 And we turn to hear from Marshall Van Alstine from across the river at BU who among other things is the person who coined the phrase and assisted society's investigation of network effects a phrase we now cast off casually but that has been kind of seismic in its way of having people think about the way that networks affect life and become embedded and gain their own momentum etc etc and Marshall I I think you've got an idea or two to share over to you well thanks John and you're you're generous for credit there I think there are lots of other folks also working in network effects I'm hoping that we can do some interesting things with that I also want to see if we can use some platform business models to speak about ways to solve this what I particularly thank Jill for her comments on political ads and creating up the context for this because one of the things I really want to address is specifically the falsity of news and the falsity of advertising in in politics I think the example just posted a moment ago deep fakes not allowed in the Twitter case but are allowed in the Facebook case is exactly the kind of thing would like to try to go into I'm hoping that I can share screen then give this a shot let's try this top two very well I'm so I have a couple of slides for a couple of different ideas I'm no if this is coming through or not so far so good I also want to apologize in advance I'm literally just coming off of the fevers my voice is a little bit horse relative to what it should be so I'm hoping that I'll be a little on the nose but yes you know it's certainly certainly relevant so I want to try to introduce an idea from information economics a lot of the different solutions for fake news have come from the computer science side or some from the regulatory side I'd like to use is some ideas from information economics and mechanism design to see if we might be able to go after it to give you a quick summer and I invite feedback on this I have a even a short write up on this will really would welcome your input on ways to approach this particular problem if we take a look at some of the existing solutions each of the lines in blue or some of the common versions of it whether it's fact-checking for crowds and algorithms whether it's media literacy and educating consumers truth chasers where folks can kind of you know follow up false ads with false information true information you know the other really popular one is tagging and product labeling know the idea of being to inform folks that they're getting bad information or banning the content or the person to remove the problem or demoting it in the newsfeed almost every single one of these has certain challenges you know whether it's the fact-checking and the product and tagging and labeling one of the biggest problems there is that they then try to discredit the labeler so if the original story is false and maybe they simply try to discredit the person doing the labeling as we've seen in some of the attacks in the media I think there's a wonderful book out on that recently you know with network propaganda or if you look at educating consumers often there's this question of confirmation bias so many folks would distrust those who would educate them empirically it turns out the truth chasers just don't seem to work they seem to have very little impact of what folks believe and interestingly enough after Alex Jones got de-platformed from a couple of folks for propagating fake news it was remarkable that his uh that visits to his website exploded so in some sense use that not only to increase his popularity but then claim media censorship and generate even additional revenues from other sources I should say by the way that's fascinating it'll be interesting to hear I think Joan Donovan who will follow may have something to say about it I had thought the conventional wisdom was that Alex Jones had been defamed as a result of that so fascinating thank you so it's interesting I think there may be a temporal version of it it's not in this slide deck but in the different slide deck Alexa I can show you the actual plot of visits to his websites and Alexa showed that his rank improved by several dozen places following that for some fairly extended period of time so I think it has eventually fallen back below but it was immediately after he actually experienced quite a rise in attention for hadn't been the platform from all of these in the categories also didn't raise a question for you which of these methods effectively changes the incentives to produce misinformation that one of the problems that identified here is really what I would argue is really that almost all these solutions put the burden on the user or on the platform rather than the author of the lies in some sense almost all of them the fact checking the tagging educating consumers the truth chasers all of these put the burden on someone else and they don't necessarily change the incentives one of the things we'd like to do is to see if we can actually put some of the burden back on the authors of the lies see if we can actually clean up the proportion of information that's misinformation in the news feed and again I want to thank Jill for the setup in some sense the problem of fake ads is really wonderful was at one extent we've got Twitter which makes a decision to stop all political advertisement Twitter globally so they just won't take any kind of political advertisement for a candidate and the opposite extreme is Martin Zuckerberg who in Georgetown said you know they won't fact check political ads of any kind they think it's up to the users to do that so no matter how egregious the lies they're entirely willing to propagate it on the premise that the users can and should decide so at the one extreme Twitter is in a position puts us in a position of no political ads which prevents political discourse and it's actually extremely difficult for newcomers to come in and gain a voice if you're a new candidate and you can't buy a political ad it's a very difficult for you to actually gain the attention that you might actually want at the other extreme you know Mark Zuckerberg solution is to let basically anything go and they don't want to be the policeman or the arbiter of it and users themselves are then forced to decide so here's one possible solution this is one of several but one specifically just to try to keep this down to 10 minutes suppose that we were to implement something that was with a standard information economic mechanism which was of the form of an honest ad guarantee it'd be really simple mechanism suppose that political ads came with a guarantee of authenticity from the author in some sense this is a guarantee or a bond which is forfeit if the representations are false but they returned if they are true now here are a couple of the properties of this basic very simple proposal the first is that no one knows better than the author whether the basis for that claim is true you've authored that so question is are you willing to stand by the claim that you're actually making it's also the case that if it's simply a monetary pledge you know an honest ad guarantee that politicians and democracies are perfectly free to lie if they wish it just becomes more expensive on the flip side what's nice about this particular mechanism is that could actually still work even in totalitarian regimes if it were adopted in this case whistleblowers could get their messages out even under totalitarian regimes it would still be possible to disseminate truth so in that case it's an instance where others might try to discourage it but you could still get it out the mechanism still works you're simply trying to pledge that something is true and in the case of damage the proceeds can be used undo the damage or underwrite another public good if that's true now here's the simple illustrations here's a an example of how this might work so suppose that an honest ad pledge would come out with something like 10 times the ad price it would simply be put in x grow there's a hypothetical fact checker suppose that it snopes or political act or hoax layer or wonderfully even a random sample of fox and cnn viewers in this case what you're trying to do is to eliminate bias you just simply make it independent in here a challenger could pay a simple challenge price more than the ad price to fact check the ad the challenge should be more expensive than an ad because otherwise you should simply take out an ad if you simply disagree now if the ad turned out to have been false then the claim of the pledge goes to the challenger if the ad is true the chain the challenge goes back to the ad buyer if there's no challenger there's no harm and the pledge simply goes back to the buyer it's simply refunded so what are the different properties of something like this in this case the ad buyer has an disincentive to lie in this case it's simply more expensive you wouldn't guarantee an ad that's going to be very that's going to be misinformation a challenger as a disincentive to challenge if the only thing you're going to do is to prove an opponent right if it's a true claim then you're really not going to bother to do that the platform such as zuckerberg's case or twitter doesn't have the authority to decide the truth in this case and the fact checker has no incentive to cheat because the challenge price is paid regardless so unlike the financial crisis in 2008 where the ratings agencies were paid by the bank to doing the rating in this case there's no incentive to cheat so that this is unbiased and neutral but notice in every single instance the cost to guarantee truth is zero because it's simply refunded in this case again the cost to guarantee the lie is expensive but the cost to guarantee truth is zero either because it wasn't challenged because if it was challenged it's then proved to be true and it's refunded in the first place the next element of this is in a technological sense or in a bottleneck sense it easily scales you only check the challenged ads and in that case the cost to cover it's all very straightforward most commonly one of the objections is okay well what about half truth in this case well the beauty of this is we have a ready example immediately at hand many of you may have come across most recent case where a pack for trump bought an ad saying that biden offered ukraine a billion dollars to fire the prosecutor investigating barisma over his son what's interesting about that is that politifact actually rated that as half true yes biden's son did work for barisma yes a billion dollars with withheld but no it was not because they were you preventing corruption it was the opposite it's because they were not investigating corruption so for half truth you could simply get half the money back what's fun about this is that gradations in this of this estimation are not only possible they're already happening uh to give you some fun illustrations of this and these are already available on there um trump claims for example he won west virginia by a margin of 42 points 42 points well ironically actually that one turns out to be true um you know or if you pick another example uh you know there's a claim democratic controlled house never asked john bolton to testify that turns out to have been false or the claim you know perhaps i was the person to save pre-existing conditions in your health care well not only is that false but the attempts to repeal the prior legislation would have meant that the existing uh protections on pre-desting conditions would have been overturned so uh politifact rated that when pants on fire false here it's simply an illustration that the gradations are possible so it should actually it should still possibly work there's other elements of this that actually have some interesting properties to it so let's take a look at a couple of other interesting examples so how do you know how to set the pledge or the size of the lie price with the honest ads guarantee what's interesting about this is it doesn't actually matter it turns out that it's actually self-adjusting suppose that exon mobile keeps claiming that fossil fuels don't cause global warming or that morris claims that cigarettes don't cause cancer something really interesting if this keeps happening because what that implies is the lie price is sufficiently low in an economic sense what's happening is that the private gain from lying is exceeding the social cost and so they're keep buying the ads the implication is the externality is so high that you should increase the lie price what this means is we've now just found an economic mechanism which is efficient search for the size of the negative externality that is in fact causing harm so in the context of global warming or in the context of cigarettes or other things of that source the honest ads guarantee actually serves as a mechanism to in some sense internalize the cosian externality exactly like the carbon tax on global warming those kinds of things actually can be priced by a market mechanisms or a market for truth in this particular context there are possible interesting thought on this is what about poor political entrants those that don't necessarily have the 10 times lie price in order to post the bond well this one actually turns out be quite easily you simply precheck the message in that case the bonds markets can easily ensure quantifiable risks again the whole idea is to create a marketplace for truth these kinds of things easily exist and it would be very straightforward to try to ensure claims of that sort now legal scholars among you might object one step further one of these is what about first amendment scrutiny it's difficult to require a pledge to speak because you can't require funding in that that would be considered a friction or a limitation on speech the beauty of this is that in political context voluntary compliance is a signal this is the source of the Nobel Prize in 2001 for information economics won by ackerloft spends and stiglitz the idea is that credible signals are those which are expensive in this you can actually signal a an important action even though the obvious illustration of this do you trust a product that is sold as is in this particular context reluctance of a super PAC to promise that it's not lying suggests that they probably are lying especially when in this particular context lying not lying is actually free because it's actually refunded in this case but you're putting something at risk which makes it a credible signal and you're going to be more likely to put out truthful messages in that context for what it's worth this is a whole this is just number two and a whole collection of possible market style interventions in a market for truth to see how this works but Jonathan asked us to keep this to about 10 minutes so those of you that are interested we're actually working on a collection of these and would actually then you know love to engage in discussion other ways to see if we could actually operationalize this so by all means jump in push back and let me know what you think Marshall thank you so much the classic contestable proposition as a way of getting people thinking about things and in fact before we turn to our third presentation and Joan Donovan I just want to if it's permissible cold call none other than Dean Martha Minow who couldn't help but feel was almost invoked by the mention of Newt Minow in Jill Lepore's presentation but also here I'm so curious Martha this seems to be like exactly in your wheelhouse to ask an incisive question about it's a very large wheelhouse but this is like dead center so I'm so curious if there's a question on your mind that you'd want to ask Marshall right now before we move on it's an ingenious idea and it makes the idea of marketplace of ideas something real rather than a distraction so I'm intrigued by it I'm interested in your last comment though about the bond market I mean it I'm trying to think about the auxiliary actors here who that might be the bond market it might be the intermediaries that do the pre-checking none of that is free so how do you think about those costs so actually what I'd like to do again this is part of a much longer discussion is to design a governance model to make to have other parties participate in numerous different ways so the governance model I would think would split the same way that we use current governance between kind of executive legislative and judicial branches I would split the definition of fake news from the adjudication of fake news and then the execution of the rules applied to fake news so you could for example get Fox and CNN to agree on the definitions of fake news then organizations such as snopes and political fact and others could be the adjudicators of fake news and then the platforms such the social media platforms could be then the executors of the decisions regarding that fake news but this but you're correct you then invite a whole host of other parties because they're financial transactions to take an interest in the risk bearing and it's entirely possible that the markets for risk bearing the hedging the insurance would be entirely happy to underwrite different candidates in different in different ways and again given that there would be mechanisms to pre-check these things you can then assign different risk levels to different categories of statements that are being made and these could apply even to the risk levels that ExxonMobil might claim relative to the risk levels that Donald Trump might claim so you could assign risk levels in different prices to exactly those kinds of things and again in marketplace here ought to be able to price these things fairly efficiently. Jonathan can I ask a question? Sure go ahead Jolapur. Yeah I just think this is a fascinating presentation I'm just wondering though given that I mean I see this is largely a problem of civic education and better informed citizenry to outsource this to the market at a time of rising populist sentiment seems to me profoundly disempowering to ordinary citizens in a democracy so I'm being your provocation with a provocation but I wonder how you respond to that. So you know my favorite article on that kind of issue is a really wonderful one done by Dana Boyd she has a fantastic article which is you think you want media literacy do you and it's a really wonderful send up of different ideas on that she makes a couple of different points um one of them is that the attempts to educate the population have a number of different effects one of them is to get the exact same kinds of pushback that you get in other contexts which who's presuming to do the education in this case so lots of folks on the conservative side might say liberals are trying to educate them and wouldn't have the right to do that so another the question they necessarily pose is what gives you the right to educate me in this in this context another point that she makes which I think is also quite profound is that the tendency to provide media literacy has empirical evidence to suggest that rather than being discriminating between true and false information the tendency is to reject almost all different media rather than to differentially reject them so it is in some ways counterproductive and it has the tendency to make people more cynical so in this case the hope is to help clean up the media stream and get people to make more truthful claims in addition to these other mechanisms but I think the market for truth has some potential to balance a lot of these different interests my hope is that we can create a number of different knobs that you can tune I'll give you a really simple example in this case a society that values greater freedom of expression so it's at the live price very low at the moment the current live price is zero you might anticipate an asian society which doesn't value expression perhaps as much but values the integrity of messaging higher it might set the live price higher but even there the beauty would be that in a different context by tuning a different knob dissidents could even get their message out rather than simply having a government say that okay you can't even discuss Tiananmen Square or other things at that point I think one of the nice elements of this is you have knobs that you can tune with different societies able to tune them to different levels