to polarize.

Yeah.

Yeah.

And just to cause a number of issues that distract us from what we really need to focus on. The organizing principles of the past are not going to serve us in the same way that we need to organize to face the challenges that are present today. Like the existential crises before us of economic inequality and the growing wealth gap, especially a racial wealth gap, the environment, and the displacement of work from artificial intelligence, those are things that an organizing principle of a nation state will not be able to solve because they’re global and they face all of us.

So on the positive, I believe this is where the internet can really shine, is that we have a global organizing principle, and it is the internet, and we have an ability to build layers, like Twitter, that ours is focused on conversation, conversational layer of the internet and wanted to make it accessible to everyone. Some people will utilize that for extremely negative things and continue to look for opportunities to game us. We’re never going to build a perfect antidote to that. We have to stay 10 steps ahead constantly. That means we need to be a lot more aware of how people are using the system and how people are intending to game it, and act faster.

I think the one advantage that we have against our peers is the fact that it is just completely open and completely fluid and completely public. I think when you have a join button or a subscribe button into a small community or a close community, you develop a lot of these really isolated sort of bubbles, where things can really fester quite quickly, and it’s also hard to see what’s going on inside. But more important ... and that becomes a real issue when you have to be a small company that’s trying to see as much as possible. But if everything is out in the open, then everyone in the world ...

I see what you mean by an advantage, yeah.

... can see it and call it out and at least raise the issue so that we can identify it much faster.

Yeah.

So I do think we have an advantage of everything being on the surface and I would hate to lose that. So that is not something that I rest on because we have definitely been gamed, and we have definitely been utilized to manipulate people. I think that intention will only grow. I think people continue to find new ways of like we saw the Russian government during 2016. We just disclosed evidence of folks in Iran taking on similar patterns and I’m certain it’ll continue to happen within the borders of this country and countries around the world. But this is where the organizing principle of a nation state doesn’t work because these are going to be global issues and require global solutions. We need to prioritize those global solutions first.

One point to recognize about that, Jack, is that if it’s true that nation states won’t be able to solve these problems, but we have the internet, which is global and does have the scale … that means that the companies that own that global scale become very much like a world government. That’s a problem because they don’t have the same accountability that elected governments have. So that, that is a huge issue.

Yuval’s new book “21 Lessons,” I’m halfway through right now, but he’s exploring a lot of these, especially like do we need a need a global world government? His answer is no. What we need is, and I might be getting this wrong because I’m halfway through the chapter, but what we need is a lot more of this kind of open communication around how these things interact, but also that the nation states prioritize, and even local governments like New York City prioritize global concerns over the local concerns.

Take the environment as an example. What are we doing to reduce our own impact on the environment? So I do believe that a lot of these services, and ours, there’s a growing concern of the power that we have over this public square that we’re serving and there’s a natural distrust of it and a natural fear of it because people fear losing their voice. People fear losing their ability to participate. People fear losing their job.

Yeah.

Because they hear “Twitter” and then they hear, they go in the timeline and they hear “algorithm,” and algorithm is displacement of work, and then five years down the line, where am I? I don’t think we’ve done enough to address some of those fears. I think the only way we can do it is more openness and more transparency.

Well, on that point, one of the things that I would really love to see Silicon Valley, you and your peers turn to, is innovation on these kinds of questions. Innovation in public explanation, innovation in transparency ...

Or terms of service.

Innovation in terms of service, perfect example. Terms of service are completely opaque, nobody knows what they’re signing up for. That requires innovation. It’s not technology, right? It’s practice.

Policies.

Innovation in listening, innovation in public criticism and feedback. It’s amazing to me that in Silicon Valley all these people who see themselves as disruptors and innovators when it comes to, for example, explaining the company, they act like the same old PR merchants from the 1950s, like absolutely zero innovation, and nobody thinks that’s a contradiction at all. I do. So I would like to see a lot more innovation in public communication and explanation because without it, as you become these global forces, if you don’t have global accountability, you’re in for a lot of trouble.

I agree.

What questions do you have for me, if any, as a user since 2008?

If you could point to one positive impact of our work and one negative impact of our work, what would rise to the top in each?

Well for me, as a journalism professor, the most positive thing about Twitter is it allows me to practice journalism school extension, which means not just teaching the people who come to NYU and can afford the tuition and show up in my classes, but anyone who is interested and allowing me to develop a constituency for my ideas, that includes journalists for sure, but also lots of people who are concerned about the quality of journalism, and to do so without gatekeepers in a way that allows me to control my own message, has been amazing.

The bad thing is the same thing that people much more vulnerable than I am complain about, which is that anybody can walk over to me in Washington Square Park and scream how much they hate me at me and to reply to them would actually bring more of that upon me. So this sort of the openness to attack ...

And the velocity.

... and the velocity of it and the scale of it sometimes. Like when a big account decides that it’s going to unload on something that I said, usually in criticism of me being part of a “liberal media,” which I’m not but they think I am, that which again, I only experience a fraction of it compared to other people who are much more vulnerable than I am, and I’m a privileged character. I have a blue check next to my name, I have tenure at NYU, so I have all these protections, but that would be the thing that dissatisfies me the most. Along with some opacity in some of the changes that you introduced. That really bothers me. Like what is the algorithm doing to my feed? I don’t actually know. I don’t know where to go to find out.

Well, that’s another fascinating area of research as well. The algorithm doesn’t know either, sometimes.

Yes, exactly.

There’s this research field in AI around explainability, and the intention is to encourage more functionality that allows, especially deep-learning algorithms, to explain the criteria, the decision-making criteria they’re using, because right now they can’t.

That’s a big issue. Big problem.

So if you don’t know why a decision is being made by an algorithm ...

Then you literally don’t know what you’re doing.

... and we’re moving more and more of our decisions through it. Like I’m wearing an Apple Watch, it tells me to stand every now and then. I have offloaded a decision around my physical health to this thing, and if it can’t tell me why it’s making that particular point, I know that it happens 10 minutes to the top of the hour, but that may change over time and that may impact something in a fundamental way that I can’t predict.

It’s like the new frontier is in opacity.

Yeah.

This article originally appeared on Recode.net.