The Vergecast - Twitter's Kayvon Beykpour on product features Twitter is prioritizing

Episode Date: October 15, 2019

Twitter’s head of product Kayvon Beykpour sits down with Verge editor-in-chief Nilay Patel and Silicon Valley editor Casey Newton to talk about how Twitter juggles its policy with trust and safety, the state of verification, the future of video in your feed, and how tweets can become more ephemeral.  Please take our survey here: theverge.com/survey Learn more about your ad choices. Visit podcastchoices.com/adchoices

Support for the show comes from Retool. Too many companies run critical operations on duct taped spreadsheets, Slack workflows, and whatever else they could cobble together. Not because they want to, but because building internal tools means weeks of waiting on someone else's backlog. That's where Retool comes in. Build custom internal tools just by describing what you need. Prompts something like,

Build me a revenue dashboard on our Salesforce data. And Retool actually builds it on your company's data, in your cloud with enterprise security built in. Go to retool.com slash Verchcast. We all need to retool how we build software. What's up, y'all. I'm Skyler Diggins, seven-time WMBA All-Star, Olympic gold medalist, and mom. And I'm Cassidy Hubbard, host and reporter for nearly 20 years covering the biggest names and stories in sports and mom. And this is Am Mom, a community for athletes, game changers, and moms of all kinds.

dropping May 14th Tap in with us Hey everybody It's down on the Vergecast On this week's interview episode Casey Newton and I sat down

with Kavon Bakeport is the head of product at Twitter Obviously you know we talk about Twitter all the time Casey covers Twitter very closely

Kavon walked us through how he got to Twitter He was actually one of the founders of Periscope Twitter acquired Periscope He rose up to be the head of product

Twitter how he's thinking about the platform how he wants it to move a little faster they've just done some big changes on the back end

to make that happen How he manages the intersection of Twitter's policies, its content moderation systems, and how that stuff is expressed in features. That stuff is very complicated. This is really interesting to hear him think about it and explain how it works. And at one point, Casey and I just asked for features over and over again. You'll see. Kvon is great. He's very direct. He's very open, very honest. There's a fascinating

conversation. Check it out. Kavon Bakepur, head of product at Twitter. Kavon Bakepur, welcome to the Vergecast. Thanks for having you. You are head of product at Twitter, which is a service I think everybody uses and very interested in. But I want to just, the first question I have is you came to Twitter through Periscope. So just talk about that process and how you landed at Twitter and how you took the role of Twitter. Sure. How far back do you want me to go? Full Life Story or Periscope.

When you were a boy, you were like, I'm going to live stream some videos. When I was a Wii lad, yeah. Yeah. Yeah. So my co-founder and I have actually been best friends since like second grade, basically. And we started our first company together, which we built, we got acquired. We were working there for five years. and we both eventually decided to leave and take some time off and travel because we knew we wanted to start another company but didn't really know what we wanted to do yet.

And in the course of our kind of like year off and traveling, we both kind of got pretty addicted to this idea of how could we help people see what was happening anywhere in the world through other people's eyes. And the way we described at the time was we wanted to build the closest thing to the feeling of teleportation. And we started thinking about this idea for two main reasons. One, you know, this was in 2013. And then obviously, like, smartphones were pervasive, these really sophisticated cameras, as we were just talking about, were super pervasive with high-speed internet connections. But yet, there is no way to kind of, like, rent the eyes and ears of someone around the world to see what was happening through their eyes. And I particularly was excited about this in the context of one of the trips I was going on at the time was to Turkey. It was just like a place I wanted to go.

And the Taxim Square protests were kind of in full swing at the time. And I had this very funny argument with my mom where she was like, no, no, no, no, you can't go to Istanbul right now. It's dangerous. And I was like, no, no, no, that's like CNN dramatizing things. Like, it's actually not tear gas like on the street that my hotel was on. But Joe and I and my co-finer were talking about like, why can't I actually see what's happening on the street that my hotel is on? Like, what if there is a way for me to rent someone's eyes and ears and kind of see through their eyes? And so that was kind of the inkling that led to us, you know, prototyping solutions to this.

and we went through kind of some, a bunch of different variations before we arrived at what became Periscope. And along that journey, probably eight months in, while we were still in private beta, Twitter found us. And a few of the bleaters, you know, Jack, Dick, a few of the other folks at the company used the prototype and reached out to us. And basically, very quickly, we're like, we would love for you all to pursue this vision with Twitter support. And that was kind of fast forward. and we ended up joining Twitter, and that was almost five years ago. It would be five years in January. And how did you get from there to now leading Twitter's product efforts?

I would say I've had like three different chapters of my time at Twitter. So for the first two years of my time at Twitter, I was hyper-focused on Periscope. My role was CEO of Periscope. We had basically a separate organizational structure with a separate office, like physically. We were like two and a half blocks away from San Francisco HQ, but we had our own space. And I was kind of laser-focused on Periscope. you know, from launching it to the, you know, two years after that. My second chapter, which started around that time, was leading the video efforts of the company, which obviously Periscope was

a part of, but there's lots of other video efforts, you know, VOD on Twitter, our professional kind of like publishing efforts. And that was kind of my foray into broader Twitter. And it was around that time that I joined Jack's executive leadership team and kind of started getting exposed to Twitter more generally. And not to say that in the first chapter, I wasn't exposed to it, but it was far less intimate than actually leading some of the areas of the product that worked. You're getting the real talk when you're on the executive team. Also that. Just getting exposed to the problems and the challenges and the opportunities that Twitter generally was thinking about,

whereas I was, again, like hyper-focused on Periscope stuff. So that was Chapter 2. And then about a year and a half ago, I guess it was June of last year, Chapter 3 is my new role leading product, which has been a really fun kind of progression because I feel like, It was important for me to also figure out my own reconciliation of being laser focused on Periscope, understanding and believing in how Periscope contributed to the mission of Twitter. But then over time, I actually, like, putting aside my role changing, I became more passionate about all the other opportunities and challenges with the Twitter product.

And so it's been a, I feel really fortunate having had the progression that I have, such that when I did take on my new role a year plus ago, I didn't feel like a total stranger to the company, to the organization, to the challenges. It's still, it's been a lot of learning to get ramped up on stuff that I didn't have nearly as much context around. But yeah, that's been the journey. It's been fun. That's right. And you're now, if not the, then one of the longest serving heads of product at Twitter. I'm constantly reminded, yes, not least of which by you. For the first year, it was kind of like a running joke for me. Like, whenever there was controversy, like, well, I've got like two months left on the job, so I might as well, you know,

tell you how I feel about the situation. There you go. But you made it anyway. Well, great. Well, knock on wood. Yeah, this interview is the end of it, Kayvon. Be careful. I feel like this is a setup job, guys. Well, speaking of setup, you know, Twitter is a product that people have opinions about, you know, from time to time. And we thought it would be a fun and novel thing to, you know, just kick around some ideas.

But it's sort of a way of getting into that. I want to take something topical, which is that the day that, the day that we we're recording this, y'all rolled out searches in direct messages. So now if you're looking for someone who you have DM'd, you can now look up their name. And I think, you know, we could all agree that is a good and obvious thing. And it seemed like it maybe took longer to arrive, you know, than it might have. And I know that Twitter has historically had some challenges around shipping things. So I wondered if you could maybe give us some insight on where you're at with the product development process and how is it that a feature like that, which again, very good, but arrives on this day in 2019.

So fair question. It might be helpful for me to maybe start with some context around DMs generally. So DMs, as much as I feel like I can pick up that you love DMs, I'd love to hear more about that. You may be surprised to know that DMs for a good chunk of time, like in the order of years, were basically as a product feature on maintenance mode. And so the reason that you didn't see a lot of features like search or any other set of things that you might reasonably expect in a modern messaging product that's private, the reason why you didn't see a lot of that stuff land is not necessarily because it took us a long time. It's because we made an explicit decision not to prioritize it. Now, rightly or wrongly, you can debate that. But we changed our perspective on that

about a year ago where we not only decided that we shouldn't have a service like DMs on maintenance mode, but that we actually should have a strong point of view on why private messaging is important, because it is somewhat counterintuitive as a service whose purpose is grounded in serving public conversation. You'd be right to say, wait a second, why are you investing in private messaging? And our perspective on that is, you know, public conversation, at least in the form that it exists today, is a pretty high anxiety form of talking. It's a combat sport. Yeah, you're getting on stage in a public square and saying something that is subject to the public scrutiny of everyone. It lasts forever.

It's subject to the popularity contest of likes and retweets and all the other social engagements we have. And it's nevertheless very powerful as a form of broadcasting something you care about. But that's like on the spectrum of talking, that's on the very like public last forever subject to public scrutiny and public replies. And, you know, there is this other side of the spectrum where you can have more private conversations, whether it's one of long or groups. And then there's lots of rich stuff in between. Like, we are having a public conversation that's in a small room right now, right? Like, we're having three people and a couple listeners in this studio that eventually will be published. Like, this analog that we're talking, that we're doing right now doesn't exist as a mechanic on Twitter,

and we think that's a miss. Anyway, I point out the spectrum to you, because I believe DMs fit in very clearly into that spectrum. It happens to be on one kind of severe side of the spectrum, which is the more like private, you have room to air out your thoughts one-on-one or with or group and sometimes you need to scratch the itch of what you're thinking to really feel like you can articulate it well enough in public. Like, we believe we should be providing a venue for people to develop their thinking before they put it on the public record. One, two, oftentimes when you see things that are happening in the world, you may want

to talk about them in a smaller forum, either with a group of people who have shared topical interest in that thing or whether literally in a private forum. And DMs very clearly fulfill that need, and we felt like we'd been underserving them. And lots of kind of table stakesy ways, which is why now that we're prioritizing, and we literally have built a team from scratch. Like, for the most part, like we have a new PM, we've got a new engineering team. Now that the team is healthy and has a charter that has been ratified, like, they're turning out, like lots of great stuff.

They're working on really great. And you haven't even seen the best of it. Like there's a lot of stuff that the team is working on that I'm excited for you all to see that isn't in public experiment yet. And we wouldn't have, that would not have been my answer three years ago because we were, it was on maintenance mode. I love it. I mean, this is exciting to me because one, just like as somebody who wants Twitter the company to succeed, messaging is the lifeblood of so many great social products. And I know that Twitter is really focused on how many daily active users it has right now.

Nothing builds DAU like a messaging product, right? So I hope eventually it spins off into a standalone thing. I hope it has all kinds of like cool widgets and stuff. Eli, would you use, like, a standalone DM app? You know, my DMs are closed, and so I actually, I think, has kind of an upside and downside. The upside is, it's a very high-quality conversation, right? It's, like, just people, like, that, you know, I'm in a follow a relationship with. But I often find that that's where the conversation starts.

That's the easiest way to get a hold of me. Like, we already follow each other on Twitter. We're, like, maybe talking in public, you know, like, make the switch DMs. And then it's like, here's my phone number. Here's my email. Like, let's move this to the place where I actually do conversations. The downside is they're closed. So, like, it's a very limited set of people, which I guess in other ways an upside.

I guess my question here is, like, there are other table stakes of a messaging service now. Like, I would like to see DMs be encrypted, right? I would like to file attachments. Like, there's all kinds of crazy things you could add to a messaging product. But some of them are really important. Like, encryption is important to me. Is that something you guys are thinking about? Definitely.

Yeah, I think there's a long list of encrypted. There's a long list of, I think, features that are table stakes, some more square. important for why DMs are important for us. And I put an encryption in that bucket. It's not one of the first things you'll see us launch. That's mostly because, you know, as a first project as we've kind of built this team from scratch, basically, we felt like there were other projects that we could sequence first, whereas encryption's a pretty gnarly one that, yeah, that's a whole, like, policy thing and you got to, yeah. There's those considerations. There's also, like, there's a minimum viable set of things that you need to build for encryption to be useful, i.e., like, encrypting

one-on-one communication is different than encrypting group conversation. encrypting media is a different level of scope than encrypting just DMs that have text in it. So, like, we're definitely thinking about all these things. But, like, philosophically, like, absolutely, we believe that we should be able to support encryption. But I'm curious, like, other than that, like, what are the other, you all are obviously power users of the product? Like, what are the other things you would expect specifically within a DM product and Twitter? I mean, I'm a believer that single-purpose tools are great, which is why I tend to, like, stand-alone apps. Some people think that's crazy.

They want the entire experience in one thing, and they get frustrated when they have to switch back and forth. But I think that if you had kind of a blank canvas like that, you could do a lot more with it. I think you look at some of the things that other social apps are doing with their messaging apps around like ephemeral sharing, I think is really interesting. Group conversations, I think, can be really good. You know, my DMs are open. And so it winds up being kind of my inbox to the world. And so maybe if you want to give me a tip, like you reach out to my DMs first. would love an encrypted way for people to do that. But, you know, I don't know. I'm just really excited to hear there's a team working on it. And, like, to your point, DMs are where a lot of work gets done that then later shows up as a better public conversation.

I had one of my very favorite writers, like, wrote something that was, like, very gently critical of me today. And, like, we didn't go back and forth in public about it on Twitter. Like, he just sent me a DM and we, you know, traded a few friendly messages. So I think that kind of thing happens a lot. Yeah. It's sort of telling to me that there are many customers who are frequently DMing for whom those same people are not frequently tweeting in public. And there's something to learn from that around the comfort of having a space

where you can control the audience who can participate in that conversation. And like DMs are so clearly that. But again, one of our like biggest strategic initiatives right now is more generally. We call it sort of conversations, which is pretty obviously simple and intuitive word. But for us, it's like that spectrum that I was talking about earlier, filling in that spectrum and being super thoughtful around where the biggest customer needs are that we don't currently fulfill as a product or that people are kind of hacking together. Because despite not having great mechanisms for this, we have customers who hack this together into the product. They create multiple accounts and they'll have some of those accounts be protected or some of them be public but have a very narrow follow graph. And they're essentially trying to do what we're describing here, but by hacking together the product in ways that it wasn't necessarily intended.

And so we can learn a lot, and we have learned a lot just by productizing those, you know, we call them help wanted signs in the jobs framework. Like people are asking for help and they're hacking the product in a way that we can learn from and productize. Yeah. Which, by the way, Twitter's history is filled with, I think some of our best innovations have been productizing things that customers have hacked together, whether it's the app mention or whether it's the hashtag. And so I think there's a lot, there's a lot to learn from that, the retweet. The retweet, yeah. Well, let's talk about another product that y'all have introduced recently, or at least, described, which is this interstitial that you're going to put in front of some tweets that might

violate your standards, but you've decided to keep up because they're newsworthy. This week, we saw the president tweet some things that suggested that maybe there would be a civil war if he was impeached. And some people wrote takes about how, you know, Twitter, if it shouldn't ban his account outright, should maybe at least have used the interstitial in that case. So it seemed like a good chance to ask you, like, how and why did you build this kind of interstitial thing? And then how does the company think about when it might actually use it? Yeah. So I would say that there's two different kind of components of this conversation,

at least in how we think about it, that I want to distinguish because I find that they sometimes are conflated together. There's this notion of what's in the public interest, which we've talked about, like, us having a perspective on as a company in our policies. And then separate but related is this product feature, which is the interstitial or the sort of like information pain that we put on on tweets. and how and when we would use that. So let's talk about both of them. So with respect to the public interest concept, really what we've said is there will be, in particular from people in powerful positions or public officials that have a certain, you know,

threshold of following and greater, we will bias as a company towards keeping up tweets, even if they violate our policies as a default. And that's not something that everyone will agree with. I know Kara, for example, like openly and blatantly disagree with that, and that's fine. But like, we have very clearly said as an organization that we believe that these tweets from these types of folks, we intend to keep on the platform, even if they violate our policies, because we believe it's in the public's interest to know. Related to that, the reason why we built the interstitial is because when we make such a decision

that something does violate our policies, but that we're going to keep it up anyway, we want to be able to make that clear and transparent and understandable in the product experience so that you don't rely on a tweet from Twitter or a blog post from Casey or a CNN article or TV broadcast or whatever to understand the fact that, oh, yeah, yeah, yeah, this did violate the policies. And here's why Twitter didn't take it down. So that is the thesis of why that feature exists. And it would only be used in the case that we deem a tweet as violating our policies, but that we keep up nevertheless because of this public interest notion.

We hear lots of feels about that. And the feels, and again, I'm curious to get your thoughts, but my understanding of the feels are there's really two different notions, right? Some folks really do disagree with that decision to keep tweets up that we believe are in the public interest. And that's by respect that, but we've made the decision that that's how we want to operate. Two, I think there has been, and this is where I think we've got work to do, there's been a lot of confusion around how we make those decisions and what criteria we use to make

those decisions. And I think in the absence of that clarity and transparency, it's super understandable that there is frustration. And like borderline sometimes like lack of trust on like how that process even works. And that's something that I think we understand I agree with and we are going to hold ourselves accountable to coming back publicly with an articulation of how we make those decisions. What is the criteria we use? And so that it's more understandable how we might apply that to a specific tweet or not. Separate from all of this. And by the way, I should say on the interstitial, sorry, on the.

the public interest exemption, there is a case where we will actually take a tweet down. And that is if one of these individuals says something that is inciting violence or harm against a private individual, it doesn't matter who that person is, we will take action on that tweet. That is a very specific exemption. It has to be a private individual, not a public figure. And again, people might disagree with that. But that is our stated policy right now. And I do believe that we have been enforcing our policies consistently given that standard. Now, separate to all of this, and I might be preempting some of your questions, but just to get everything out on the table, separate from all of this, there is an independent question around

what do we believe that policies should be around which we use a remediation like the interstitial or what other such product features should we create that allow us to take actions that are different than just keeping content up or taking content down? I believe that one of the challenges we've had as a company over the last 13 years is that our primary form of remediation has been, do we take this tweet down or not? And that's a super heavy hammer that doesn't provide a flexible enough product experience around how you remediate different forms of severity. And the interstitial is like scratching the surface of one of these such remediation that gets you away from just enforcing based on take down or take up. and I do believe that there's other things that we should be doing and thinking about that go beyond that. So that's kind of like maybe like three different components of how we think about the problem.

I mean, one alternative is that the official Twitter account could quote tweet it and say, damn, now that's a bad tweet, something to think about. I don't think we need that. We have you for that. Yeah, it's true. I will point out a bad tweet from time of time. I will say I am actually mostly aligned with you on this policy stuff. I think that if you were to ban some really bad politician, that person's fans would just create an account. and that politician would be posting elsewhere and some bot would be screenshoting it and posting it on Twitter.

I actually think the technical question of how you stop someone from using Twitter is maybe harder than it looks, particularly when it comes to a politician who has millions of followers. I also do get really worried about politicians using the platform to inside violence. And I hope when the time comes that a politician really does use the platform that way, that Twitter sticks to its guns. But, you know, the alternative is that Twitter has to spend a lot of time deciding, like, what are the boundaries of free speech in America? And I just don't really want to trust any public company with that. So, maybe. I mean, the thing you're describing, when you describe remediation against policy, I mean, you're the head of product that is a huge Sillian Valley tech company.

And that, like, quite frankly, is some cop shit, right? Like, you have to build a product that enforces policies in a way that I think Twitter, in particular, sits right at the center. of. You have a trust and safety team, a legal team that's developing policies, rules for how people behave, and then your product either incentivizes people to behave in certain ways or punishes them when they don't or provide some transparency into why other decisions get made. I think I'm fairly sure that is like a unique moment in software history. Like I don't think like Windows 3.1 had that architected into it. And I'm very curious how you think about that. The traditional head of product role is like, you know, I'm going to build some cool software features.

People are going to use them. They're going to use a product more. Hopefully we'll make some money. Right. Like, this is now very much directly. When does Twitter censor the president in the United States? Right. And like, how are you thinking about that role, the split between product and policy and how to manage all of the many demands that you have?

Well, I guess you said a lot of things there. I'm trying to tease the part what the question is. Well, do you believe Windows 3.1 had a policy? speech enforcement. Yeah, but it's like entirely new for the product to enforce the law, right? Like a normal person walking on the street, you know, if Congress changes some law, they just keep walking on the street. Twitter changes its policy about what you can and cannot tweet.

The product can actually affirmatively enforce it, right? And that's not normal necessarily. That's much more impact on individual people and how they behave. So how do you think about balancing that responsibility against keeping the platform as open as I think Twitter's values want it to be? Yeah, I mean, I think what you're highlighting is that, We are, we're in this fascinating and fortunate but challenging position where Twitter plays a really important role in the world, I believe. And we have to be super conscious about that role and that responsibility and take it very seriously.

And that's why I feel like there's a fair share of like incredible excitement, passion and like slight degree of being terrified about like doing our, having having the roles that we do. I personally find it extremely challenging and really gratifying. But I think maybe more specifically to your question is I don't, I think it's really important that we not divorce things like the product from the policy and the enforcement around the policy because it is at the end of the day like an end-to-end experience. And I think quite frankly, one of the reasons why we've maybe over the last, you know, decade haven't moved as quickly as possible to solve some of the foundational issues that we're now hyper-focused on is because we did think of them as somewhat divorced.

And we didn't really acknowledge and prioritize as much as we should have early on some of these really critical things. So like the fusion of those things together, I think, is why we've got a lot of work to do and why, like, health is such a crucial priority for us that is not a policy priority versus a product priority versus an engineering product. It's just like it's a Twitter priority and it's an end-to-end set of things that we have to do. And there's an interplay between how we can, you know, we have existing rules. we have to make sure that we're enforcing those rules better. But then sometimes the rules aren't right. We've got to create new rules. But then also sometimes the product isn't right.

And we are using as a crush the policies and the enforcement of those policies to really make up for deficiencies in the product. And so the creation of new product vectors might then lead to new rules that then have gnarly enforcement criteria around them. So I think it's an interplay that we have to be super thoughtful of. And there isn't any one silver bullet. Like we have to kind of do all of the things. like enforce the current policies that we have better. Think of where our current policies need to be evolved. And also, quite frankly, where we've underinvested the most,

evolve the product experience. For example, and this is like a hypothetical example, if you had the ability to tweet and decide who can reply to your tweet publicly or not, that like a whole class of what we would consider abusive today, i.e., like people getting all up in your mentions and replying to your tweet and having like a dog pile on a conversation you're trying to have. Suddenly that abuse vector is fundamentally different because if you introduce a product mechanic that gives you the control of who can respond to your tweet publicly.

So that's just one example of I think we've been underinvesting in how product mechanic changes and in particular incentive changes can then alter the ecosystem of how people interact on the platform. So where does that policy team sit with you in that kind of decision matrix? I mean, to me, the most interesting thing about every sort of person in your kind of position I talk talk to is where do we locate trust and safety in policy, right? Like, if the product and the policy are influencing each other so directly, is that one integrated team? Is it, you know, we're giving presentations to each other every week? Like, how does that actually work for you all? Yeah, I think, you know, it's something that we talk about a lot. And I think the company's gone through its own

evolution of how both in terms of structure and decision making and, and priority. Like, there are times where we just didn't prioritize any of this stuff from a product standpoint. So I think the way I think about it now is something like, I mean, this is true for everything, but in particular health takes a village. It takes the policy team. It takes our sort of, we call the Twitter services team, which does a lot of our, which does all of our enforcement. It takes the product engineering, design, resources, public policy team. At the end of the day, like, I personally feel accountable for the product experience. And so we have to make decisions that are the right fusion of policy and enforcement, but also like the scaffolding of how the product

experience gets used. And those are highly collaborative conversations that in particular are complex when it comes to creating new product vectors. And when it comes to, you know, the more, the less new they are, you can kind of decompose them very clearly. Like if we have a policy right now, like the enforcement, there's a, you know, the enforcement of that policy is relatively less complex than, hey, we're trying to build this whole new surface area that actually has a different thesis around how we want people to use it. And like, what should the net new policy for this feature or this interaction be? And how would we enforce that? And those are as we, in particular, as we evolve the product in new ways that I think are fundamentally different than

how Twitter has worked in the past, like those will be equally complex. Support for the show comes from Framer. Framer is an enterprise-grade, no-code website builder used by teams at companies like Perplexity and Muro to move faster. With real-time collaboration and a robust CMS, with everything you need for great SEO, not to mention advanced analytics that include integrated AB testing, your designers and marketers are empowered to build and maximize your dot com from day one. So whether you want to launch a new site, test a few landing pages, or migrate your full.com, Framer has programs for startups, scale-ups, and large enterprises to make going from idea to live site as easy and fast as possible.

Learn how you can get more out of your dot com from a Framer specialist or get started building for free today at framer.com slash verge for 30% off of Framer pro annual plan. That's framer.com slash verge for 30% off. Framer.com slash verge. Rules and restrictions may apply. Support for the show comes from Upwork. The days of doing it all, all by yourself, are over. There's no romance and burning out while you're trying to scale. Instead, you can check out Upwork.

Upwork helps grow your business by giving you fast access to specialize talent across more than 125 categories so you can fill skill gaps, launch projects faster, and scale without committing to full-time headcount. And finding the right talent is easy. You can browse profiles, review past work, and get help scoping the role so you can get started quickly. Seriously, you could connect with the right freelancer in just a few hours, especially when you sign up with Business Plus.

Their AI-powered shortlisting pairs you with the top 1% of talent in under six hours. No endless searcher required. You can visit upwork.com right now to post your job for free. That's upwork.com to connect with top talent ready to help your business grow. That's upw-w-r-k.com. Upwork.com. Let's talk about another kind of nexus of product and policy, which is verification. So I was thinking about y'all last week when YouTube really stuck their foot in it over verification.

If you don't remember, they announced that they were going to take away checks from various people and that also they had been just verifying anyone who had more than 100,000 followers, and they've since sort of recanted all about that. You know, verification might seem like a niche issue, but to me it kind of goes to the heart of can I trust that I'm talking to the people that I think I am on this platform. And, you know, if I can trust that, I might use it more, and I might have a better experience.

So I would love to get an update from you on verification. You guys have officially paused this program, but you do continue to verify some people, and we'd just kind of love to hear what, what the heck is going on there? Yeah. So I'm happy to start with maybe revisiting why we paused and then how we're thinking about it now and like on the spectrum of pause to launch to where is it now.

So we paused verification for a couple of reasons. One, we just did not have a rigorous or consistent enough set of criteria that we felt like matched our intent behind the verification feature. We were sort of conflating the notion of authenticating whether someone is who they say they are with the notion of whether they are like a public figure. And those are actually quite distinct things that you would validate in different ways. That and, you know, we did not have a sufficiently transparent and self-service process by which people could apply. And on top of all that, it just was as much as I agree with you, it's it is a critical feature actually.

It's a very powerful incentive, actually, that exists in the product experience, one of the only account-level incentives that we offer, other than the follower account, which is also very powerful. All that being said, it just was not our most important problem. And, you know, we decided to put it on pause, as messy as that is. And what we mean by that is, like, doing the work to fix everything that I said, we decided was not a top priority. In particular, relative to all the more important problems we had within health that we can talk about if you're interested. So that's why we paused it. And, you know, we did say, you know, when we made the pause decision that we will continue to, as unideal as it is, we will continue to make sort of Twitter-centric, like behind-the-scenes verification decisions when we think it's in the public interest to do so. Again, that's not ideal because there's no rigor around the criteria or process and blah, blah, blah, blah, blah.

But that is the state that we entered when we made that decision. So fast forward now. What's different is that we do, I was talking about how DMs were on, maintenance mode and we've made it a priority. We similarly have decided that building a strategy around identity needs to be a priority. For all the reasons you just said, it's actually like a pretty critical way of understanding the credibility of who you're talking to. And by the way, verification to me is one of the things that we feel as important as part of the strategy is really teasing apart what we mean by verification because there's actually, we believe that we should be

exploding into more decomposed features what the notion of verification actually means. Like you can actually decompose identity into lots of different things, like validating whether someone is a human or a bot, or validating whether someone is who they say they are from like a real name standpoint, or validating whether someone is a public figure is another form of that. And I think we just had this very coarse feature with the blue tick that kind of meant a lot of things to a lot of people. And two, the criteria around it wasn't particularly clear to folks, nor was there a self-service process.

So we're now building a strategy that allows us to, we think, solve those problems in the right order. And so this isn't something that internally we consider paused anymore. It's going to, you know, it'll take some time for us to actually manifest the work into something that people will be able to see. But that's different than, like, a year and a half ago, my response would have been similar to your question about the edit button, which is like, cool. It's a problem we could solve. It's not anywhere near the most important problem. So we're not solving right. But did you preempt the edit button question already?

I find with Casey, I have to. I actually preempt some of my tweets about it. product launches with the edit thing because at this point it's like it's such a reliable meme that you know the first comment's going to be yeah but edit button so I just figured I know my audience well look I'm very excited because once you start working on verification that will mean that your last remaining problem to solve is the edit button and then maybe that can also come out of pause look you think that's the last remaining problem I think is this is why Casey's not in charge of product at Twitter yeah that's very true I mean I will say it's a feature you should have wait I want to

focus on verification for one second case you brought up YouTube I think you two had a botched rollout, botched communication. They did have a relatively good core idea, right? Which is we're going to have different kinds of badges for different people, right? So if you're an actor, you get the verified actor badge or whatever it is that they were going to do. At the end of this all, are you still just delivering blue check marks, but through different routes? Or are you going to show different kinds of things? Do I get a different verified badge that says, this is actually Neelai Patel versus, I don't know, Darth?

like, this is the real Darth. I think there's a couple dimensions that are interesting. So one, in terms of what the product experience would want to do, like, and what the customer wants. Like, what exactly are you verifying, I think is one question. And to me, there are lots of potential dimensions like, like, are you verifying humanness? Or you verify, which what I mean by that is, like, someone is an individual, validating that someone is an individual versus a bot versus a business. Like, that is a dimension of verification that you could choose to implement in a product.

validating credibility is another dimension. And how you do that, I could imagine lots of different solutions, too. It seems like the approach YouTube was trying to take was kind of around credibility as denoted by some, like, category that YouTube decides exists. I think that that's not an inherently flawed idea. I think maybe the way they implemented it didn't go over super well. And I actually don't, I didn't really, like, grok the specifics of how their solution worked. But I think that the notion of validating someone's credibility on.

a certain topic is actually pretty interesting, especially in the context of Twitter where, you know, Casey Newton might be the journalist that you know from The Verge, but actually he's got like a quantifiable, perhaps, level of credibility around WWE that if he's like jumping in... It's true, by the way, yeah. Sorry? That's true, by the way, yeah. Well, you would think at least. I'm giving you the benefit of the death.

Thank you. But I think that's an interesting notion where, given how people talk on Twitter, you know, one of the challenges, I think, is that people kind of like boxed in into the follow graph that they have. One of the things that we hear often from customers is I feel like I have a lot of followers who follow me for a certain dimension. But, you know, like, I remember having this conversation with M.G. Siegler. I hope he doesn't mind, but he was one of our investors back in the day. And he was talking to me about Twitter. And he's like, I'm followed by the VC community. But whenever the Michigan game happens, all I want to do is tweet about the Michigan game. But, like,

all of my, like, investor community and Silicon Valley community friends are like, oh, I want to mute MG. That's me during every WWE. pay-per-view, by the way, which is why I recently created an alt account, but go on. Well, exactly. And I think this circumstance that we're talking about here actually cross-cuts a lot of interesting parts of the potential product experience that we're excited about, be it giving people a space and control around having conversations with people who have shared interest. That's one of them. Two, how do you even find the people who have the same shared interest as you? Like, that's kind of one of the superpowers of Twitter is helping you stay informed about

the things and talk about the things you're interested in. But it's actually quite, challenging and indirect to do. The entire service is predicated around following accounts. And following accounts is very powerful, but very distinct than following an interest. Like if you care about the WWE, you have to know the people who are experts who have credibility in that topic or in that sphere. And that's a lot of work. It's not as simple as following the WWE account, i.e. Casey might have some good thoughts, maybe. And so that's, I think identity plays into that.

We have an initiative we're calling interest that is a huge, I think will be a fundamental change to how we all use the Twitter experience that you're starting to see the beginnings of with what we're launching. Like we had one of my favorite launches recently is we elevated lists, right? Like you can swipe between your lists. To me, that's like scratching the surface of our work to give you more control around kind of diving into what's happening. specific interest sphere that you have. And today those interest spheres are a collection of specific accounts in a list because lists existed as a product primitive. But you can imagine forming more sophisticated spheres that are collections of people, collections

of interests where Twitter then does the work to find the people who are most relevant and only show you their tweets if it's about that interest rather than like every fucking thing that Casey can says. And I let's swear. Oh, I mean, I never would, but you can. I don't believe you. Anyway, so I think I realize it's sort of, it touched on identity and then some other stuff, but I really do view identity as a cohesive layer that we have to think about in the context

of the whole cohesive application. I love that you guys are focusing on this, and I hope you find various ways to, like, elevate and display that stuff. I've always thought that verification features should be open to way more than just, like, the journalists and the politicians and the public figures. If somebody is on Twitter and they have verified their real name and their phone number, number, like, I want to know that, right? It's very helpful to know, you know, when you're dealing with a Russian bot versus an American, right, for all kinds of reasons. So I think verification

and all of the kind of maybe subclasses that you described is just like such a great way for y'all to start thinking about it. Yeah, and there's some explicit signals of credibility that, you know, customers would opt into. And there are some implicit ones that we've already built into the product experience, right? Like, it's no coincidence that when you look at your tweet, we now show you what client you tweeted from. It's no coincidence now that when you look at your profile, we show you automatically, like when that person joined Twitter. And these are all sort of implicit signals of credibility, I think, where we haven't done as much work as on the explicit signals. And I think that's a really interesting, there's a lot of interesting potential there.

Yeah. Let's do a little bit of a curveball. How does Twitter think about ephemeral content? And will I ever see Twitter stories? That's your curveball? I mean, not like a crazy, you know. Well, you preempted the edit button, man. That was,

that's like the last one. He's got like the ultimate curveball and the hopper. You've talked so much about policy and trust. What about the edit button? That's how you get it. I would go back to,

you know, we were talking about that spectrum, right? On, you know, on one hand, you've got private. On the other hand, you've got public. Today, really the two form factors

we support are tweets as a mode of conversation and DMs in another mode. And arguably, maybe somewhere in the middle you've got protected accounts, which is kind of like sits in there. To your question of ephemerality, I view that as another dimension that is really important for some customers, for some specific set of circumstances where you want to talk to people, but you're not quite sure you want it to last forever yet. And so I think as a dimension to focus on,

as a specific customer problem, like absolutely, I'm very interested in exploring how we might give customers more control. Where effemerality is just one of those dimensions. I think there are other dimensions that, you know, while we can get excited and talk about effemirality because there's lots of other kind of standards of how other apps do this, I think other dimensions like control around who can see

or control around who can participate is really critical. You know, we talked about my example of you tweeting something and, you know, in fact, actually, the Kara Jack conversation is a perfect example of this, right? Like, Kara and Jack were trying to have a conversation in public. But this is real quick. This is a sort of thing that happened on Twitter.

Karraswish, our good friend and colleague, talk with Twitter CEO, Jack Dorsey, and they had sort of extended back and forth on Twitter this year. Yeah, I believe that the intent behind that conversation was to have a fireside chat where the two of them were on stage and the world could watch. But it was actually quite difficult for them to do that for a number of reasons. One, like the mechanics of the conversation and following it were really challenging. And a lot of that were addressing with our work in the Little T app, the, the public beta that you've hopefully seen. But I think another one of the issues was it's actually

quite difficult to have a fireside chat when you have like a billion people screaming in your year. Like imagine we had like tens of thousands of people in the studio with us right now talking into our ear while we were talking to each other. No, thank you. And so I think that's another dimension of how our conversations features work or don't work today that's really important to us. That and ephemerality, I believe, is up there as well. So you will see us, you should expect to see from us various product features, because there's no silver bullet for all of these things, but you should expect to see various product features that try and nail different intersections of the spectrum. I'll tell you my request is I would love to set all my tweets to just go private

after a year, basically. And the reason is just that cultural standards change enough that I either have to delete all of the tweets on a regular basis or I could just like set them private, because we've now seen bad actors sort of digging into people's old tweets, taking them out of context, and like wreaking all kinds of havoc. So I would just love a way to never have to think about that again, basically. I mean, I pay some, I pay other companies to delete my tweets for me. I'll just pay you. I mean, it's, in terms of, what service do you use for that? Tweet, deleter.

Dot Russian hackers. It's like, it's not, I don't love it. They're all super shady looking. Yeah. Like, I really truly do not love that this company has like... Your Twitter login. Yeah. That's actually kind of shocking to me.

It's clear that you care so much that you're willing to... I mean, journalists careers end because someone digs up an old tweet, you know? It's like, it really is a very existential. I realize we're like in pretty niche territory, but... No, no, we're actually not, though. Like fear of speaking in public and fear of retaliation or fear of being harassed. And harassment, harassment means many. different things to many people or fear of being held accountable for something that is like

not what you meant. These are some of the biggest reasons why people don't tweet, which is why like we actually take this very seriously. And there are many different product solutions. I mean, auto deleting tweets is something that we could do. We have, we have thoughts on other things that we could do as well. But like the point is like this problem we're hyper conscious of. And if we believe that getting getting people to feel comfortable talking in public is critical and that there need to be bridges to that mechanic because talking in public is pretty terrifying for most people. So yeah, we're super interested in coming up with multiple solutions to solve it. Do you think about literally auto-deleting tweets as part of ephemerality, or do you think that's a

different kind of product? No, I do think it's a form of ephemerality for sure. I would say it's a, don't take this the wrong way. I think it's a less interesting solution to the same problem. But it's absolutely a form of ephemerality. But we're in. interested in exploring a couple other solutions that have the same potential effect of you not having to worry about what you say lasting forever, but giving you some of the other control that I think is missing. Because I don't think the ephemorality alone solves the most important problem. But we may realize that we should still offer that. So I'm not dismissing it. I just think that we've got some other ideas around how we might solve it in interesting ways.

This is a curve. It's not really curval. It's it's thematically connected. But I was just thinking about how a lot of people I know experience Twitter and things people personally experience. And I think I'd be remiss if I didn't ask you about harassment. Like Twitter as a service is a primary vector of harassment and abuse for a lot of people, something I've experienced waves of. Something in particular, like the young women on like the Verge staff experience quite a bit. How are you thinking about fixing the product elements that enable harassment in that way? Good question. So first of all, everything you just described is why health has been our number one priority for the

the last three years. Health, I think, really distills down into a few different things. Conversational health where abuse and harassment are kind of central focus areas of is obviously one bit of that. And then sort of information integrity is another piece of that. Like if as a company that believes that our purpose is serving public conversation, that conversation isn't healthy, if you can't trust the integrity of the information on it, then what is the point of the whole thing? Like, the whole thing is existentially at risk if we can't insure those two things. So That's the basis for why health has been our top priority for the last few years. In particular, conversational health is critical because the fuel that helps people stay informed on Twitter is atoms of content that were tweeted by real people.

And if we are not able to serve the needs of those people, whether it's the tools they need to talk or whether it's the risk mitigation to ensure that they're protected and not being abused or harassed, then we can't expect them to feel comfortable talking in the first. place, which jeopardizes the whole purpose of the platform. So that is why we've prioritized it. The way I think about it to your question is we have to do a mix of things. There's no silver bullet, but we have to clarify our rules, make sure we have the right rules. We have to enforce our rules the right way. We have to make sure that we're building, that we have to make sure that we're proactively enforcing those rules because a lot of the problem right now is customers have to face the burden of experiencing and then reporting the abuse, which is, even if we have the best reporting flows, it's almost too late at that point. And so one of our

biggest initiatives this year in 2019 is something that we call proactive health. And literally our goal is how do we enforce our rules before our customers have to enforce them themselves through reporting? And so I think by January of last year, about like 15% of our terms of service violations that we actioned were done proactively. At this point now, it's about 50%. So we've made a huge amount of progress actually just enforcing our rules. more proactively so that you don't have to report it or even see it in the first place. So that's one way that we think about it and one set of things that we do. And then the other is actually building more product mechanics that give you control around having a healthier experience

so that you don't rely on the rules and Twitter's enforcement of those rules in the first place. So one example of something that we've launched recently that, again, is scratching the surface, but it's an example of what you can expect to see more of is we launched a feature that we call author-moderated replies, where you can hide globally a reply to your tweet such that no one else can see it in the conversation thread. And that's the first time we've ever given anyone the power to make a moderation decision that affects what other people see. Obviously, you've been able to block or mute people, but those are sort of local decisions. And we think this is important because we want to change the balance of power between the various actors in the conversation. You've got a few different actors.

You've got the original tweeter. you've got the replyers who are participating, and then you've got the audience, like everyone else. And the balance of power has been totally off-kilter in the past. Like people replying have had too much power in abusing or harassing, which again, like everyone has different interpretations of what that means to them. And we want to change that. And one way we're changing that is author-moder replies where you can hide something.

I would eventually love, like our plan is to open APIs around this. You can imagine developer tools that allow you to create super customizable actions around what you want automatically hidden in your replies. So that we shouldn't have to build all of these first party features. Like we should build core ones like you being able to hide a reply. And you might kind of like an automator you can create super complex actions around automating things. Like you ought to be able to create rule sets that you can share with other people. So that's an area. A lot of the conversation we were just having around ephemorality in terms of giving people control around who can participate.

They're in service to the same goal. If we give you more control around how conversations unfold, that is one other thing that will help you feel like you can have healthy, controlled conversations about topics that are interesting or sensitive. Whereas, you know, some people don't feel comfortable having certain types of conversations right now because they're entirely dependent on us enforcing a global set of rules that spans cultures, geographies, time zones. And, like, that's just inherently not scalable. Do you think you have to shrink that? I mean, that's something I think Casey writes about the interface all the time. And we talk about maybe internet scale is not the right scale for the world. Like, you need to create smaller communities more proactively.

Is that something, you know, you're talking about lists and maybe interest groups, but just around people and individuals and audience. Is that something you want to do more proactively or let people do themselves? I mean, the way I would articulate it is, the way we think about it is we want to invest more in decentralizing moderation capabilities. Right now, it's entirely centralized. And I think in order to pursue a concept like that, we have to think about other product mechanics where that sort of decentralization makes sense. Because one of the beautiful things about Twitter is that it's extremely fluid, right?

There is no clear space. Like everything exists in the Twitter sphere. And it's a very powerful but fluid thing about the platform. And so it's one to think one of the really interesting problems that we're looking to solve, you know, moving forward we've been talking about a lot of recently. is how we can decentralize more. And so yes is the short answer to your question. But I think this is why we have so much fascinating work in front of us. And so many problems that we have to focus on, but we're deliberately trying to be,

and hopefully you all see this with the work that we're doing, we're trying to be a lot more ambitious about what problems we sign ourselves up to solve. And we're trying to be a lot more ambitious about the speed at which we're solving them. And believe me, I'm sure people are still frustrated. with a pace that we're moving, but like we have, we're trying to deliberately increase that pace and really solve some of these hardy problems. The truest curveball of them all. Apple is putting out a catalyst, which is their framework to bring iOS apps to the Mac.

Twitter onstage, Demote is one of the highest profile catalyst apps. We haven't heard a lot about it from developers. Are you involved in that app? Do you think that framework's going to work? Do you think that iPad apps on the Mac or the future, is that something you're excited to be a part of? What I'm really excited about is that Apple is making it a lot easier to actually port iOS specifically iPad apps onto the Mac. And I think, you know, we had to make a tough decision a couple years ago, a few years ago,

to deprecate our Mac presence because the cost of maintaining yet another client was just too high. And we couldn't keep up to pace with, you know, the features that we had on our other clients. And so, you know, we just had to focus. So what I'm excited about with the catalyst stuff is that that equation is a lot different now and that the ease at which developers can, you know, bring their iPad apps to life on the Mac is a lot easier, relatively speaking, in a way that can ensure that as we do product development on iOS, that it's coming to the Mac at a far lower cost. And so, yeah, I'm very excited about it.

I've been involved. I'm, you know, I see the builds. And, you know, we're waiting for Catalina to launch, obviously. And we're excited to bring it to customers. Are you spending cycles being like, we got to make this as Mac-like as possible? Do you have those kinds of events? I mean, that's like one of the big questions, right? Are these just going to be sort of bad iPad apps sitting on top?

I mean, Apple's own catalyst apps are like not great. Are you trying to perfect it and make it feel very native? We definitely, like, one of the lenses through which we're looking at the product experience is like, what are the core aspects of this feeling like a desktop app that we should make sure we build into the experience that aren't as in the forefront when you think of it in the iOS context? For example, like Dragon Drop. Dragon Drop is a pretty critical fluid sort of aspect of the experience that, I, feel like has to work right in the Mac app context. And so yeah, we definitely like, we look through

that filter as we think about the product experience. And, you know, especially in the context, is like, what do we put in the V1? We may not get everything right in the V1, but like we are not, I don't think we would go through the effort if we just wanted to have like a lazy port of something that feels out of place. So that's, we definitely want to make sure it's, it feels great like anything else we launch. All right. Last question. Bringing it full circle. You are a video person. What is the future of video? Is that my brand? That's, yep, that's how you're known and always will be. No, I'm just kidding. What type of person are you? A tall one. So, you're a video person, among other things. And what is the future of video on Twitter

and should it look like TikTok? It's a loaded question. No, it's open, I don't have feelings. What does it looking like TikTok mean to you? What do you think of when you say that? Well, Twitter once owned Vine where people sort of made wonderfully creative entertainment in looping six seconds, videos and then that went away and then another company came along and made new looping videos and people are now doing very entertaining things and it's like very popular at least for the moment. So I just wonder if you think that is part of video on Twitter or if you think Twitter on video on Twitter will look totally different. So the reason I ask the question is because I feel like there's two different aspects of what you're saying. Like what does video

on Twitter look like is different than like what creation tools should we be creating? Because like you see a lot of TikToks on Twitter. The main way I find great TikToks is actually on Twitter. and they render quite fine on Twitter. What we don't provide and what we haven't invested, you know, anywhere near the same amount of time and effort in is I think a lot of the basis of what companies like TikTok do is build the tools that help creators make those, that interesting content, which is somewhat distinct from distribution and discovery. You know, I think that the way we think about it is like we have to prioritize a core set of jobs that we believe customers hire us for. creating super compelling short form video content is not a main job that we get hired for. People do it still. Like, there's lots of things that aren't in our focus that people end up using Twitter for.

Like, it's a big product. But I don't think this is an area where we're, like, extremely differentiated. And so it's not, like, a top priority for us. But I do think that the ecosystem is obviously super fascinating. And, you know, I've personally spent a lot of time thinking about how you can give people tools to create compelling video content because, you know, that's kind of what my startup did too. You are a video person.

But yeah, I think we need to make sure that we're prioritizing the things that are core to to why people hire Twitter every day and make those things better before we try and do the ancillary stuff that we're less differentiated in and I think is less important for our purpose. All right, Kavon Bagpore. Thank you very much for joining the Vergecast. Really enjoyed it. Thank you guys. And specifically, thank you for staying for the entirety of our conversation.

All right. My thanks to Kvon Bakeport, had a product Twitter. Also, thanks to Casey for joining me. That was really fun. We've got to do that again. We'll be back on Friday with the chat show, back on Tuesday with the interview show. On and on it goes, hit me up. Let me know who you want me to interview.

These suggestions have been very helpful. I'm at Reckless on Twitter. I also just love hearing your feedback on the show, tell us how to get better. It is tremendously useful. And it's also just great to hear from Odie. So hit me up at Reckless on Twitter. We'll be back on Friday with the chat show.