Talk & transcript
The Science of Presidential Elections
David Shor
Transcript
Download textAutomatically generated transcript. May contain errors. Select a timestamp to watch that moment on YouTube.
I mean everyone come forward come forward come forward take talk to the person next must speak very quietly until the people in the back feel awkward quiet quiet thank you can you guys hear sh if you guys if you guys want to hear the talk yes um come come forward oh my God we're in The Matrix um that's cool um okay so I'll plug in my computer in a
second how many [Music] people how many people is this your first time here oh my gosh wow so many people um how many people is this your second time here that's a lot how many is it your third time we have any three timers okay Lots I can't really see anything so um fourth probably a few fourths okay some fourths that's cool um how many of you have been to Sweet Mouse this is my first time here it's cool
it's like we're in The Matrix this is really cool sour sweet Mouse sugar Mouse sugar Mouse sugar Mouse my bad don't okay we are going to get started here and um so if you don't follow us uh follow us on Instagram and Twitter uh take out your phone take look take a picture follow us tag us if you tag us on Instagram or Twitter uh that's helpful for us we we'll remind you later um okay so I'm going to talk a little bit about what is um decentralized science so uh how many people here come from like a science background oh a lot and what medicine
counts you can raise your hand um no need for that um and then how many people like come from how many people hear because of SBC like Stanford and blockchain oh some okay a lot okay so decentralized science is like a combination of uh science and blockchain so what does that mean so de decentralized uh like defi was like using the blockchain for financial as a financial primitive to sort of remove some of the intermediaries from uh the from people from like the middlemen that was slowing down the financial system and decentralized science is sort of using that same idea where guys in the
back um the decentralized science is like the same idea where instead of like trying to remove um instead of using the blockchain for Financial incentives uh you can use the blockchain to simplify science so imagine a world in which like the the like the internet was a giant lab and you didn't have to work in like a university setting in order to do science and like you could do science on the blockchain there was no one in charge the best ideas came to the top um it was open source it was collaborative and so this is what decentralized science is instead of doing science in in a university setting you're doing science in like an internet native way
and there's a lot of really cool benefits to this that we'll talk about and what's what's interesting is I don't think that this is too far from the future like we're just on the cusp right now of like a big transformation where uh universities are sort of losing some Prestige and you know Sovereign and individuals are becoming more and more important so I'm going to talk around like five different things that's distinguish like traditional science which we're going to call Tradition trai from descent calized science so trai it relies on the first one is funding so trai relies on like government grants to make it work and the problem with this is that government grants oftentimes just fund the same projects over and over again and that's that's okay but it
doesn't allow for new moonshot style ideas whereas in decentralized science it's more of like a um a meritocracy where people can uh fund things that matter to them and instead of it just being the same things funded over and over again uh new ideas can come and sort of uh more more aggressive ideas um the second one I want to talk about the differentiates traditional science from uh decentralized science is publishing so existing uh publishing is sort of relies on peer review and everyone thinks of peerreview as the gold standard in which there's nothing better than it but if you actually have been on a peerreview committee or have you if you've submitted a paper for peerreview you know that it's very Pro that it's not actually that great so the people
who review the project don't actually care a lot about it and because they're not paid and they're sort of just trying to get it done as quickly as possible they have no skin in the game um so and often times no one like looks carefully at the data it's just they just sort of know who's submitted what and uh approves their friends in decentralized science it's more like a prediction market for academic papers where the people could bet on papers that they think are going to be very successful and if they're correct that that research is very important they will get rewarded with money or tokens um data sharing so in order to participate in in order to In traditional science the sort of the end result is often times a PDF paper posted
on archive now that's okay but if anyone ever wants to sort of reproduce that experiment it's very hard to do that because either the methodology is so complex or um uh you know it's just they can't get access to the data in decentralized science all the data is posted on chain so you can get access to that immediately um this is a big one I think so one of the things that decentralized Finance did that traditional that sort of was really important compared to traditional Finance was that it opened the financial system up to thousands and thousands of more people and I think there's something similar that could happen in science where there should be like a thousand times more scientists than
there are today and the reason there aren't is because um in traditional science you need to apply to a university you need to get accepted you need to find a pi you need to get funded and all of these things are steps that step that are in the way of like people getting access to an institution it's artificially supresses the number of sort of scientists that are out there sort of similar to like the AMA how there's like that automatically uh sort of suppresses the number of doctors that are there so I think with like decentralized science you're going to try and open it up and make it permissionless so there could be a thousand times more scientists um another one is like IP or I would also call it like um putting
together the incentives so in a research lab you sort of your incentive is to get a paper published and then to get that to you get use that to get an academic role the problem with that though is like it would be much better to just tie the work to the actual underlying like uh tie the status that you get from the work to the underlying work instead of having it go through all these intermediaries so in a traditional University you have to you if you create some sort of Ip you have to uh put that into the licensing office you have to get access to it and then you don't by the time you're done with it you actually don't get you don't get any sort of tangible benefit from it with the centralized science you know every research project is a is a is an IP is a
pool of Ip that people can contribute to and then earn based on whe whatever they contribute um governance is defi allowed for new types of governance to happen so any Dow that exists has like liquid democracy or futy um which are all very new styles of deciding which researches which experiment should be conducted um how resources should be allocated and in a traditional setting it's just sort of the pi who might already have been a professor for 40 years and doesn't really care they're calling the shots so that governance model is sort of broken and with the centralized science it's a much more liquid active and dynamic way of doing it um reproducibility is another one um so like in a traditional
science setting uh you know a lot of universities are coming under Fire because they their resarch is either fake or he's not easy to reproduce but when you put all of the data for a research project on chain um it's really easy to reproduce it because every piece of data has um some uh has provenance associated with it so you can just click a button and an AI agent can go over it again so and then the last one is like um AI so I don't think there's going to be any like scientists human scientists in the future like all science is going to be done by AI bods because they're faster and just like have much more knowledge so oop is this going
forward okay whatever um so a like all so you need to start putting data on chain where these AI Bots can automatically just get access to it so I think the future of dii is really really bright there's um I think the future of dii is really really bright um we're going to democratize science make it permissionless um there's going to be a thousand times more uh people and and scientists and Bots involved and uh it's going to be much much more reproducible okay so um now I will talk a little bit about decentralized science New York City
which is why you're which is presumably why you're here um here are some of the past speakers we've had uh people doing mushrooms psychedelics bees AI Martin scy the guy who invented nfts the research Hub guy um PE we we're really into longevity um we had a professor from uh Albert Einstein come talk about tissue replacement we had the guy who did the enhanced games which is basically like the Olympics on steroids um we had someone who's doing focused ultrasound um for the brain we had someone who's doing bio electricity um when we had a guy who is using a
camera in your toilet to measure your poop so we go for the weird speakers for sure okay so those are the speakers um we got a lot of people out here I was this I started this group just because I wanted to create a space for people interested in science in New York City I thought there was a lot of people playing like back gamon and stuff but not a lot of people discussing interesting ideas so I thought this would be like a good thing to do um we have a website where we put all our our videos and all our recordings so if you go to a dside .nyc we have all our videos the video will be posted tomorrow morning and we have a lot of cool upcoming events which I'll share
with you right now I like okay so this is a big one uh we got Brian Johnson and he's going to do you guys know who that is yeah okay so August 19th um if you can scan that QR code in the bottom uh I think maybe if we do that maybe it'll be a little bit easier so so that'll be really cool this is the guy that basically like uh is at the Forefront of longevity and trying to live forever or trying to not die not trying to not die um he's also a really cool guy um I think he's pushing I think he's I think he's making it making it
seem possible that we might be able to live forever and he's been thinking about he's not just like a a Johnny come lately he's been thinking about the is for a really really really long time I was reading some of his stuff from 2018 and it was actually really cool [Music] um next up we have um Joe Zer uh who will be coming on September 19th uh she was one of the original people that was um interested in biohacking and crisper um very very cool person she's also been doing this for a very long time uh like an OG of crypto uh and she'll be here on September [Music] 19th okay um next we have a nor sadiki
uh classmate um she's doing a very cool thing on embryology and trying to reduce the amount of uh birth defects by looking at the embryo after it's been fertilized and screening it for genetic diseases it's a very cool project and it's working and it's live so that'll be a very that'll be a great project a great to and she's very cool um next we have Sheridan he's going to be talking about penile optimizations yeah okay he's been researching this topic and he his whole premise was that he he came up he's like I think we need to do a talk on this uh he's a teal fellow a Founder um very cool guy and
he's been studying this you think there's a lot of like uh bad research out there and uh he'll be he'll be talking about what he's found uh that's on November 11th Okay so so those are the talks okay next we have a project that we've been working on um where's Brian Brian Brian you want to come up here yeah come up here come up here yay Brian thank you Brian for uh Brian is the technical Brian's technical door but but actually he's behind the scenes that actually like builds everything that
we're doing here so kind he's the he's the tech wizard um that builds like the project that we're both working on it's called science validator Network which I'll talk about in a second and um so what we've been working on today is a way to like you know decentralizing science happens a lot onchain but it can also happen a lot offchain and we wanted to think about ways in which um we could use this amazing community that we've built today um to like deepen it a little bit so we decided to launch like a crypto token for D NYC that we built um so what does it do um we're trying to make it so that um basically what will
happen is like speakers for the next one will be voted on based on uh like a community vote instead of just like one per one or two people deciding who to invite so we're trying to like decentralize it a little bit um you can vote on the next speaker you'll be able if like the Brian Johnson event going to fill up we already have like 500 people RPD um so like we're going to have to cap so we're going to try and limit it to people who've been here before um we have a community group chat so we're going to token gate that based on people that have been to an event at least once so there's some common ground in the community um and maybe we'll we have this cool shirt we'll give away um we if you guys bring a sponsor in at some point um we will uh we will uh split the
money with in the Dow so we're thinking of a lot of cool ways to actually like make um let make this community into like a decentralized onchain community y Brian um okay um so if you want to get the tokens for the D if you want to get D NYC tokens uh you can scan this we'll also send an email in a little bit so that you can claim your D NYC tokens I'll leave it there oh also I'll take this opportunity if anyone wants to sponsor these events um feel free to uh
send me a message I think this is at this next page oh we're going to have a leaderboard for D NYC as well which is going to be cool so you can see who are like the top dite nycers oh go back sorry
sorry did everyone get it okay yay oh and if you're a speaker you get d NYC tokens too so uh double double so we have a leaderboard oh follow us on Instagram and Twitter tag us tag us tag us uh on dite NYC on Twitter and on Instagram tag tag tag tag also tag um sweet Mouse and thank you for giving us this space and also tag uh Williamsburg Pizza uh is the guy is Williamsburg pizza guy here is he here he was going to come I don't know okay uh tag Williamsburg Pizza they gave us the pizza um this is my contact info uh if if
you're if anyone here is interested in sponsoring one of the upcoming events uh you can email me M Fisher 1000 at gmail we've had a couple of people interested in sponsoring the Brian Johnson or the pile Health one so uh these all these are these there are real costs associated with doing this event and it basically comes out of uh my pocket so if you're interested in donating to sponsor another event uh feel free to email me okay uh quickly uh this event is also sponsored by science validator Network it's a it's a science validator network is a new way of like collecting scientific data for this Aid driven scientific future um you know science is only going
to get us so far uh AI is going to take us to the next level um we need better data sets the currently the problem with data today is it's all scraped from Reddit and the Reddit data is often times very bad um so we're going to try and create create a very precise data set um with very high quality data and every person that contributes it to it is incentivized for their contribution with a crypto token um there's a validator network that we've created that uses Ai and prompts to validate the scci the the accuracy of the data um this is a different way of doing science it's not sort of a causal thing where you have a hypothesis an experiment results and a conclusion this is sort of
based on modeling of the accurate modeling of the world and letting the science the the AI come up with new ways of doing it um partially we're doing this we're hiring if anyone is interested in working together with us on this project and also hosting some of these events or mostly it's on the science valary Network we're hiring we're backed by the top um VCS in crypto including poly chain Dow five bellagi and blizzard um oh and that's all okay anyone have any questions questions one question two question one question okay one question come on yes go
ahead oh Rebecca's also here she's she's not here today but she's in charge of our social media anyone else okay all right so without further Ado I'd like to welcome David shore on stage [Applause] um so I'll give David a brief introduction I met David in New York City um and he was doing some really really interesting stuff at the intersection of politics and Science and polling which he'll talk about and uh he worked with Obama's uh reelection
campaign they called him the uh Nate Silvers of OB Obama's personal Nate Silvers and uh I'll let him take it away so let's give him a warm Round of Applause [Applause] all right uh thanks everyone for coming uh and I'm you know I'm going to spend today tonight mostly talking about politics uh so hopefully that's okay uh so first uh and all right it looks like everything on the screen works the mic is basically working f up so uh first I'm just going to talk a little bit about my company uh I am the co-founder of a shop called Blue Rose research we're about 45 people
uh we work with basically every Progressive group uh that you might have heard of or more accurately maybe like 70% of the progressive groups that you might have heard of uh our main thing is that we spend uh most uh most of our time basically doing testing I'm going to talk more about this in the presentation but really up until very recently nobody knew what actually worked in politics and what didn't work and you know uh machine learning and statistics is helping us finally shed light on that so the first thing and I just want to make sure people can see the screen but the uh first thing uh is that when when you are working on a political campaign
there are really only two questions that matter first is are you persuading voters and are you mobilizing voters uh mobilizing voters complex topic honestly I'm not going to talk that much about it tonight but persuasion uh I'm going to try to go into some depth on so the big problem in politics probably the fundamental problem in politics is that if somebody sees an ad or hears a speech and changes their mind we have no way to know that um because there are no tracking pixels in people's brains you know yet anyway um and you know so what that means is that uh you know the signal that you truly care about is invisible to you but there are a bunch of other signals out there there's you
know who clicks what's the clickthrough rate on your emails or your ads you know how many views did your thing on YouTube get and the the problem is that the only people who engage with Democratic political content are highly engaged educated largely white liberals and if you choose what to tell swing voters on the basis of what excites liberals you're going to have a bad time uh so this is a perfect example of this so uh you know I'm going to just start this was uh Hillary Clinton's mirrrors ad she put uh it won a bunch of awards uh she put about $8 million behind it uh you know in the week before the election and
you know the basic idea behind it is that there are a bunch of uh there a bunch of little girls and they're looking in the mirror and then Don they see across the mirror instead of themselves Donald Trump saying bad things you know racist things sexist things makes fun of uh disabled people at some point um and then you know one of the women at the end she priz it's very powerful I guess um and people really like you know people like me really liked it um the problem is that when we actually get a randomized control trial where some people saw it and some people didn't and then we saw how they voted after it turned out that it made people more likely to vote for Republicans uh which is bad uh at least from my perspective I'm sure there are
some Republicans in the room but uh and you know this is not a rare thing uh you know we actually find across all of the tests we've done that uh roughly one in five pieces of content that Democrats make with the intent of persuading people to vote for Democrats actually makes people more likely to vote for republicans and so you know my personal theory of change which is I spend most of my time doing is let's identify those pieces of content and not show them to people uh simple uh and so just you know uh just to kind of go through this is a diagram of how what we do Works uh you know the basic idea is people go into a survey we ask a bunch of questions and
then there's a control group that does not see an ad and then a treatment group that does see an ad and then afterward you can compare you can you can ask them who they're voting for and if the control group is 50% you know haris and the treatment group is 52% haris then you can causally say that the ad change their chance of voting for you by 2% uh and so that's the basic mode of analysis it's easier said than done it turns out but that is that is the idea um and so you know this method of analyzing things and corporate space people call it Grand lft in politics uh no one it's really nobody did this until about four to four
or five years ago in 2016 it cost and you know I remember uh you know when I worked in 2012 uh you know a campaign person I was in analytics we were in a thing called The Cave they like locked all the nerds in a closet basically and so uh one campaign leadership person comes in he says oh you know I'd like to I'd like to test my ads and see whether they work or not and so then me and you know all the other people we go and we're like all right well you know they it'll cost about $80,000 and it will take six weeks um but then we'll get you an answer and then he rolled his eyes and left the room uh you know said muttered some stuff under his breath and so you know basically we've spent and the industry has spent you know uh the
last eight years really driving down the cost of things in 2016 it cost about $30,000 to test an ad we've gotten that down to about $300 and what's exciting about that is is that now it's extremely cheap to test them and so for the first time in the history of political strategy you know it's actually very it's actually very straightforward to find out what persuades people and what doesn't um so you know this is just you know some graphs of our of our volume just to give you a sense of scale um you know we've done something like 30 million uh survey interviews since 2021 we've tested tens of thousands of pieces of content and you know it's just very cool because at that point when you have 30,000 pieces of content you can really learn and patterns about what works and
what doesn't and so here uh this is a graph you know every single one of these bars is a message that we tested for one of our clients and you know what you could uh you know uh what you could see here is and I guess I'm really lucky that the screen ends right where my hand hand is but like these messages over here are about five times better than the average message and then all the way over there you see a bunch of uh a bunch of messages that again actively make people for repu which is bad uh to me uh and uh so kind of a lot of the job of politics is that campaigns are really this massive decentralized uh you know Endeavor where
you have thousands of different groups each of whom you know cumulatively employ tens of thousands of people and they're all constantly saying stuff you know whether it's on TV whether they're posting things on social media and the job of you know uh of a campaign is how do you make sure all of the those people say the stuff that's over here and don't say the stuff that's down there in red uh so the wor uh the well you know uh I I think uh it's basically what you would expect you know uh I'll get to that I swear you know uh all right so now I'm going to show you a wall of numbers um
uh this is not appropriately sized but I'll just walk you through it so the basic idea is that we're trying to measure uh important for every single issue we have 40 different issue areas things like inflation Medicare education you know whatever we basically measure how important voters find it and then we also measure how what whether or not voters trust Democrats on that particular issue and you know the issue landscape for this election you know when Democratic strategists are trying to figure out you know uh what to do is that if you look at the top is issues that voters care about right now it's the cost of living inflation the deficit
uh crime National Security and these are all terrible issues for Democrats um you know partly because of the moment but really these have always been bad issues for Democrats you know the basic dynamic in two- Party politics is that Center left parties tend to have an advantage on issues like healthcare or you know equality while the right has you know an advantage on things like fiscal discipline on crime you know people talk kind of about how there there's a mommy party and a daddy party and you know that's the trust thing is always there but what people care about really varies a lot in 2020 for example uh de Democrats had a really easy strategic
position which was that the two issues that voters cared about the most were uh Co and Healthcare and those also happen to be the issues that people trusted Democrats on the most uh and so all we really had to do was talk about those two things and ignore the people who were like hey let's not have police um and so that was a much easier problem now the problem is that the issues that voters care the most about are issues that voters just don't trust Democrats on while the issues that people do trust Democrats on you can't see this because it's it's blocked but it's like abortion and climate change uh basically uh voters don't care about a lot right now
and so the basic Strategic Mission for Democrats is you have to get people to care more about abortion uh and then you also have to get voters to trust Democrats more on crime and the deficit and inflation um so that's the Strategic Outlook um this is a different graph this is just showing the difference between what Rich Democrats care about and what poor Democrats care about and this is just kind of a helpful thing to see um at the very top is climate change you know I actually I once um I once had uh I once had breakfast with a CBC member uh member of the Congressional Black Caucus and he he uh while we were having breakfast he was just like you know it really what really strikes me is
white people are obsessed with climate change you know I I don't get it um you know uh and you really uh this just kind of comes through it's something the rich Democrats care a lot more about well if you look at what poor Democrats care a lot about it's unemployment housing poverty you know and I think that's kind of unsurprising but there's kind you know the problem is the people who work like all of this polling is very important because the people who work in politics on both the Democratic side and the Republican side are all kind of people like me you know they live in cities they're young they're educated um and so you know the stuff that we care about is you know not really representative uh of the stuff that
persuadable voters care about and so you know a lot of the challenge in politics is how do you like step outside of this bubble that you live in and you know uh in general if you want win it's better to talk about the stuff at the bottom versus talking about the stuff at the top um so this is just another graph um which is just a series of test results um but it turns out talking about things that are more popular makes you win more that's an incredibly controversial thing to say in Democratic politics but uh it is it is true we've done the tests you know it turns out um and just kind of we've pull a couple hundred different policies in a consistent format and one of the things that kind of comes through
very quick clearly um is just that voters care the most about concrete issues that affect people's lives you know uh things like healthare things like education while a lot of the things at the bottom I won't say them out loud but there are a lot of things that you would expect um so what this table shows this is just kind of showing uh how effective different topics are and that kind of answers your questions a little bit um and you know just turns out that what topics politicians choose to talk about really matters a lot that's another controversial statement in Democratic politics but it's true uh where the very best thing you know for Democrats to talk about is Social
Security and Healthcare that's been true for 40 years um and it continues to be true well to answer the other guy's question you know in terms of what's in the bottom it's you know student loans I hope that doesn't make anyone mad or uh or you know talking about January 6th but you know you just kind of see things that concretely affect you know uh the median voter's uh life and you know it's just important to remember that the median voter is about 50 years old you know um like and and so you know that's just kind of what works how you say it matters so this is from our uh this is from our treatment library and this is showing uh our single best message on
immigration you know what you could see over here is that immigration is close to the bottom in general it's a pretty terrible topic for Democrats but if you talk about it in the right way uh the overall average of immigration messages is 20 is is 25th percentile uh but by talking about it in the right way you can go from 25th to 60th um or 70th and so this this immigration message you know just kind of talks about common sense Solutions uh fixing a broken system that incentivizes people to break the law and you know the whole return to these kind of testing is to learn how to talk about things in a way that doesn't alienate conservatives or alienate Trump
voters um and it turns out that on basically every topic there are ways to do it and you know this is a this is just kind of showing over here uh some good messages at the top and some bad messages in the bottom uh you know I'll just read the top one and then the bottom one and then you can kind of get a sense for things you know the first is Democrats say you should support the Democratic candidate because they support pro-immigration legislation this is about fixing a broken system if we don't want migrants crossing the border we need to work to now we need to work now to reform legal methods of Entry by reducing wait times for visas improving background checks and Staffing our immigration system in courts that's a pretty reasonable thing that you can say to your uncle you know uh without him
yelling at you uh and that was a 70th percentile message relative to the kind of things that our clients come up with which is very good for immigration baselines like 25 the flip side is that our worst uh immigration message is Democrats say uh that you should support you know Democrats because they support pro-immigration legislation immigration detention in the United States is inhumane abusive and unjust immigrants are entitled to the same basic rights human rights as everyone else Biden must prioritize this one's a little old Biden must prioritize reducing the number of people behind bars and immigration deten detention centers across the country and you know I I personally my politics I agree with that sentence uh turns out
most swing voters don't and that's one of the really challenging things you know when you when you do this work is that uh as I said before the people you're trying to persuade and the people you're trying to mobilize are not like you uh if they were they wouldn't be swings you know uh the key thing you know that characterizes swing voters uh is that they don't have strong party ties and they don't buy into comprehensive ideologies and so that means that you know you have to purposely suppress yourself and not say things uh that you think people won't agree with and you also have to take frames that just don't come very naturally to you um and so anyway that's the end that's the talk I since there's a lot of crypto people here I just
thought I would post the quality Market odds of the election outcomes uh and there we go uh thanks [Applause] uh so I think uh the plan is for me to take questions um but I'm gonna oh cool all right I'm just gonna let's do one oh great ftic right everyone [Music] [Music] smile all right um I'm just gonna go in order I got start yeah so I noticed on of the slides
you had a breakdown of the effect on of one of the messages by like age groups and different things like that I was wondering how often is it the case that there is this kind of heterogeneity in the message and if there is heterogenity in terms of like statistical terms do you think that that actually reflects like a substantive difference in the way the message is being received like how you differentiate those that's a great question uh actually you know sometimes I say that's a great question is filler but no that's actually it's a great question so you know I think a lot of the news coverage of my industry is really focused on microt targeting you know it started with uh Mark Penn he was
talking about soccer moms and I think everyone got really into this idea of let's say one thing to one person and another thing to another person and you know the reality is that politics doesn't work like that you know for two reasons you know the the the first reason is that even if you could find the you know 50 people in Bridgewood For Whom defund the police is a good message like the Republicans exist and they will record you saying that to them and then they will broadcast it out to everybody else um but you know what we see is look I mean there are differences and what persuades people and it's like very interesting sociologically to to study that you know probably the biggest AIS
or the biggest two axises is some things appeal more to liberals some people uh some things uh you know appeal more to conservatives and the other age it turns out young people and old people are persuaded by very different things you know because they're very different much like liberals and conservatives but at the end of the day campaigns that win win by finding broadly persuasive messages that appeal to everyone and I think that's probably the biggest misconception about okay where does crypto fall in the political football right now is it left issue is a right issue yeah you know it's interesting we did so we did a poll uh of crypto users uh that I tweeted
about maybe like admittedly in 2022 so things may have changed since then uh but you know overall we find that uh crypto owners are a democratic leaning group uh for the basic you know it's about it was at the time it was something like 61% Biden um and you know that's in some ways unsurprising for the basic reason that it's a group that's very young very educated lives in cities um also disproportionately non-white you know which is also part of that um but there's also a dynamic where the more crypto you have the more Republican you are and we're all on Twitter we we see that um but you know I I I see the crypto space as kind of classically cross pressure um because on one hand
you know they are you know mad about a variety of regulatory uh actions that and positions that have been taken by folks on the left but on the other hand you know the kind of people who are into crypto should be Democrats like if you just like found like in general if you like look at hyper educated young men and who live in cities like they're a pretty Democratic leaning group and so you kind of have these two things kind of going against each other it's one of the big reasons and you know it's one of the big reasons why they're especially persuadable and why you're seeing both sides kind of fight over their votes right now and then the other thing I'll say about crypto you know outside of polling is I think what's pretty interesting about crypto is there's really only two groups of single issue
uh there are only two single issue advocacy groups that really exist in politics uh one is Israel and the other is crypto there are lots of ISS there are issue advocacy groups that are part of a coalition like if you look at you know the labor unions or Planned Parenthood or the NRA they're all basically in one way or another appendages of the Democratic or Republican party um but uh and so as a result their spending tends to be pretty pretty concentrated one party or the other crypto is one of the only ones that spends on both sides because they genuinely care a lot about their so
that's well uh if you're interested there's the there's a really good uh I think quanta article about this you know uh in general uh prediction markets are usually more accurate than polls for the basic reason that polls are an input to them uh it's interesting that different vetting markets seem to have different levels of accuracy uh like you know if you look like uh manifold for example in 2022 had a higher accuracy rate than poly Market metaculus eat all of them uh that's very funny it's been a long time since I worked in finance but it's very weird to me that the prices are not the same everywhere but I know people are working on that um
but Rel to their partisanship what are some politicians that you think are replacement and some below replac do do you mean he he asked which politicians are above replacement or below replacement do you mean that in terms of ability to win elections or policy or how much I like them or in winning elections and what drives your over performance INE how does that relate to your work another great question uh so I think one of the single biggest misconceptions about politics uh is just that you know people who uh uh I think people who work in politics and who care a lot about
politics tend to have fairly ideologically extreme views and they want radical policy change I'm one of those people that's why I got into politics um but it turns out that actually sorry it turns out if you look uh at what yeah it turns out if you look at what politicians are the most popular like if you just go and look at who are the most popular Senators who are the most popular Governors it turns out that it's usually you know uh people like Joe Mansion or people or the actual answer are a bunch of Governors you've never heard of people like Phil Scott he's a republican governor in Vermont uh you know berham who was almost the BP but he's the governor of North Dakota and
you look at them and you're like I've never heard of these people they've never done anything and it turns out voters love that uh this is the single biggest misconception about politics is that swing voters are deeply suspicious of both parties do not buy into either liberal or conservative ideological super structures and generally speaking do not want anything to happen um status quo bias is just the biggest thing in politics and that makes it really hard because both parties are controlled by people who want to do things uh but you know whether or not they stay in you know the people who decide what party controls uh things don't want things to happen and so it's funny this is called
thermostatic uh it's a terrible name but in political science people call this thermostatic public opinion that basically when Democrats win elections uh the public becomes more conservative and when Republicans win elections uh the public becomes more liberal uh I this is very you know this is a pretty robust effect it's surprisingly large if you look at graphs and I think it doesn't really feature you don't hear activists talk about it very much because it's very annoying from the perspective of wanting to change things um all right another question there a ton of questions all right I'm going to go over there and then I'll get back question
yeah so he asked basically it doesn't see uh when you listen to politicians talk it doesn't seem like they're talking about the good things it seems like they're talking about the bad things another great question and you know the answer to that is if you actually look at ads uh made by Democratic campaigns they tend to be you know they're pretty by this metet rate good in that they're boring you know they're like oh we want we want to lower healthare costs for seniors you know we're laser are focused on the cost of living you know whatever uh and that's because you know when the the people who run the Democratic party have an enormous amount of control over what ads end up being about but when it comes to
you know what are random people talking about on Twitter you know the Democratic party actually is a coalition of probably like a thousand different groups many of which are quite small all of whom you know to be frank have to do a lot of online fundraising and it does just kind of turn out that the messages that help you raise the most money uh are very different than the messages that help you win and that's just cuz democracy is a really weird thing you know like in in business if you can get 1% of the public to buy your product you will become fabulously wealthy and that's incredible but in politics uh you know you need that doesn't get you anywhere and so it's just you there's fundamentally different strategies that
get you to 1% or 3% then they get you to 50 and that's just one of the hard challenges in the space um I'm GNA uh in the corner do do that quickly yeah yeah okay uh I'll take it yeah what are the top three messages that the campaign for Harris President should be pushing in the Russ Bel states to get a victory what are the top three messages the haris Campa the r Bel States uh cost of living Healthcare abortion that's basically as I said yeah um uh and how much
yeah sorry all right uh let's uh let's just have a couple people say questions and then I'll synthesize uh let's do that group in the back over there from the custom perspective it's way more effective to say that messages not the pro things like to say that Republicans have not tried right
[Applause] yeah oh yeah yeah I think I understand your question um also I just want to say you know I I used to live in I used to live in Russia and when I told them about my research and election forecasting nobody was impressed it turns out um but uh so he's in some ways kind of asking about positive messaging versus negative messaging and and just whether it makes more sense to boost yourself versus attack your opponent and you know what I'll say about that uh is that positive messaging almost always does better um there are exceptions um the it's it's a very funny thing like you go to the donor meetings and you would really think like I think a lot of leftists have incorrect models of what happens in these things you know they
think it's like there's some rich donor he has a mon and he's like I'm I'm going to take my money out unless you you legalize coal but actually rich donors just like are rich people who watch a lot of MSNBC and so then they come in and then they get mad and they're like oh you're letting the Republicans walk all over us uh we need to punch them in the face and then you go ah negative advertising I've never had not thought of that we will give that we'll get right on that we'll give a shot um but the basic problem is that as I said before swing voters like Democrats less and also hate Republicans less than I do and so it just ends up not really working you know the med and swing voter already has a negative view of the Republican party but also a negative
view of the democratic party if swing voters were like in a place to absorb our criticisms of the other side they would not be swings they would be Democrats um all right next question next question next question all right how much does the party actually listen like how much how much is research like ground Truth Versus like H we'll kind of think about it uh 5050 you know I mean the reality is I think there's a mix of sometimes there's a lot of there's a lot of Coalition politics that prevents people from saying the most ACC you know the most effective thing but then the flip side is if you don't listen to the Coalition politics
then they start on you and so you know there are real constraints politics is hard what was theising thing un um okay probably the most surprising thing cuz you know a lot of the stuff I've been saying is like ah defund the police is bad Healthcare is good electric cars are really bad like if I had the choice between uh if I had a choice between a Democrat embracing defund the police or electric cars it would be pretty close um like voters really really hate electric cars I didn't expect I don't drive I'm I don't know so
con Harris you know I don't I don't want to get in trouble you know I I mean I I have boring answers to this I'm a big I'm a big ret Wimer fan you know I like Josh I don't know um uh I think people who want to ask questions should probably just like come to the front it's hard what's your number one uh tweet a lot uh I think it's the answer um um you know the honest answer is I don't know my best guess is that it's pretty close um there are like I I've met I've met Republican versions of me they exist
you know the other side has a lot of money too
I so he asked are llms being used in politics the answer is yes both
sides the really the really big thing is that old people used to be silent generation and now they're Boomers and boomers are just a lot more liberal than silent and so you know this just Democrats do a lot better with old people than they used to and they're doing worse with young people so it's a very interesting time you
know it is fun to be involved he's asking how this how I think the space will evolve it is fun being in a zero sum game you know it's very sobering uh I think many of us are in Zero Sum games um I assume LMS are going to take over everything but that's the direction uh I hope so uh that you okay I'm G to get an answer to this though and I'm sorry so you know I think a lot of people in politics sorry a lot of people see like this space as like kind of fundamentally illegitimate you know they're like oh you're just manipulating people you're radicalizing people and you know what I'll say after having been
a Pooler for a while is that every time we've ever done a test it has said you should be more moderate you should be less divisive you should be more positive I think that a lot of the negativity that exists in our system kind of is because both parties are dominated by highly ideological people who are young and you know uh and hate each other and I think that you know in our political process when you just kind of look you know educated people are more likely to run for office they're more likely to uh to donate they're more likely to vote and so at every single step the views of educated people are uh are overweighted except election day and
polling is basically the only chance that you know regular people get a chance to say what they think um I'm going to do one more question and then [Applause] [Music] [Applause] [Music] yeah you know there's another misconception in politics you know like people like to divide things into persuasion and mobilization where they're like you know we have the swing voters like the like the fat people in Ohio with the lunch mailes and then we
have the base which is like which they imagine as like I don't know like super liberal young kids in college campuses uh but the reality is you know your persuasion targets are people who you're pretty sure are going to vote but have a 50-50 chance of voting for you and your get out the vote targets are people who would probably vote for you if they voted but you're not sure if they're going to turn out or not and that group it turns out is overwhelmingly workingclass overwhelmingly nonwhite overwhelmingly young um but the important thing is overwhelmingly modern like if you look at Democratic persuasion tet like the list that Joe Biden is going I'm sorry Tom Harris uh is going to use uh toig you know to figure out who who did not on their
doors 60% of those people identify as Moder uh only 20% of them identify as liberal and so it turns out that you know the same messages that uh kind of get uh moderate people you know like swing voters to vote for you are generally the ones that get these kind of you know young workingclass people of color to support you as well there are some trade-offs but I think it's less than you think um all right uh I guess one
more uh we do our best but it's hard uh you know uh but yeah I I I think you know the policy polling project that we did was you know this kind of cool idea of after the 2020 election we got all the major stakeholders you know people like Joe Mansion Squad all the different groups they all submitted policies and we tested them in a standardized way just so that folks could could make these optimization problems and you know it's been it's been an interesting ride that's like what I'm trying to do as influence policy but it turns out politics is hard and complicated so you know T um all right that's all the questions thank you so much be hang out here thank you guys for coming out enjoy the rest thank you to Sugar Mouse as
well you like up and say anything uh thank you to williamsberg Pizza thank you to all for coming we'll hope to see you for Brian Johnson on August 19th until then bye
[Music]
Dan [Music] [Applause] see [Music] [Applause] [Applause] [Applause]
like every day [Music]
my [Applause]
[Applause]
[Applause] [Music]
[Applause]
no
do
[Music] [Applause] his [Music]
[Applause] me and then
[Music]
middle