🎂 Raffi Krikorian & Alissa Starzak Fireside Chat

Hi, this is Alissa Starzak. Welcome back to everyone at Cloudflare TV. I am the head of public policy at Cloudflare and I'm here with Raffi Krikorian, who was the former CTO of the Democratic National Committee and currently managing director of engineering at Emerson Collective and I am so happy to have you here today. Thanks for having me. So I want to start back. So it's Cloudflare's birthday week. We've been around for 10 years. So I want to do the birthday thing and basically ask you how you started out, where you started out 10 years ago, and then we can figure out how you got to where you are now. Oof, 10 years ago. So 10 years ago, I was a year in at Twitter, I guess. So like I was one of the, I was like the 50th employee, give or take, at Twitter and I joined in 2009. So in 2010, I was just getting up to speed again. Well, so how did you end up at the DNC? So you joined the DNC in 2016, right? Yeah, I mean, so I was running the Uber self-driving car team at the time based in Pittsburgh, and I distinctly remember I was in San Francisco visiting HQ on Inauguration Day 2017. And sort of just like aghast, kind of, sorry to get political, I was a little aghast with what I was watching on the screen. And literally that hour, I started pinging all my friends who had joined the Obama administration, asking, and like, some of these people have asked me to join with them, but I didn't want to move to DC at the time. And it kind of like, felt like I missed out on that opportunity to join a historic administration. So I reached out to all of them like, where do I go? What do I do now? And they introduced me a bunch of people that are very generous about it. And over the next month or so, I finally had a conversation with the chair of the Democratic Party, Tom Perez, chair of the DNC. And we talked through what a role could be we talked through what he needed, he talked through what I thought about, and then we settled on me coming on board at the DNC. Great. So what was that like? So where did you start with? What did you do? Just give us a description, obviously, the DNC in 2017, after all the events in 2016, probably a lot to do as a CTO. Yeah, I mean, like, I, you know, the chairman Perez and I, I remember our first conversation, kind of looking at me like an alien of just like, we just didn't speak the same language. Like, I'm sitting here, a techie, he's a very seasoned political policy, government operative, he used to be a Secretary of Labor, like, I'm sitting here, and he's like, well, I need someone to clean up my cyber problem. And I'm like, I think we should talk more than just a cyber problem. Yeah, you're right. I mean, like, I was, I was essentially brought in to help clean up after the 2016 Russian hack of the DNC. And there was a lot to do that. There's things like, we still use email on frame. Like, there just wasn't good information security protocols put around, we have introduced two factor mess, introduced two factor, we introduced end to end secure messaging via signal, basically had to like reboot the entire culture on how we think about technology. And that's only on the defensive side. I like this, as I used to like to say to the chairman, like, fixing our security problem would just ensure we didn't lose again, it wouldn't ensure that we won. So I also spent a lot of time digging in and trying to understand the technology foundations of the party, things like, how do we model voters? Like, what's the data warehouse look like? How do we run analytics at scale? Where are all the voter outreach and voter contact mechanisms? And we slowly like chipped away and tried to like, make sure all those were shored up, and like really understood what political campaigns needed for a lot of time with people in the 2016 campaign, and just tried to build the most robust system we could to get us ready to move fast in 2018. And then again, 2020. So do you feel like it's in a better place now? I mean, do you feel like how much how much of the work was backward looking in the sense that it needed to be done, and it was sort of restorative? And how much of it is actually was it affirmative, where you could actually finally move forward in a in a in a future looking way? Yeah, I mean, that's a great question. I mean, a lot of the work was backwards looking. So like, you know, our data warehouse in 2016, where every single campaign data scientist would do work effectively, would go down for hours on end inside the 2016 campaign, like, one of my one of my, one of my data managers was the data director for North Carolina in 2016 cycle. And she would talk about how like, well, the data warehouse went down, I'll get that I'll just go take a nap in my car, because it's gonna be a couple hours before it cuts back online again. So we got that to a place that that's just not a thing this cycle, like this cycle, analytics at scale is is working flawlessly. And I would like to say that we spent a lot of time getting like the new and sexy voter contact technology in place. But honestly, the pandemic has just upended all of that, like, no one predicted, this is the type of campaign we would have to run. So it's kind of like reverted most, most political folks in the campaign folks got like, first principles and trying to understand what we need to do in this world. Yeah. So how do you think it has shifted things? So here we are in this in the pandemic, everybody's virtual, in some ways, it's more important than ever. How do you think about it? How do you how do you reformulate if you're if you were in that same role now, which you're not, you're not, you're not in the same role exactly now. But if you were in the same role now, how would you describe what had to be done? How does it look different? Yeah, I mean, Democrats are really, really good at knocking on doors. And they're really good at sort of like getting out there and canvassing neighborhoods and talking to people. And so like a lot of what we spent time on was like, how do we make that even better? How do we make that even more efficient? How do we get it so that when you knock on someone's door, you can start a really good conversation with them and get some data from them that we can aggregate and have a good sense voter contact models and how how the campaign is going and how the candidate is going. And we, we spend a bunch of time also rebooting our digital advertising work and trying to figure out what would be appropriate micro targeting things like that. But had we known this was going to happen, we just would have like not spent any time on door knocking. I mean, like, it's only this week that the Democrats start door knocking again. So like, we wouldn't have spent any time with that. It's been a lot of time instead, like, rethinking campaigns to look like a digital marketing, digital marketing agency, like, how would I get Google manage a marketing campaign, or we'd be we would build that tech and brought that brought that to the campaign instead. But, you know, do you think anyone's doing it? Well, I mean, so here's this new brand new world, in some ways, to your point that no one anticipated. Do you think you've seen anyone able to pivot effectively? I mean, what does it look like to you from the from now from the outside? I mean, no. In reality, that might be fine. Like if everyone hasn't pivoted correctly, then everyone's in the right place. But no, I mean, we haven't figured it out. And part of the problem is that like, this happened, like at exactly the wrong time. I mean, you can't plan for these types of situations. So like, you know, a lot of campaign technologies locked and loaded sometime in the summer, and then you sort of just like operate it and run it at scale as your race for the election. But like, we locked this country down, like in March. So there just wasn't enough time to get like a whole new train of thought, like, all the online voter registration work has clearly been soaring, because you can't, you can't get that up there in person. But we had a lot of plans are like, at large, the progressive movement had a lot of plans of how to do in person at scale voter registration that just got wiped out. Right, right. Well, so now that it has shifted online, I guess, I guess that raises a whole set of questions about are there are there new opportunities in the online space? Are there new opportunities to engage given that we are in a pandemic and things are strange, right? What can people do online? So I think one of the challenges even in the both on the on the political side, but also just thinking about how we interact, it's shifting, right? All that it's constantly shifting because of the pandemic. So how do you think about that from a political campaign standpoint? How do you engage people make them actually feel like they're a part of something? For example, what does that look like? Yeah, I mean, like, in a lot of ways, 2020 was going to be probably the year of text messaging and like mobile device messaging, and trying to make sure that we sort of reach voters on the device on the device they use most where they are on the stuff that they're most interested in. So in some ways, that conveniently still holds true here in campaign times, like, we can still text message a whole bunch of people and keep them engaged. I don't think anyone has really figured out how to get the same excitement of bringing people together to see a candidate to talk to a that over zoom or other technologies. Like, I mean, I really like seeing the list of the same across from you at a table. So I don't think anyone's like figure out that part. But I will say all preliminary data shows like record enthusiasm, at least on the left hand side, pretty sure on the right as well for this for this cycle, like, people are figuring it out, they're stumbling, they haven't figured out exactly the right tech to use, but they're cobbling together what they can, they're figuring it out. Well, so what about the election generally? So here we are, we have we have lots of stuff happening on the security side, we have questions about the post office, we have, you know, we have this huge range of things. Do you think we're up to it? I mean, from a from a country standpoint, where do you think we are actually on the just election security side, as someone who's in tech? I mean, I'm very involved in this space. Yeah, I'm generally always an optimistic person. So I think that like, yeah, I think, again, we'll, we'll, we'll figure this one out. But there are a lot of open concerns and open gaps. So, you know, everything from the post office itself, and operations, or logistical operations of actually moving the number of ballots we need to move, actually to the act of counting up the tallying, you know, I'm very concerned about the different, you know, Secretary of State websites and all the county websites, which are going to post results, concerned about those security, and whether or not they can stay up, or whether they're gonna be tampered, how do you know if it's true or not? I'm also worried about just the information environment that most voters are walking into, like, look, we're, this is going to be the record breaking election when it comes to the number of mail in or, or drop off ballots, that's going to happen, like, and in order to pull that off, that's a voter education game at this point, like, you know, Pennsylvania has some really crazy rules and intricate rules of how you get everything just right, in order for your ballot to count if you're going to do it by mail or by Dropbox, North Carolina, Michigan, like all these states have such complex rules. If you're a first time voter, or, or if you're an older voter that might get easily confused by the number of rules that are going on, it's gonna be so easy for you to not vote correctly and have your vote not count toward an election. So I think like we're walking into a landscape where people aren't going to get all the information they need, and we need to still figure out how to break through and get them the information you need. And then when and if it doesn't go well, we on the political side are going to need to figure out every single person who's ballot didn't count, we're going to have to chase them, we're going to need to get them back, we're going to need to get them to actually cast their vote correctly this time, figure out what went wrong. There's gonna be a lot of logistical stuff that we're gonna have to work through electronically and physically if you're race for election. Right, right. Well, so I mean, so thinking ahead, so if we, let's anticipate we were five years and so let's, let's say, for the sake of argument that this election actually works, we actually manage, you know, what do we improve next time? So it's almost too late in some ways. So we can do a lot of sort of short term things for in anticipation of this election. But at the same time, we can do a lot of short term things. So we can do a lot of short term things. We can do a lot of things that we've seen pop up in this election where we've seen pop up in a pandemic. So what are those? So you mentioned the Secretary of State, we've been thinking a lot about how we how we can provide security services to entities that might not have support for those and for from our Athenian project. But what about the bigger things? What about long term? You know, where do we go? How do we how do we think about this from a policy standpoint down the road? Yeah, I mean, like, I can try to put in a few buckets. I mean, there's like, the not sexy bread and butter stuff that we need to get right next time, which is like, you know, a bunch of states have been voting by mail for a decade, if not longer. And we need to get that experience from those states to every other state as quickly and as fast as we can. Like, we just need to be not a question of whether or not the state of Pennsylvania and sort of keep on picking Pennsylvania, but it could be any state, not a question whether or not they can administer an election at that state ever again. And whether that means test runs, or whether that means like all the other times we need to go, we need to give this a shot and need to get it right. Doesn't matter. We just need to figure that out. But then let's talk about the information landscape that we're walking into, like, voters can get so trivially confused right now in this cycle, both accidentally and potentially maliciously, there's a lot of information out there. How do I know what's true? And I'm actively googling or actively searching for like, how do I cast my ballot? Where's the nearest ballot box, stuff like that. What's the voter registration date? I mean, there's been some evidence. So talking about malicious, there's been some evidence of bad actors trying to purposely confuse people and purposely put dates out there. So from a policy standpoint, we need to figure out how to stop that. Like that could be policy, either at like the social platform level, or the quote, unquote, news level of trying to try and figure out how to arbitrate what's true and what can actually go out, whether it be from official channels, organic channels, or stuff like that. But it could also be policy work that could be handled by the government side of trying to hold companies better accountable for the kind of content that might be going out. I'm not saying that like, companies should, I'm not saying that like, companies should be held accountable to every single menu item and whether or not they're neutral arbiters and things like that. But there are some clear official pieces of information that should not be tampered with on its way out the door and on its way for voters. But there's an opportunity there to maybe fix or do something. So actually, that's an interesting point. In part, one of the things that we're seeing driving it is is the fact that we have lots of social media. And so people are trying to promote things on social media. But of course, you can promote information that's not true, too. So how do we think about what's official and how we sort of keep that separate from something that might not be as official, particularly someone who's coming from a party side. So in some ways, you're putting out information all the time. How do we keep those lines clean? How do we make sure that people are getting the information that they should be getting, where it's not necessarily just a company? I mean, how do you think about it from a government side or a policy side? Not someone just deciding what's true or not. Totally. I mean, well, if it were an easy answer, we would have done it. But I mean, I think there's opportunities here to figure out what are official accounts, what are unofficial accounts. And then either changing boosting algorithms or changing prominence of those accounts over other organic pieces of propaganda showing up. Then there's another aspect of potentially doing better fact checking, frankly. And it's something that most social media networks seem to have abdicated at this point. And I'm not saying to suppress speech, but instead, perhaps, to label it better, like what's actually true, what's not true, and then providing resources so that people can read a bit further and understand a bit further. There's also opportunities, like I mentioned, like boosting algorithms. There's also opportunities to deweight things that are of more questionable value versus those which come from official statements. So I think there are definitely a few things that social media companies could be doing. The question, really, is how do we incentivize them correctly in order to actually do that work? And that's potentially where government intervention or government regulation could come into play better. Because right now, most of them are just incentivized by pure stock price. And that doesn't seem to be working, frankly. So there needs to be another incentive injected somewhere in there. So given all of that, what do you think of all the decisions to stop doing political advertising right around the election, for example? How do you think about that from a practical standpoint? How does it affect the campaigns? What does it look like? Will it help people? What's your sense? I think it's well-intentioned. I want to say, again, I'm an optimistic person. I want to say it's a well-intentioned move, but I think it has pretty bad downstream ramifications. So for example, again, I apologize. It sounds a little more political than not. But the most highly engaged content on Facebook right now is usually more conservative-leaning content. There's been a bunch of reports about this. And so if you simply eliminate campaign advertising, that doesn't provide the left an ability to sort of counteract or provide a different point of view at the same scale and same reach as that kind of commentary. So I think if you eliminate, and what the plan is is to eliminate all election-related paid advertising the week before Election Day, that just means that it's all going to go into organic, and we're not regulating, we're not fact-checking anything on the organic sense, which will fundamentally put liberals and left-leaning candidates at a disadvantage in the situation. So I think there's those kind of unexpected consequences that will show up by simply doing like point fixes as opposed to like a holistic think-through of what needs to be done. So to what extent, I mean, on all of these issues, it seems like in some ways we're playing catch-up, right? We didn't think about what was going to happen with the possible amplification or disinformation. Do you think it's possible to catch up? I mean, are we at a point where there is a way forward either through regulation or elsewhere? I mean, what's your sense of what comes next? I mean, I think it is possible. I think it's not in the next 30-something days, but I think in the next four years. But I think like it's a, we have to put a concerted effort to it. Like we're basically, you need to ask an entire industry, and I say ask in the most light way, it might not be asked, but we need to ask an entire industry to change how they do business. Like right now, and I think there's like some metaphors we can draw around here. So maybe you can draw a metaphor of social media content to, this may sound more pejorative than it means, but to pollution effectively. Like we figured out how to regulate companies that pollute the environment and are causing climate change. We're not perfect at it yet. We're making steps in every day. But the answer hasn't been you have to stop polluting tomorrow. The answer is like setting up different incentive structures, whether it be cap and trade or carbon taxes or things like that. And I just have to wonder whether or not there's something similar we could be doing on social media and social content related issues. Like we're not asking you to go full on regulate and shut it down. We feel that would cause a backlash fairly immediately, but we need to just introduce points of friction. We need to introduce some kind of check and balance into the system so that these companies are accountable to more than just their stockpiles. Well, so here I am on the public policy side. So now I'm very curious about that because we have Congress who is a little behind the eight ball on some of these issues sometimes. So how do we get that to work? So how do we get to a point where if we do believe that there's some other structure that needs to be in place where people can look forward enough to figure out what that looks like and understand the technology and understand the underlying issues that they're actually trying to address while still being flexible so that we can continue to have innovation? Well, first off, if you believe in a system of checks and balances, then what you need is an appropriately staffed check and balance. Right now, I would say we don't have the right kinds of people with the right frames of thought sitting on the governmental side to even have this conversation. So instead what we have is the government complaining effectively and tech basically saying we're going to self -regulate ourselves. Again, I wish that could work. I just don't see a path forward to making that happen. So what we need to do is we need to recruit actually really good tech talent, technology thinkers into government. We need to recruit them through programs like Tech Congress, maybe, which places technology staffers into different congressional committees or places them into different congressional offices to sort of be a staffer there. And so we need more people who are educating lawmakers, educating policymakers so that they can actually have reasonable conversations with the tech side. I mean, look, every single time tech executives go in front of a senator house committee, it doesn't look good. Like the kinds of questions aren't great. They're like, they might be deep in hope, but it's not like a generative conversation. I'm not expecting a general conversation on day one, but I feel like there's like an intellectual mismatch on this topic. And so like, we need to sort of like equal the playing field so that we can have that type of conversation. And I think it really only starts with getting the right people into Congress first to have those longer term conversations. So we either need to elect them there or we need to get them as staffers is what it really comes down to. So I'm going to actually switch topics a little bit. I think you're at Emerson Collective now, right? So, and you're doing all sorts of interesting things. So what are you working on in your current role? How do you think about social good in the role that you're in now? Yeah. I mean, we think, you know, my team at Emerson thinks a lot about how do we use technology to either amplify the works of Emerson and our partners or to raise the tide in certain fields, maybe fill a gap in places that incentives don't align. You know, one of the things that we've been working on is there doesn't actually turn out to be a map. Let me back up for a second. Like when we entered the pandemic, one of the concerns that we were all had was what would economic fallout do to children and specifically what would that mean in like food, potentially newly food insecure population. So if all these people are losing their jobs, how are they going to get food and how are they going to keep their families fed? And one of the things we've found out, like we started poking around the space and trying to understand what does, you know, crassly what the supply and demand look like, like where are people who need the food? Where is the food? We really started understanding the network of food banks and food pantries and all the different players in the space that are working so hard to make sure that people are fed. But when it comes to like actually a last mile problem, there seemed to be a low information gap. Like there seemed to be places, especially with this new population of food insecure folks of like, they don't know where to turn to to get food. Like they've never had to do it before. And so when you poke at that a bit more, it turns out that there isn't even a map of all food pantries. There isn't a map of all the different organizations that could help you in your time of need. Like we talked to the New York City Food Czar, we talked to a prevalence in New Orleans, D.C., Chicago, and there isn't a comprehensive list on a state by, on a city by city level, regardless of the nationwide level. So we've actually been creating a tool that helps volunteers actually call food pantries across the U .S. and sort of get an inventory, times that they're open, whether they have kosher food, do they have formula for mothers, like that kind of stuff, so that we can then appropriately get that information and match people to places that are nearby that can help them in times of need. So tech actually helping the world in the current place that we're in, which is exactly where we want to be, right? I mean, look, I don't believe that we can ever use, like, I don't believe that in this, in the sector that we operate in, that tech is, should ever be a disrupting mechanism. I feel like there's some, like, really clear gaps that if we can just, like, connect person from point A to point B faster than they would have done before, like that's super power, right? Like we can make things happen faster. We can make it happen faster than they would have done that before or easier. That's what we should do. Like, there are so many kids and so many people that are now hungry because of the pandemic. Right. Well, and the reality is, I think, where we are now, these are things that we can build, right? I mean, we're in a place, I mean, that's, that is the benefit of being in the year 2020, which 2020 has been kind of a crazy year. So, but there's so much that we can do now that we might not have been able to do five or even 10 years ago. So, we have a little under, we have about four minutes left, but I want to, I want to actually touch in a forward-looking way. So, here we are thinking about the future of the Internet too, right? So, social good programs like that, which are really kind of building to need, where are we in 10 years from now? Do we have more of that or do we have more, do we identify more problems that we've created than ones that we've solved? No, I mean, I don't know who is to answer. I mean, if we look back 10 years. Optimism. Optimism. I'm looking for some optimism. I mean, when I, like going back 10 years, like the way we thought about social media and social media networking was that like it flattened the world. It gave you access to everything. Like everything became equal. Like there's a very utopian vision, but in some ways, a super naive one, because it's sort of based on the same principle that created the Internet, which is like this naive, like, be lenient in what you accept, be strict when you send out on the protocol level, since that propagated all the way up to our like society somehow. But like, I feel for the next 10 years, like we're kind of on this knife edge where like, we can actively choose right now. It's like similar to climate. We can actively choose right now that we as a society are going to do a lot of work to try to get the right checks and balances in place and to build an Internet society that, you know, I'm not going to say that mimics the real world, but has some more real world consequences and dynamics as part of it. Like part of the reason why we as like people who wander the streets and talk to each other in the real world, like keep a society together is that we have a social contract that's hard to break when we look at each other in the eye. And we've like dissolved that when it comes to Internet. And this is the chaos that we get into. So we need to like actively choose now in order to build that better society. Or we could not. And then it becomes like further more of this dystopian nightmare that we live in that could get even worse and further factionizing the country and further effect like, that seems bad. I don't want that. I want the other one. But I think it's going to require a lot of work for us to get there. We have to choose as a society to do it. Well, and it then gets into the question of how we choose, right? So we have the question of government and private sector and how those pieces fit together, which there may be a lot of opportunity over the next 10 years, but it's hard to know exactly what happens, right? So, so on that sort of just last last words of with, you know, what's your sense of on the, I won't ask you election results, because that's mean. But what's your sense of like, one big thing that could come out in the next 10 years, that's going to that's going to change things that's going to that's going to push us make sure we're on the right side of the knife edge, if that's something or you just stay on the knife edge. I'm not sure which actually where you want to be on the knife in that. Yeah. I mean, it's a good question. I mean, I think there's so many opportunities, whether it be like how to educate our kids better, or there's so many opportunities on how to like, how to like, and like the future of work and how we can all be working from wherever we want. It's still contributing to society in a, in a global kind of way. I mean, it's actually, it's just like exact conversation I have with like, some of my friends in Armenia, like, you can participate in the global economy, even though you're in this landlocked country in the Caucasus. Yeah. So I think there's like, I think that I think like, I think we still I still lean on the fact that you can instantly communicate across the world without any boundaries. It's an amazing and powerful thing. We just need to figure out how to like, make us all still accountable in the same way that we're accountable in real life. And build those social structures and their social contracts, right? That's exactly right. Well, Rafi, I'm so glad that you joined me today. I won't make you sing happy birthday because that would be mean. That's not what we're going to require anyone to do. But thank you. Thank you for joining us on for this week. And, and thank you for all the things that you've done and are planning on doing here on out. No, thank you. It's been an absolute pleasure.