Skip to content
The Joe Rogan ExperienceThe Joe Rogan Experience

Joe Rogan Experience #1258 - Jack Dorsey, Vijaya Gadde & Tim Pool

Jack Dorsey is a computer programmer and Internet entrepreneur who is co-founder and CEO of Twitter, and founder and CEO of Square, a mobile payments company. Vijaya Gadde serves as the global lead for legal, policy, and trust and safety at Twitter. Tim Pool is an independent journalist. His work can currently be found at http://timcast.com

Joe RoganhostJack DorseyguestTim PoolguestVijaya Gaddeguest
Mar 6, 20193h 25mWatch on YouTube ↗

EVERY SPOKEN WORD

  1. 0:0015:00

    Five, four, three, dos,…

    1. JR

      Five, four, three, dos, uno. Come on, TriCaster.

    2. JD

      Yeah.

    3. JR

      Live?

    4. JD

      Yes. So-

    5. JR

      All right. We're live, ladies and gentlemen. To my left, uh, Tim, Tim Pool.

    6. TP

      (clears throat)

    7. JR

      Everybody knows and loves him. Vijaya, what is it, how do you, how do I pronounce your last name?

    8. TP

      Vijaya.

    9. VG

      Vijaya Gadde.

    10. JR

      Vijaya, not Vijaya, Vijaya.

    11. VG

      Yes.

    12. JR

      Vijaya...

    13. VG

      Gadde.

    14. JR

      Gadde. And your position at Twitter is?

    15. VG

      I lead trust and safety, legal, and public policy.

    16. JR

      That's a lot. That's a lot. And Jack Dorsey, ladies and gentlemen. Um, first of all, thank you everybody for doing this. Appreciate it.

    17. TP

      Thank you.

    18. VG

      Thank you.

    19. JR

      Feels, th- there feels-

    20. TP

      (clears throat)

    21. JR

      ... all of a sudden there's tension in the room.

    22. TP

      (laughs)

    23. JD

      (laughs)

    24. JR

      We were all loose, we were all loosey-goosey just a few minutes ago.

    25. JD

      There's no tension. There's no tension.

    26. JR

      And everyone's like, "Uh-oh, this is really happening." Here we go. Um, before we get started, we should say, because there were some, uh, things that people wanted to, uh, have us talk about. Um, one, that the Cash App is one of the sponsors of the podcast. It's been a sponsor for a long time and also a giant supporter of my good friend, Justin Wren's, Fight for the Forgotten Charity, building wells for the Pygmies in the Congo. This is very important to me and I'm very happy that you guys are a part of that, and you are connected to that. I don't, uh, that's, I mean, it's easy for someone to say that doesn't have an influence on the way we discuss things, but it doesn't. So, if it does, I don't know what to tell you. Um...

    27. TP

      I'm gonna mention too, just 'cause I don't want people to come out and freak out later, I actually have like 80 shares in Square, which isn't really that much, and you know, but...

    28. JR

      But it's something.

    29. TP

      It is, it is. So I don't want people to think, you know, whatever, you, you're the CEO of Square, I think, right?

    30. JR

      Yep.

  2. 15:0030:00

    Do you wanna, do…

    1. VG

      And I'm open to, to having those discussions. I'm not f- I'm sorry, Tim, familiar with your specific cases, but I'd love to follow up with you and really try and understand.

    2. TP

      Do you wanna, do you wanna pull it up?

    3. JR

      But I think-

    4. VG

      Do you wanna see the tweet?

    5. JR

      ... we, we definitely-

    6. TP

      Uh-

    7. JR

      ... need to pull that up.

    8. TP

      ... so it's, uh, B-I-T.L-Y/antifatweet, all lowercase.

    9. JD

      The, this is also an evolution in prioritization as well. One o- one of the things we've come to recently is we do, we do need to, we do need to prioritize these efforts both in terms of policy, enforcement, um, how we're thinking about evolving them. Um, one of the things that we wanna focus on as number one is physical safety. And this leads you immediately to something like doxing. And right now, the only way we take action on, uh, a doxing case is if it's reported or not. What we wanna move to is to be able to recognize those in real time, at least in the English language, recognize those in real time through our machine learning algorithms and take the action before it has to be reported. So we're focused purely right now on going after, uh, doxing cases with our algorithms so that we can be proactive. That also requires a much more rigorous appeals process to correct us when we're wrong. But we think it's tightly scoped enough, it impacts the most important thing, which is someone's physical safety. Once we learn from that, we can really look at... The, the biggest issue with our system right now is all the burden is placed upon the victim, so we only act based on reports. We, we don't have-

    10. TP

      So-

    11. JD

      ... a lot of enforcement, um, e- especially with, with more of the, more of th- more of the takedowns that are run through machine learning algo- uh, um, deep learning algorithms.

    12. TP

      But if, if something is reported, a human does review it eventually, or are there a series of reports that you never get to?

    13. JD

      There's, there's probably reports we don't. I mean, we, we prioritize a queue based on severity, and the thi- the thing that we'll mark severity is something like physical safety or private information or whatnot. So generally, we try to get through everything, but we have to prioritize that queue even coming in.

    14. TP

      So-

    15. JR

      W-

    16. TP

      ... if, if someone threatened the lives of someone else, you would, would you ban that account? Would you tell them like... like let's say someone tweeted three times, "Kill these people, I want them dead," three times. Is that-

    17. VG

      Yes, that's a violation-

    18. TP

      You didn't-

    19. VG

      ... of our rules.

    20. TP

      ... you didn't ban him though.

    21. VG

      And I, I don't know why that is.

    22. JR

      Well, let's, let's pull that up, Jamie.

    23. TP

      That's, uh ... I, I don't know if, I don't, I don't necessarily wanna give out s- specific usernames because then, uh, people just point the finger at me and say, "I'm getting these people banned." But-

    24. JD

      Yeah.

    25. TP

      ... you know, during Covington, this guy said multiple times to, he wanted his followers to go and kill these kids.

    26. JD

      Yeah.

    27. VG

      And-

    28. JD

      And, and we have to look at that, but we also have to look in the context 'cause we also have, I think we talked about this a little bit in the last podcast, but we, we have gamers on the platform who are saying exactly that to their friends that they're going to meet at the game, in the game tonight. And without the context of that relationship, without the context of the conversation that we're having, we would take the exact same action on, on them incorrectly.

    29. TP

      Yeah, absolutely. That I, that I understand. I think in the case of Covington though, this user was so high profile. He's a verified user, he's got something like 20,000 followers, and it was highlighted by numerous conservative media outlets saying, "Wow, this guy's..." it's screenshotted, it's being shared. I mean, you had a Disney producer in, like saying, a picture of a wood chipper with a body being thrown in it saying that's what he wanted to happen.You know, I- so I- I do know that some of these accounts got locked.

    30. JR

      A Disney producer was doing that?

  3. 30:0045:00

    Well, that- …

    1. TP

      like, "Well, of course. Of course Twitter is going to enforce the social justice aspect of their policy immediately," in my opinion probably because you guys have, uh, PR constraints and you're probably nervous about that. But when someone actually threatens me with a crime and incites their followers to do it, nothing got done.

    2. VG

      Well, that-

    3. TP

      And I'm not the only one who feels that way.

    4. VG

      Well, Tim, that's a mistake. If someone ...... acts in that manner and threatens to hurt you, that's a violation of our rules.

    5. TP

      Right.

    6. VG

      Maybe there was a mistake there and I'm happy to go and correct that, and we can do it offline so we don't fear any sort of reprisal against you, but that's a mistake. That's not an agenda on my part or on the team's part.

    7. JR

      Would this be a manual-

    8. JD

      And we don't, we don't have any PR constraints.

    9. JR

      Would-

    10. JD

      Th- that is not how we-

    11. TP

      So why did you ban Alex Jones?

    12. JD

      We can get into that.

    13. VG

      You wanna get into that?

    14. JR

      Yes.

    15. TP

      Absolutely.

    16. VG

      Are you ready for Alex Jones?

    17. JR

      Sure. (laughs)

    18. TP

      (laughs)

    19. VG

      All right.

    20. JR

      Oh, I've been ready for Alex Jones.

    21. VG

      Let me pull this up.

    22. JR

      (laughs)

    23. VG

      Um-

    24. TP

      Well, le- let, lemme say this, the reason I bring him up-

    25. JR

      Okay.

    26. TP

      ... is that Oliver Darcy, one of the lead reporters covering Alex Jones and his content, said on CNN that it was only after media pressure did these social networks take action. So that's why I bring him up specifically, 'cause it, it, it sort of implies you were under PR constraints to get rid of him.

    27. VG

      I think if you look at the PR that Twitter went through in that incident, it wouldn't be that we looked good in it.

    28. JR

      You, you have to-

    29. VG

      And that's not at all why we took action on this. (laughs)

    30. JD

      Sorry, but-

  4. 45:001:00:00

    Well- …

    1. VG

      to express that opinion.

    2. TP

      Well-

    3. VG

      If he's doing it in a manner that's targeted at an individual-

    4. TP

      Mm-hmm.

    5. JR

      Repeatedly.

    6. VG

      ... repeatedly-

    7. JR

      Repeatedly.

    8. VG

      ... and saying that-

    9. JR

      Okay, but what about-

    10. TP

      So-

    11. JR

      What about sp- like-

    12. VG

      ... that's where the intent and the behavior comes in.

    13. JR

      You know what's going on with Martina Navratilova right now? Martina Navatrilova. Why, why can't I say her last name?

    14. TP

      Nava- Yeah, I don't know. (laughs)

    15. JR

      Navatril- t- I don't think I've ever said it. Martina Navatil- tilova. Is it Tilova, Trelova?

    16. VG

      That doesn't sound right.

    17. JR

      Anyway, epic world-class l- legend tennis player, right? Who happens to be a lesbian. Is, um, being harassed because she says that she doesn't believe that trans women, meaning someone who is biologically male who transitions to a female, should be able to compete in sports against biological females. This is something that I agree with. This is something I have personally experienced a tremendous amount of harassment because I stood up when there was a woman who was a trans woman who was fighting biological females in mixed martial arts fights and destroying these women. And I was saying, "Well, you, you just watch this and tell me this doesn't look crazy to you." Um-

    18. TP

      Well, I-

    19. JR

      Go ahead.

    20. TP

      Go, go ahead.

    21. JR

      Well, my point is, you should be able to express yourself. And if you say that you believe someone is biologically male even though they identify as a female, that's a perspective that should be valid. I mean, that, this is someone's, someone's... This is a... First of all, it's biologically correct, so we have a problem in that if your standards and your policies are not biologically accurate, then you're dealing with an ideological ide- you know, an ideological policy. And just because, I mean, I don't, I don't wanna target trans people. I don't wanna harass them. I cert- I, I'll call anybody whatever they want. I mean, if you wanna change your name to a woman's name and identify as a woman, I am 100% cool with that. But by saying I don't think that you should be able to compete as a woman, this opens me up for harassment. And I never-

    22. TP

      Will-

    23. JR

      ... reported any of it, I just don't pay attention to it. But, but go ahead.

    24. TP

      But in, in going into like Meghan Murphy, for instance, right? You can call that targeted harassment if Meghan Murphy, who is, uh, for those that are don't, don't know, she's a-

    25. JR

      Mm-hmm.

    26. TP

      ... radical feminist who refuses to, uh, use the transgender pronouns. If she's in an argument with a trans person over whether or not they should be allowed in sports or in biologically female spaces and she refuses to use their pronoun because of her ideology, you'll ban them.

    27. VG

      Again, it depends on the context on the platform, and it's also, uh, wa- not banned permanently. Like you get warnings.

    28. JR

      Well, she was banned permanently, but let's be clear-

    29. TP

      Like she was warned, yeah.

    30. JR

      ... about what happened.

  5. 1:00:001:07:04

    Yes. …

    1. JR

      miss context? I mean, it seems to me that there's a lot of-

    2. JD

      Yes.

    3. JR

      ... people that say things in humor-

    4. JD

      (clears throat)

    5. JR

      ... you know, they, they-

    6. JD

      Or, or, or slurs within particular communities which is perfectly reasonable in our case.

    7. JR

      Right, right. So-

    8. JD

      So yes, there is a danger of the algorithms missing context and that's why we, we really want to go carefully into this and this is why we've scoped it down first and foremost to doxing, which is at least... First, it hits our number one goal of protecting physical safety, like making sure that nothing done online will impact someone's physical safety on- offline on our platform in, in this case. The second is that there are patterns around doxing that are much easier to, uh, see without having the context. There are, there are, um, exceptions-

    9. JR

      But-

    10. JD

      ... of course 'cause you could dox, um, someone's public, you know, uh, a representative's public, uh, office phone number and email address and the algorithm might catch that, not have the context that this is a US representative and this information is already public.

    11. JR

      So essentially this just, it highlights how insanely difficult it is to monitor all of these posts. And then what, what is the volume? Like what, what are we dealing with? Like how many posts do you guys get a day?

    12. VG

      Uh, hundreds of millions of posts a day.

    13. JR

      And how many human beings are manually reviewing any of these things?

    14. VG

      I don't have that, that number. A lot.

    15. JR

      A lot. Thousands? Hundreds of thousands? How many employees do you guys have?

    16. VG

      We have, uh, 4,000 employees around the world.

    17. JR

      That's it?

    18. VG

      Yeah.

    19. JR

      (laughs)

    20. JD

      We have, we have, we have 4,000 employees. The, the reason-

    21. JR

      That's crazy though, but stop and think about that. 4,000 people that are monitoring hundreds of millions-

    22. JD

      No, no, no, no, no.

    23. JR

      ... of tweets?

    24. JD

      No, no, no. We, we have a, we have a, we have a small team who's monitoring, uh, tweets and some of them are employed by us, some of them are contractors throughout, throughout the world.

    25. JR

      So 4,000 employees total.

    26. JD

      4,000 employees who are engineers, who are designers-

    27. JR

      Okay.

    28. JD

      ... who are lawyers, policy experts.

    29. JR

      So the number of people m- actually monitoring tweets is probably less than 1,000?

    30. JD

      Uh, well, the reason we don't give out specific numbers-

Episode duration: 3:25:10

Install uListen for AI-powered chat & search across the full episode — Get Full Transcript

Transcript of episode DZCBRHOg3PQ

Get more out of YouTube videos.

High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.

Add to Chrome