! E L M E N T SE "#! $!%&' &( "#! )!*(+, !"#$%&!'(&!"' &)$ $#*+,&!*" *- &)$ .$""/ .0$%% ("1 &)$ $22(",$+ 3),03) 2,01$0% ALSO FEATURING • The Digital Pandemic • The Force on the Front!"# $%&#'(')&$)!# '#*#)'+" ,-$'%). -/ 0-*!-% +-..#(#-. +/ ' 0 2 0 21 · 1 &* 2 % ! 1 7 ·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’% '#!6!"*+ '3:$0(!)$% %$( -'#(/ !" 1996. !"# $%&%!'( )'*$#+%, Averting Negative Externalities of Disinformation Through Platforms #2*" 0*((!0/ “This is the cost of disinformation” – Atul Gawande, responding to nurse Jodi Doering’s story of her COVID-19 patients’ denial of their condition1 1. INTRODUCTION Eight months into a global pandemic and it appears the spread of the novel coronavirus is supplemented by the spread of disinformation. One story, in particular, illus- trates just that – South Dakota ER nurse Jodi Doering re- counts a few days’ worth of COVID-19 patients in a Twit- ter thread: “The ones that stick out are those who still don’t believe the virus is real…They tell you there must be an- other reason they are sick. They call you names and ask why you have to wear all that ‘stu3’ because they don’t have COVID because it’s not real” (Orth, 2020). How did we get here? Doering’s story draws a destitute parallel to developing countries. Despite e3ective remedies like mosquito nets and deworming pills, people in developing countries ne- glect to use them at large, partly citing a lack of available information, or misinformation more generally (Pulford et al., 2011). In the United States, with an abundance of readily accessible information at our fingertips, how are we, too, in a position where people neglect reality? Outside of the externalities inherent in a pandemic, widespread disinformation hinders and undermines the very integrity and the accuracy of fact. Disinformation has the potential to influence civic engagement, election results, and, most unfortunately, downstream health outcomes. In search of a remedy, we must begin at the root of the problem: sec- tion 230 of the Communications Decency Act. In so doing, the evaluations and recommendations in this paper contribute to a literature on the e3ects of media and disinformation on behavior and health outcomes (Bursz- tyn et al., 2020; Banerjee et al., 2019a; Bursztyn et al., 2019; La Ferrara, 2016; DellaVigna and La Ferrara, 2015; La Ferrara et al., 2012; Chiang and Knight, 2011; Jensen and Oster, 2009). Prior work has revealed that media ex- posure can increase hate crimes (Muller and Schwarz, 2018; Bursztyn et al., 2019) and mass killings (Yanagiza- wa-Drott, 2014); it can also a3ect domestic violence (Card and Dahl, 2011; Banerjee et al., 2019b), fertility choices (La Ferrara et al., 2012; Kearny and Levine, 2015), and respons- es to natural disasters (Long et al., 2019). Related to this analysis is contemporaneous work calling for a reinterpretation and a reassessment of section 230 of the Communications Decency Act as it stands today. Lotty (2020) argues the current interpretations of the scope of section 230 immunity wrongfully deny individuals who have been sexually harassed or assaulted an opportunity to hold online services accountable for causing or exacerbating their harms (Lotty, 2020). Thus, Lotty prescribes a necessary revision to section 230 to better align with contemporary views regarding the role of the internet user and the responsibility of technology companies to deter sexual misconduct. In a similar vein, Sloss (2020) responds to the Myanmar military carrying out brutal attacks via Facebook against Rohingya Muslim communities in Rakhine State in 2017. Going further than Lotty, Sloss argues for a statutory exception in 230 to permit civil suits against internet companies on account of their alleged complicity in genocide, war crimes, or crimes against humanity (Sloss, 2020). With a civil liability exception, Rohingya plainti3s could then bring a state tort law claim against Facebook alleging that Facebook was negligent in permitting its platform to be utilized to spark mass violence against the Rohingya in Myanmar. As the law stands today, such a case would be immediately nullified under the current federal preemption defense to state tort law claims. Irrespective of the study, the written law, or the lack there- of, section 230 most certainly demands a statutory modifi- cation – a modification which provides a positive incentive for companies to internalize the risk prevention of dis- seminating misinformation on their platforms. The re- mainder of this paper proceeds as follows. In section 2, I provide a brief history of the Communications Decency Act, followed by a substantive overview of the clauses un- der section 230. In section 3, I present a platform-by-plat- form analysis of their prior and current e3orts to prevent the spread of disinformation. In section 4, I introduce po- tential amendments to the current legislation, present no- table downstream implications, and provide empirical evi- dence of disinformation as a negative externality to society. Section 5 concludes. 2. HISTORY 2.1 Inception of The Communications Decency Act Introduced in February of 1995 by Senator James Exon, the federal Communications Decency Act (CDA) was cre- ated to combat a growing issue of extensive pornography and obscenity on the internet (Cannon, 1996). As passed, the CDA extends the “antiharassment, indecency, and an- tiobscenity” restrictions currently placed on telephones to 20 !"!#!$%& : : &'()$* 2021 interactive computer service providers (“ICSPs”). The bill was promptly met with opposition from some lawmakers and interest groups who opposed the idea of interfering with the Internet, along with concerns that the act was a “violation of free speech and … of the right of adults to communicate with each other” (Cannon, 1996). Opposing the CDA on the basis of these concerns, coupled with the 1995 New York Supreme Court decision Stratton Oakmont v. Prodigy Services Co., representatives Christopher Cox and Ron Wyden proposed the “Cox-Wyden Amendment” to Exon’s bill, some parts of which would ultimately become section 230 of the CDA as it stands today. In Stratton Oakmont, the court held that an online bulletin board service provider could be held liable for defamatory content posted by users on its platform, given that the web- site proactively monitored, screened, and removed o3en- sive user content, thereby serving as an editor and pub- lisher of all posted content and assuming legal responsibility for it (Stratton Oakmont v. Prodigy Services Co., 2018). By contrast, a 1991 New York case, Cubby, Inc. v. CompuServe, Inc., held that an ICSP that did not regulate third-party user content evaded liability for libel since it did not know of and had no editorial control over posted de- famatory material. Astonishingly, Stratton Oakmont as- serted that an online service that does nothing by means of monitoring for problematic content can never be legally responsible for the content of its users; in addition, a ser- vice that takes good-faith steps to screen such content sub- jects itself to liability (Ardia, 2010). As a direct response, the Cox-Wyden Amendment was pro- posed to encourage online services to take proactive mea- sures to improve online safety and regulate objectionable content without fear of increased liability (Ardia, 2010). Thus, this amendment allowed private ICSPs to address the problem of online indecency, while concurrently up- holding the Representatives’ policy goal of fostering the “vibrant and competitive free market that presently exists for the Internet and other interactive services” (Stratton Oakmont v. Prodigy Services Co., 2018). Subsequently, Congress passed the Telecommunications Act of 1996, which included the Communications Decency Act and the Cox-Wyden Amendment – legislation presently regarded as section 230 of the CDA (Ardia, 2010). 2.2 Section 230 of the Communications Decency Act Section 230 of the Communications Decency Act is the most significant piece of legislation regarding internet regulation, praised by commentators for allowing the In- ternet as we know it today to prosper. This statute largely restricts the scope of potential liability for interactive com- puter service providers based on content transmitted by third-party users of the service and an ICSP’s good-faith actions to limit access to objectionable material transmit- ted by users. The ICSP definition extends to “any informa- tion service, system, or access software provider that pro- vides or enables computer access by multiple users to a computer service,” which includes most websites and In- ternet-based apps (Stratton Oakmont v. Prodigy Services Co., 2018). As written, section 230 bears the name “Protection for pri- vate blocking and screening of o3ensive material” (Strat- ton Oakmont v. Prodigy Services Co., 2018). Section 230(b) lists the policy objectives, which include: promoting the continued development of the Internet, preserving the free market that is the Internet, spurring technological innova- tion that maximizes user control, eliminating “disincen- tives for the development and utilization of blocking and filtering technologies,” and guaranteeing the enforcement "[Section 230 of the Communications Decency Act] largely restricts the scope of potential liability for interactive computer service providers based on content transmitted by third-party users of the service and an ICSP's good faith actions to limit access to objectionable material transmitted by users." 21 !"# $%&%!'( )'*$#+%, of federal online security laws (Stratton Oakmont v. Prodi- gy Services Co., 2018). Section 230(c) sheds light on what constitutes “Good Samaritan” neutral actors and publish- ers online. Below the heading “Protection for ‘Good Sa- maritan’ blocking and screening of o3ensive material,” section 230(c)(1) states: “No provider or user of an interac- tive computer service shall be treated as the publisher or speaker of any information provided by another informa- tion content provider” (Stratton Oakmont v. Prodigy Ser- vices Co., 2018). Successively, section 230(c) safeguards civil liability for any “provider or user of an interactive computer service” given “any action voluntarily taken in good faith to restrict access to or availability of material that the provider or user considers to be obscene, lewd, lascivious, filthy, excessively violent, harassing, or otherwise objectionable,” irrespec- tive of if such material is protected by the constitution (Stratton Oakmont v. Prodigy Services Co., 2018). Overruling Stratton Oakmont, this legislation issues ICSPs that make bona fide e3orts to monitor and screen content immunity from legal actions based on user content gener- ated or transmitted on their platforms. Nevertheless, the Act explicitly does not permit ICSP immunity from liabil- ity for specific o3enses: violations of federal criminal law, communications privacy law, intellectual property law, and federal sex tra4cking law (Stratton Oakmont v. Prodigy Services Co., 2018). However, not included in the legislation: what constitutes such bona fide voluntary action, which explicit actions separate platforms from publishers, and how these specif- ic, yet ambiguous policy goals can be e4ciently achieved.1 At present, courts are progressively interpreting section 230 to broadly restrict the scope of potential civil and state law liability for ICSPs, thus fostering a safe environment for web-based corporations against a multitude of claims. How those companies do so is, evidently, up to them. 3. PLATFORM-BY-PLATFORM EVALUATION As of now, technology companies and media platforms are left to self-regulate, experimenting on their own with vari- ous moderation strategies and models of platform gover- nance. Essentially, individual companies are tasked with not only aligning their platform objectives with the needs of their users but to promote the principles of the common good as well. 3.1 Twitter In an interview with Boston College Magazine, Colin Crow- ell, former vice president of global public policy and corpo- rate philanthropy at Twitter, discusses the ramifications of Twitter flagging one of Donald Trump’s tweets for the first time (Figure 1). Signaling a new approach to combatting disinformation, Crowell asserts that “navigating the emerging and evolving online terrain of disinformation to protect the integrity of vital civic conversations, without succumbing to excessive censorship, is critical to safe- guarding the internet as a vibrant platform for human ex- pression” (Tempera, 2020). Here, Crowell echoes Twitter CEO Jack Dorsey’s overarching commitment to serving the public conversation. Although Twitter does not have an explicit policy on general disinformation, Twitter does present specific guidelines and policies on the following areas: violent threats policy, glorification of violence policy, platform manipulation, synthetic and manipulated media policy, and, most notably, civic integrity policy. According to Crowell, “when you have a platform as large as Twitter, it is important to recognize when there is a need to help users understand what they are viewing with additional context,” and Twitter’s civic integrity policy does just that (Tempera, 2020). Updated in October of 2020 ahead of the presidential election, Twitter asserts they will “label or remove false or misleading information about how to participate in an election or other civic process” (Twitter, 2020). This policy includes but is not limited to posting or sharing content that may suppress participa- tion or mislead people about when, where, or how to par- ticipate in a civic process. Further, Twitter explicitly com- mits to removing and labeling false or misleading information intended to intimidate or dissuade people from participating in a civic process, in conjunction with any intent to undermine public confidence in an election. On the same webpage, Twitter includes the subheadings: “What is not a violation of this policy?”; “Who can report violations of this policy?”; and “What happens if you vio- late this policy?” (Twitter, 2020). Thus, Twitter makes it abundantly clear both the ramifications and the qualifica- tions of their policy violations in an e3ort to transparently combat negative externalities to society. But is Twitter do- ing to prevent the spread of disinformation with this ap- proach? In defense of Twitter’s civic integrity policies, Crowell praises the decision to take action. The flagging feature, according to Crowell, “provide[s] a screen that [says], basi- 22 !"!#!$%& : : &'()$* 2021 cally, this tweet is in violation of Twitter rules, but it is re- maining on the service in the public interest,” as figures like Donald Trump are democratically elected heads of state (Tempera, 2020). Firmly, Crowell asserts that “Twit- ter can and should take action when online speech risks o5ine harm,” touching on negative societal externalities like disenfranchising voters and downplaying risks of CO- VID-19 (Tempera, 2020). Previously, Twitter only had the binary choice of either removing a tweet or leaving it up; now, Twitter allows journalists and the public to comment, links access to factual information, and takes steps to limit the ability of content to reach viral spread (Tempera, 2020). In accordance with section 230, simply flagging tweets does not invoke publisher status on Twitter’s be- half; however, Twitter begins to stray from the comfort of the “neutral actor” position.2 The existence of such policy pages represents an act in good faith on part of Twitter to screen for objectionable content, thus shielding them from civil and state liabilities. Nevertheless, Twitter is well within its right as an ICSP to prohibit attempts to use its services to manipulate or disrupt civic processes, along with distributing false or misleading information. 3.2 Facebook In a testimony before the United States Senate Committee on the Judiciary, Facebook CEO Mark Zuckerberg verbal- izes Facebook’s mission to “‘give people the power to build community and bring the world closer together’” (Testi- mony of Mark Zuckerberg Facebook, Inc., 2020). Zucker- berg touts that Facebook removed false claims about civic processes and displayed more than 150 million warnings on posts, after review by independent third-party fact- checkers (Testimony of Mark Zuckerberg Facebook, Inc., 2020). However, in response to Twitter’s fact-check labels, Zuckerberg takes a di3erent approach, “[believing] strong- ly that Facebook shouldn’t be the arbiter of truth,” in an interview with Fox News (Halon, 2020). Nevertheless, Facebook, too, makes an explicit commitment to stop the spread of misinformation. Under the heading, “Working to Stop Misinformation and False News,” Facebook makes its objectives clear in foster- ing accurate information on its platform. Specifically, fo- cus is given to three areas: disrupting economic incen- tives, building new products, and helping people make more informed decisions (Facebook, 2020). Citing finan- cial motivations behind fake news, Facebook commits to removing the economic incentives for tra4ckers of misin- formation – doing so through third-party fact-checking or- ganizations, strict enforcement of their policies, and apply- ing machine learning to assist response teams (Facebook, 2020). Reinforcing Zuckerberg’s stance against becoming arbiters of truth, Facebook is “building, testing and iterat- ing on new products to identify and limit the spread of false news,” engaging with their community “Oversight Board” to elicit help in fact-checking (Facebook, 2020). In doing so, Facebook is explicitly making bona fide e3orts to screen for misinformation, upholding an impartial posi- tion, rather than a publisher position in working with ex- ternal, unbiased groups. Through the “News Integrity Ini- tiative,” Facebook is pulling together over 25 funders and participants, including tech industry leaders, academic institutions, non-profits, and third-party organizations, to help people make informed judgments about what they read online.35 As a result of this initiative, Facebook posi- tions itself to advance global news literacy, bolster trust in journalism, and better inform the public conversation. Irrespective of hate speech and promotion of violence vio- lations, Facebook takes a hands-o3 approach to content moderation. Per the “Community Standards” page, Face- book makes its position clear: “We want to help people stay informed without suppressing productive public debate,” justifying the rationale behind not removing fake news from the Platform (Facebook, 2020). Rather, Facebook makes good faith e3orts to reduce the circulation of mis- information by adjusting their algorithms so as to “display [ fake news] further down in the news section,” thus avert- ing viral spread (Facebook, 2020). Essentially, Facebook encourages users to decide for themselves what to read, "Voluntary action superseded by profit in no way advances the e5ort on combatting the spread of misinformation; thus, there must be a disincentive to the 1 step forward, 2 steps back approach." 23 !"# $%&%!'( )'*$#+%, what sources to trust, and what content to post, all the while attempting to not overstep their boundaries as a neu- tral actor. 3.3 YouTube In the time leading up to the 2020 presidential election, a YouTube ad likening Democratic nominee Joe Biden to Venezuelan socialism circulated in the state of Florida (Donald J. Trump, 2020). Despite a published fact-check by the Associated Press, YouTube showed the ad more than 100,000 times in the eight days leading up to the election (Merrill & McCarthy, 2020). Trump went on to win the state of Florida by roughly 375,000 votes – the largest margin in a presidential election there since 1988 – and carried about 55% of the Cuban American vote (Mer- rill & McCarthy, 2020). In retrospect, this ad illustrates key gaps in the policing of misinformation by Google, the par- ent company of YouTube; while nominally prohibiting false claims in advertising, Google rarely takes down po- litical ads (Google, 2020). However, YouTube takes a more nuanced approach to this issue. Per the “How does YouTube combat misinforma- tion” page, YouTube engages a range of practices: remov- ing content in violation of policies, raising up authoritative sources for news and information, and reducing recom- mendations of borderline content and harmful misinfor- mation (YouTube, 2020). In an attempt to satisfy the vol- untary action conditions of section 230, YouTube algorithms go as far as to not proactively recommend con- tent that comes close to violating Community Guidelines; thus, taking action to limit the spread of disinformation. To determine what qualifies as “harmful” misinformation, YouTube states that they consult “external human evalua- tors” and experts to determine whether content promotes unsubstantiated conspiracy theories, or inaccurate infor- mation – reviewing “hundreds of thousands of hours” of video each day (YouTube, 2020). Unarguably, YouTube explicitly makes commitments and takes action to remove objectionable content, within their rights as an ICSP. De- spite these articulated commitments, no capacity exists in which YouTube can be held accountable for defaulting on their expressed promises on deterring misinformation. Circling back to the YouTube ad; if the Associated Press fact-check does not qualify as an external expert evaluation on unsubstantiated conspiracy theories, what does? Negat- ing their own policy, company spokeswoman Charlotte Smith claims that the campaign ad “does not violate our policies…and we’re not going to attempt to adjudicate ev- ery claim or counterclaim” on the platform (Merrill & Mc- Carthy, 2020). Albeit not a direct violation of section 230, this event underscores the urgent need for companies to be held accountable for what spreads on their platforms, rather than merely relying on an honor system for e4cient platform governance. Given people watch over a billion hours of video any given day on YouTube, financial incen- tives could be a driving force behind the platform falling short of enforcing its policies (“Press - Youtube”, 2020). In 2020 alone, the Trump campaign spent $106 million – $37.2 million in the last month of the campaign – on both YouTube and Google search ads alone (Thompson, 2020). Incalculable, however, is the e3ect running these ads had on disenfranchising and dissuading potential Latin Amer- ican voters in the state of Florida. With tens of millions of dollars in prospective ad revenue on the line, the clauses currently set forth under section 230 do little to counter any perverse commercial incentives at companies like YouTube. Thus, although YouTube makes voluntary ef- forts to publish their commitments like Twitter and Face- book, the commitments themselves appear vague and un- corroborated by decisive action. 3.4 Google In 2019, Google was heavily criticized in a 2019 Senate Investigation for its role in spreading misinformation during the 2016 presidential election (Wells, 2019). In October of that same year, a Senate Intelligence Committee released a report describing just how fast misinformation can spread on the platform; just days after the 2016 election, a false news story claiming that Trump had won the popular vote ranked higher on Google than accurate stories (Cassady, 2020). Although a mixed bag in its history with politics, Google’s present-day initiatives to combat the spread of misinformation are centered around fact-checking. Specifically, the “Google News Initiative” (GNI) is support- ing “First Draft,” a nonprofit protecting communities from harmful misinformation, with a $6.5 million grant (Mantzarlis, 2020). With an immediate focus on coronavi- rus information, the funding goes directly to fact checkers fighting misinformation around the world. In response to misinformation circulating about COVID-19, Google is supporting journalism fellowships at Stanford University to “collate data from around the world and help journalists tell data-driven stories that have [an] impact in their com- munities” (Mantzarlis, 2020). Additionally, Google says it 24 !"!#!$%& : : &'()$* 2021 has made Google Trends data available in “localized pages with embeddable visualizations,” in an e3ort to help jour- nalists understand and meet people’s information needs around the world. However, despite the promise of these initiatives, they appear reactionary. In March of 2020, the Global Disinformation Index found that 1,400 sites spreading COVID-19 misinformation earned a collective $76 million in ad revenue, with the vast majority coming from Google (Figure 2). Thus, Google seemingly under- stands the dilemma: they can take down a conspiracy theo- rist’s YouTube channel for making false claims about CO- VID-19; however, continuing to run ads on the website negates the entire purpose. Voluntary action superseded by profit in no way advances the e3ort on combatting the spread of misinformation; hence, there must be a disin- centive to the “1 step forward, 2 steps back” approach.3 4. PROPOSALS AND IMPLICATIONS Passing judgment on the regulation of free speech entails a balancing act, as Zuckerberg calls it, “‘of competing eq- uities’” (Testimony of Mark Zuckerberg Facebook, Inc., 2020). In some cases, the right thing to do from a safety or security viewpoint is not the most ideal for privacy or free expression; thus, there will always be somewhat of a trad- eo3 between disappointing people and promoting the common good.4 Deciding upon such tradeo3s is anything but straightfor- ward and passing down this moderation to individual companies raises much ambiguity surrounding the en- forcement of section 230. Given the relative ease of switch- ing between platforms, augmentation of the statute is of the uttermost importance – working with technology com- panies, rather than against them. 4.1 Statutory recommendations First, Congress should create a statutory exception in sec- tion 230(c) that permits civil liability against any ICSP’s complicity in the spread and circulation of disinformation. Not uncommon, last year lawmakers passed a bill that added a new exception to section 230 that positions plat- forms to be liable for any third-party content that facilitates sex tra4cking (Romano, 2018). In addition to modifications in this clause, section 230(c) must unambiguously specify what constitutes “any volun- tary action taken in good faith” to provide clear guidance to all ICSPs on the explicit steps they must take (Stratton Oakmont v. Prodigy Services Co., 2018). Parallel to Sloss (2020) in his push to hold companies ac- countable for complicity in mass genocide, a statutory ex- ception that withdraws civil liability could also work in internalizing the negative externalities of disinformation. Under section 230(c), such an exception to liability would arise in the case that a company fails to prevent transmis- sion of inaccurate information if that information: a) would be understood by ordinary readers as incitement or inducement to mislead or suppress people of factual infor- mation and the written law; and b) there is a significant risk that any recipient of the inaccurate information is ex- posed to harm. The statute should also include a duty for companies to remove such content within the 24-hour time frame since the content’s inception. Nevertheless, the statute should continue to preserve immunity from civil liability for any company that makes a reasonable, bona fide e3ort to comply with content removal, but is unable to do so for a justifiable reason. To do so, however, such bona fide e3orts must be clearly defined in section 230(c). Between Facebook and Twitter alone, the approaches to taking preventative steps in combatting misinformation are noticeably di3erent. On the one hand, flagging viola- tions and linking them to factual journalism raises con- cerns of anti-conservative social media bias; however, there is currently no evidence to support this exists (In- gram, 2019). On the other hand, allowing users to simply decide for themselves what to read does little by means of taking tangible, good faith steps in countering false infor- mation. Ideally, it is the work of journalists to hold those in power accountable and to provide context to readers; however, given the scale of platforms like Twitter and Facebook, the clauses in section 230 must recognize the need to take action. Thus, to define what proper good faith action is in practice, the legislation should mandate as- pects of both the Twitter and Facebook approaches in out- lining what such voluntary steps look like for all ICSPs. The approach should: require companies to take action to "There will always be somewhat of a tradeo5 between disappointing people and promoting the common good." 25 !"# $%&%!'( )'*$#+%, limit the ability of disinformation to go viral through algo- rithmic modifications, flag inaccurate information linked to third-party fact checks, and allow content to be seen such that people can debate it, support it, refute it, or dis- sect it, and do so publicly.5 4.2 Consequential incentives An exception to section 230 that withdraws liability from companies that perpetuate disinformation could have a preventative e3ect by providing a positive incentive for companies. Typically, the mens rea for civil liability is a neg- ligence standard, rather than specific intent. Such a negli- gence standard would incentivize companies to adopt sen- sible measures to inhibit third parties from violating their community guidelines and policies on misinformation. In fact, given the global reach of major platforms like Twitter, Facebook, and Google, these companies arguably have greater power to prevent negative externalities to society than do some national governments. One could argue that ICSPs do not need additional incen- tives to limit users from posting inaccurate information because they have already taken steps to address the prob- lem. For example, after hearing about violence in Myan- mar, Facebook removed accounts associated with the Myanmar military and hired Burmese language speakers to monitor content (Ellis-Peterson, 2018). However, Face- book did not take such actions until after users in Myan- mar already used the platform to spark genocidal violence (Stecklow, 2018). Thus, a federal statute subjecting compa- nies to potential civil liability would incentivize compa- nies to act proactively to prevent harms before they occur, rather than issuing reactionary responses. To avoid liabili- ty, companies would need, among other things, to hire teams to specifically monitor objectionable content; hence, the liability risk provides a financial incentive for compa- nies to incur that cost to avoid future adverse judgments. Without such an incentive, ICSPs would neglect to incur the costs of risk prevention until after it is too late. Additionally, we can leverage aspects of behavioral theory in order to “nudge” platform users into taking their own steps in combatting the spread of disinformation.6 In the introduction to their book Nudge, Thaler and Sunstein present a nudge as “any aspect of choice architecture that alters people’s behavior in a predictable way without for- bidding any Options” (Thaler & Sunstein, 2008). Such an intervention closely mirrors what Twitter already does: flagging tweets in violation of its community guidelines, nudging users to follow the link to accurate information. As Thaler and Sunstein put it, “Putting the fruit at eye level counts as a nudge. Banning junk food does not” (Thaler & Sunstein, 2008). Twitter, through flagging, plac- es accurate information at eye level, rather than banning it altogether. Along with linking accurate information, Twit- ter can also leverage advancements in machine learning in order to curate feed algorithms in such a way that pres- ents information in an objective, unbiased manner. In to- tality, such behavioral nudges demand more in the way of freedom of choice and less so in the way of government constraint and compulsion. 4.3 “Revoke 230” is not the solution As recently as December 2020, President Trump threat- ened to veto an annual defense bill unless Congress re- voked section 230 of the CDA (Brandom, 2020). Just one year prior, in December of 2019, now president-elect Joe Biden, in an interview with The New York Times, said “‘sec- tion 230 should be revoked, immediately should be re- voked, number one’” (The New York Times Editorial Board, 2020). Given the Communications Decency Act is now over 20 years old, lawmakers are trying to amend the statute to best align with present-day needs; simply put, all-out revoking section 230 is not the solution. Above all, section 230 serves one main purpose: protecting ICSPs from the liability for the speech of others on their platforms. If it were to be revoked, former Twitter VP Crowell suggests doing so “creates a dilemma for the com- panies because they would have to over censor and start taking down anything that might remotely run a liability risk,” noting it would cause a ripple e3ect in the timeli- ness of the internet (Tempera, 2020). Thus, before any content could be posted on an ICSP, it would essentially "Revise, revoke, repeal — whatever ‘R’ word you choose — the debate over the future of section 230 boils down to one, simple idea: the law exists to protect people." 26 !"!#!$%& : : &'()$* 2021 have to be approved by lawyers, having the involuntary e3ect of making the largest companies more powerful. Larger companies, like Twitter and Facebook, have the fi- nancial means to a3ord legal teams; as Crowell puts it, “it would really create a mess” (Tempera, 2020). Section 230, albeit not perfect, is essential to allowing plat- forms to exist while still within their discretion to do some moderation. Protecting everything from Facebook and YouTube to the recently rebranded 8chan, the central point of section 230 is to provide platforms with the certainty that they can adopt specific, good faith moderation prac- tices that users deem necessary without fear of liability. Without this statute as it stands today, the entire internet environment would look vastly di3erent, some platforms shutting down and others stopping moderation altogether. 4.4 Empirical evidence for a negative societal externality “We’re not just fighting an epidemic; we’re fighting an info- demic” – Dr. Tedros Ghebreyesus, WHO Director-Gener- al70 Reforming section 230 to account for disinformation draws reasonable skepticism. Notably, Matthew Waxman, a professor at Columbia Law School and specialist on na- tional security law, states that “Disinformation is a broad category that’s tough to define, and resulting liability for its harms is very uncertain,” suggesting that there is some ambiguity to the negative externalities (Inside Cybersecu- rity, 2020). Waxman’s skepticism is mainly centered around disinformation as it relates to foreign campaigns; however, domestic disinformation “campaigns” pose real, calculable e3ects on the health and safety of U.S. citizens. Bursztyn et al. (2020) presents the e3ects of news cover- age of COVID-19 by the two most widely viewed cable news shows in the United States, Hannity and Tucker Carl- son Tonight, on viewers’ behavior and downstream health outcomes (Bursztyn, 2020). Carlson warned viewers about the severity of COVID-19 from early February, while Hannity originally dismissed the associated risks before incrementally adjusting his position in late February. Via constructed epidemiological models, as exhibited in Ap- pendices C and D, di3erent e3ect sizes provide empirical evidence that misinformation is an important mechanism driving the observed e3ects on increased cases and deaths. Given notable, inherent externalities in a pandemic, mis- information may have detrimental e3ects far beyond those on viewers themselves by a3ecting disease transmission trajectories in the broader population (Bursztyn, 2020). Although confined to television networks, this study sheds light on the potential e3ects media platforms have by dis- seminating misinformation, as exacerbated during a pan- demic. Hence, misinformation on mass media of any kind can and does have empirically calculable consequenc- es to society at large. 5. CONCLUSION Revise, revoke, repeal – whatever ‘R’ word you choose – the debate over the future of section 230 boils down to one, simple idea: the law exists to protect people. How can we, as lawmakers and policy drafters, orient the clauses of sec- tion 230 such that the statute exists to protect all people? Looking forward, Congress must: a) enact a statutory ex- emption in section 230(c) that permits civil liability for any platform’s complicity in the spread or harboring of disinformation; and b) unambiguously define and recom- mend good faith steps all ICSPs can put into practice. Integral to the main objectives of a section 230 modifica- tion is to ensure more reliable and truthful content ap- pears, spreads, and lives on online platforms. It would be wrong to assume this is solely a debate over the future of free speech; thus, pleas to revoke section 230 in its totality are both unsubstantiated and grounded in misaligned pri- orities. Platforms like Facebook and Twitter, for example, are aggressively taking action to suppress the spread of disinformation; however, they are plagued by the subjec- tivity of their own policy decisions. Hence, there exists a deep desire for heightened collaboration and information sharing between the tech and social media industries and the U.S. government. In the very near future, the path of least resistance in determining how the internet should be governed is to solve these tensions and ambiguities to- gether as a society, in a way people feel is both reasonable and equitable. To do so, lawmakers must continue to invite input from technology companies, citing their individual moderation approaches, as well as input from the public in order to make progressive strides in the forthcoming year. 27 !"# $%&%!'( )'*$#+%, -2..*!%!'"6+, 6..!'$/7 Appendix A Figure 1: May 26, 2020 tweets made by Donald J. Trump Notes: Pictured above is the first instance of Twitter flag- ging tweets made by the President of the United States. Marked with the label “Get the facts about mail-in ballots,” users, upon clicking, are taken to a “Twitter Moments” page, which links articles and tweets from journalists and experts that debunk misinformation claims. Brad Parscale, the Trump Campaign manager, labeled this action as a bla- tant display of Twitter’s political bias.75 Appendix B Figure 2: Share of ad revenues generated from COVID-19 disinformation sites Notes: The top three companies shown (Google, OpenX, and Amazon) generate 95% of the ad revenues to the site in the sample. Google ad services alone delivers $3 out of every $4 that these sites earn in ad revenues. Important to note: for Google, the ad shares and revenues for Google AdSense/AdX and Google DFP have been combined. Appendix C Figure 3: Reduced-form and 2SLS estimates of e3ect of dif- ferential viewership on cases and deaths (extended) Notes: Figure 3 displays day-by-day reduced form (Panel A) and 2SLS (Panel B) estimates on log one plus cases and log one plus deaths. Panela A portrays day-by-day e3ects of the instrument, FoxShared, on log deaths and log cases, conditional on state fixed e3ects and a large set of con- trols: Fox News’ and MSNBC’s share of cable in January 2018, Fox News’ share of television in January 2020, the population density of the country, the log of the country’s 1)%#. ) 1)%#. 0 28 !"!#!$%& : : &'()$* 2021 total population, the number of predicted TV’s turned to non-Fox channels during Hannity, Tucker Carlson To- night, and The Ingraham Angle, the population-weighted latitude and longitude, the percent in the country living in rural areas, the percent white, Hispanic, and black, the percent over the age of sixty-five, the share of men and women lacking high school degrees, the share of men and women lacking college degrees, the fraction of the popula- tion lacking health insurance, an age-adjusted measure of the average physical health in the county from 2018, the percent under the federal poverty line, log median house- hold income, the unemployment rate, the 2016 Republi- can vote share, and the log total number of votes cast in 2016. Panel B presents day-by-day e3ects of the standard- ized di3erence in viewership of Hannity vs. Tucker Carl- son Tonight, instrumented by FoxShared and controlling for state fixed e3ects and the same set of covariates as in Panel A. Standard errors are clustered at the DMA level and report 95 percent confidence intervals. Appendix D Figure 4: Implied COVID-19 curves Notes: Panel A of Figure 4 plots, in black, the logarithm of (one plus the) mean number of cases in each day across all counties. In gray, the figure plots the implied counterfac- tual values (based on the 2SLS estimates) for a county with one standard deviation higher viewership di3erence between Hannity and Tucker Carlson Tonight. Panel B replicates Panel A, taking log one plus deaths as the out- come rather than log one plus cases. This is presented with 95 percent confidence intervals on the counterfactual estimates. Standard errors are clustered at the DMA level. 1)%#. ) 1)%#. 0 29 !"# $%&%!'( )'*$#+%, !'$'&"!- 1. Here, “e!ciency” denotes the completion of the core objectives: promoting further development, preserving the free market of ideas, allowing for innovation concurrent with user control, and to do so in a manageable, secure way. But, in a way least reminiscent of George Orwell’s “Big Brother” in 1984, yet still prevents the negative exter- nalities of disinformation. 2. Important to note, commercial entities should not have the role of determining what constitutes fact over falsehood online. Linking factual articles to tweets gives rise to a multitude of ethical consid- erations regarding information and perspective biases, beyond the scope of this paper. 3. Evaluating platforms like Reddit and Yelp, for example, are key for future study on this issue, yet are outside the scope of this paper. 4. Again, ethical considerations arise when the onus falls upon companies, or any governing body, to determine the “right” thing to do. In a way, defaulting to the indivudual companies on platform moderation serves to limit the role and presence of government in our everyday lives. 5. For possible future study, it would be of importance to evaluate the various court interpretations of the term “publisher,” as it either relates or di"ers from the term “platform” in both legal provisions and court decisions. 6. #e nuances behind both qualitative and empirical behavioral theory are crucial to examine in the context of this topic; however, they exist outside the scope of this paper. +!(!+!'8!- “An Update on Our Political Ads Policy,” November 20, 2019. https://blog.google/technology/ads/update-our-po- litical-ads-policy/. Ardia, David. “Free Speech Savior or Shield for Scoun- drels: An Empirical Study of Intermediary Immunity un- der Section 230 of the Communications Decency Act.” Loyola L.A. Law Review 43, no. 2 (2010): 373,410. The New York Times Editorial Board. “Opinion | Joe Biden Says Age Is Just a Number.” The New York Times, January 17, 2020, sec. Opinion. https://www.nytimes.com/interac- tive/2020/01/17/opinion/joe-biden-nytimes interview. html. Brandom, Russell. “Trump Calls for Last-Minute 230 Re- peal as Part of Defense Spending Bill.” The Verge, Decem- ber 2, 2020. https://www.theverge. com/2020/12/2/22037118/trump-section-230-repeal- ndaa-rider facebook-twitter-moderation. Bursztyn, Leonardo. “Misinformation During a Pandem- ic.” The Becker Friedman Institute of the University of Chicago, June 17, 2020. Cannon, Robert. “The Legislative History of Senator Ex- on’s Communications Decency Act: Regulating Barbar- ians on the Information Superhighway.” Federal Commu- nications Law Journal 49, no. 1 (November 1996): 52–94. Cassady, Daniel. “Google Announces Measures to Crack Down On Misinformation Ahead Of Presidential Elec- tion.” Forbes. Accessed December 16, 2020. https://www. forbes.com/sites/danielcassady/2020/07/31/google-an- nounces-measures to-crack-down-on-misinformation- ahead-of-presidential-election/. Twitter. “Civic Integrity Policy,” October 2020. https:// help.twitter.com/en/rules-and policies/election-integrity- policy. Communications Decency Act, 47 U.S. § 230 (2018). Facebook. “Community Standards,” 2020. https://www. facebook.com/communitystandards/false_news. Inside Cybersecurity. “FCC Seeks Input on Regulatory Re- view of ‘Section 230’ Liability Protection; Issues Touch on Countering Disinformation,” August 4, 2020. https://in- sidecybersecurity.com/share/11499. Ghebreyesus, Tedros. “Speech at the Munich Security Conference.” Munich, Germany, 2020. Lotty, Alexandra. “Apps Too: Modifying Interactive Com- puter Service Provider Immunity under Section 230 of the Communications Decency Act in the Wake of ‘Me Too.’” Southern California Law Review 93, no. 4 (Spring 2020): 855–921. Mantzarlis. “COVID-19: $6.5 Million to Help Fight Coro- navirus Misinformation.” Google, April 2, 2020. https:// blog.google/outreach-initiatives/google-news-initiative/ covid-19- 65-million-help-fight-coronavirus-misinforma- tion/. Merrill, Jeremy, and Ryan McCarthy. “Trump Won Florida After Running a False Ad Tying Biden to Venezuelan So- cialists.” Electionland, November 12, 2020. 30 !"!#!$%& : : &'()$* 2021 “Press - YouTube.” Accessed December 16, 2020. https:// www.youtube.com/intl/en GB/about/press/. Global Disinformation Index. “Research – GDI.” Accessed December 16, 2020. https://disinformationindex.org/re- search/. Sloss, David. “Section 230 and the Duty to Prevent Mass Atrocities.” Case Western Reserve Journal of International Law 52, no. 1/2 (Spring 2020): 199–212. Stratton Oakmont, Inc v. Prodigy Services Co. (Supreme Court, Nassau County, New York 1995). Tempera, Jacqueline. “Behind the Scenes.” Boston College Magazine, Fall 2020. Thaler, Richard, and Cass Sunstein. Nudge: Improving Decisions About Health, Wealth, and Happiness. Yale University Press, 2008. Thompson, Alex. “Trump Deploys YouTube as His Secret Weapon in 2020.” Politico, September 6, 2020. https:// www.politico.com/news/2020/09/06/trumpyoutube election-comeback-408576. Volz, Georgia Wells, Robert McMillan and Dustin. “Senate Faults Google, Other Tech Giants for Role in Russian Elec- tion Meddling.” Wall Street Journal, October 8, 2019, sec. Politics. https://www.wsj.com/articles/senate-report- seeks-to-curb-possible russian-influence-in-2020-elec- tion-11570556521. Working to Stop Misinformation and False News | Face- book Media. “Working to Stop Misinformation and False News,” 2020. https://www.facebook.com/formedia/blog/ working-to-stop-misinformation-and-false news. YouTube - How YouTube Works. “YouTube Misinforma- tion - How YouTube Works.” Accessed December 16, 2020. https://www.youtube.com/howyoutubeworks/our- commitments/fighting misinformation/. Zuckerberg, Mark. “Testimony of Mark Zuckerberg Face- book, Inc.” Testimony presented at the Hearing Before the United States Senate Committee on the Judiciary, No- vember 17, 2020. https://www.judiciary.senate.gov/imo/ media/doc/Zuckerberg%20Testimony.pdf. Pulford, J., Hetzel, M. W., Bryant, M., Siba, P. M., & Muel- ler, I. (2011). Reported reasons for not using a mosquito net when one is available: a review of the published litera- ture. Malaria journal, 10(1), 1-10. Gawande, A. [@Atul_Gawande]. (2020 November 16) This is the cost of disinformation. It’s not only about the reality of deaths but also health & economic damage. MI stats on COVID hospital patients: Typical survivor age = 50-70. At 60 days, 40% either lost their job or were still too sick to work. http://acpjournals.org/doi/10.7326/M2… [Tweet]. Twitter. https://twitter.com/atul_gawande/status/132840 9592959823875?lang=en Orth, J. [@JodiOrth]. (2020 November 14) I have a night o3 from the hospital. As I’m on my couch with my dog I can’t help but think of the Covid patients the last few days. The ones that stick out are those who still don’t believe the virus is real. The ones who scream at you for a magic med- icine and that Joe Biden is [Tweet]. Twitter. https://twitter. com/JodiOrth/status/1327771329555292162 Halon, Y. (2020, May 28). Zuckerberg knocks twitter for fact-checking Trump, says private companies shouldn’t be ‘The arbiter of truth’. Fox News. Retrieved 2020, from https://www.foxnews.com/media/facebook-mark-zucker- berg-twitter-fact-checking-trump Donald J. Trump. (2020, October 20). Salga a Votar [Vid- eo]. YouTube. https://www.youtube.com/ watch?v=BtA1i0ujGrQ About. First Draft. (2022, January 12). Retrieved 2020, from https://firstdraftnews.org/about/ Romano, A. (2018, April 13). A new law intended to curb sex tra4cking threatens the future of the internet as we know it. Vox. Retrieved 2020, from https://www.vox.com/ culture/2018/4/13/17172762/fosta-sesta-backpage-230-in- ternet-freedom Ingram, M. (2019, August 8). The myth of social media anti-conservative bias refuses to die. Columbia Journalism Review. Retrieved 2020, from https://www.cjr.org/the_ media_today/platform-bias.php Ellis-Peterson, H. (2018, August 27). Facebook removes accounts associated with Myanmar military. The Guard- ian. Retrieved 2020, from https://www.theguardian.com/ 31 !"# $%&%!'( )'*$#+%, technology/2018/aug/27/facebook-removes-accounts- myanmar-military-un-report-genocide-rohingya Stecklow, S. (2018, August 15). Special reports from Re- uters journalists around the world. Reuters. Retrieved 2020, from https://www.reuters.com/investigates/spe- cial-report/myanmar-facebook-hate/ 25 32 !"!#!$%& : : &'()$* 2021 */-" &( 6+"9&+: !"# $%!!&# '( !"# )'**# +'*,+: !"# )!'-. '( /,&&,%* *+(%01#%2 © Creative Centenaries. (n.d.). The Battle Of The Somme comic: The story of William McFadzean | CCEA. CEA. Retrieved January 19, 2022, from https://ccea.org.uk/learning-resources/understanding-1916/creative-centena- ries/battle-somme-comic-story-william 13 15 28 28 28 28 29 29 (,34-# 1 © originally tweeted by User:@realDonaldTrump (account suspended) on 26 May, 2020. #%)!#- 5-'+&%*%!,'2 '( 1916 © originally uploaded to the English Wikipedia by w:User:Jtdirl (https://commons.wikimedia.org/wiki/ File:Easter_Proclamation_of_1916.png), „Easter Proclamation of 1916“, marked as public domain, more details on Wikimedia Commons: https://commons.wikimedia.org/wiki/Template:PD-old (,34-# 2 © GDI report: Ad-funded COVID-19 Disinformation: Money, Brands, and Tech (https://disinformationindex. org/wp-content/uploads/2020/07/GDI_Ad-funded-COVID-19-Disinformation-1.pdf), „Share of ad revenues generated by COVID Disinfo Sites“. (,34-# 3, 5%2#& % © originally published in Bursztyn, Rao, Roth, Yanagaziwa-Drott: Misinformation During a Pandemic (https:// www.nber.org/papers/w27417), „RF estimates of e3ect of di3erential viewership on cases and deaths (extend- ed)“. (,34-# 3, 5%2#& $ © originally published in Bursztyn, Rao, Roth, Yanagaziwa-Drott: Misinformation During a Pandemic (https:// www.nber.org/papers/w27417), „2SLS estimates of e3ect of di3erential viewership on cases and deaths (ex- tended)“. (,34-# 4, 5%2#& % © originally published in Bursztyn, Rao, Roth, Yanagaziwa-Drott: Misinformation During a Pandemic (https:// www.nber.org/papers/w27417), „Implied COVID-19 Curves, Panel A: Estimates on Cases“. (,34-# 4, 5%2#& $ © originally published in Bursztyn, Rao, Roth, Yanagaziwa-Drott: Misinformation During a Pandemic (https:// www.nber.org/papers/w27417), „Implied COVID-19 Curves, Panel B: Estimates on Deaths“. 35 38 43 45 %-!,(,+,%& ).2%5)#) $%)#0 '2 (#--'-#&#+!-,+ !422#& 642+!,'2) © Mike Davies Intel Labs, Intel Corporation et. al: (https://commons.wikimedia.org/wiki/File:Artificial_syn- apses_based_on_FTJs.png), https://creativecommons.org/licenses/by-sa/4.0/legalcode !%$&# ,,, ,2(#-#2+# 5'/#- %20 5'/#- 4!,&,1%!,'2 © Tasbolat Taunyazov, Weicong Sng, Hian Hian See, Brian Lim, Jethro Kuan, Abdul Fatir Ansari, Benjamin C.K. Tee, and Harold Soh: (http://www.roboticsproceedings.org/rss16/p020.pdf), „Table III: Inference Speed and Power Utilization“ $&%+"# &%*'2! %! "#+&%, *'2!%2%, /,!" "#- )!40#2!), 1893 © Unknown authorUnknown author (https://commons.wikimedia.org/wiki/File:Blanche_Lamont_with_stu- dents.jpg), „Blanche Lamont with students“, marked as public domain, more details on Wikimedia Commons: https://commons.wikimedia.org/wiki/Template:PD-US !"#'0'-# 04--%2!’) *43)"'! © originally published in Report of the trial of William Henry Theodore Durrant (https://archive.org/details/ reportoftrialofw00durr/page/10/mode/2up), marked as public domain. 60 61 61 61 3-%5" ,&&4)!-%!,23 !"# )",(!) ,2 5-,+# %20 74%2!,!. )455&. © originally published in 31.11 E4ciency and Deadweight Loss (https://saylordotorg.github.io/text_economics- theory-through-applications/s35-11-e4ciency-and-deadweight-loss.html), „Figure 31.8 Deadweight Loss“ 3-%5") )"'/,23 !"# 24*$#- %20 8%&4# '( *#-3#-) %20 %+74,),!,'2) © originally published in Statista: Mergers and acquisitions in the United States (https://www.statista.com/ study/65232/mergers-and-acquisitions-in-the-united-states). 3-%5") )"'/,23 3&'$%& 74#-. )"%-# $. )##&+! )#%-+" #23,2#) © originally published in Statista: Global market share of search engines 2010-2022 (https://www.statista.com/ statistics/216573/worldwide-market-share-of-search-engines). 3-%5") )"'/,23 0'*#)!,+ 4) )#%-+" 74#-. )"%-# $. )#&#+! )#%-+" #23,2#) © originally published in Statista: Market share of search engines in the United States 2008-2022 (https:// www.statista.com/statistics/267161/market-share-of-search-engines-in-the-united-states). */-" &( 6+"9&+: +'-5'-%!# !#+"2'&'3. ,2('-*%!,'2 )#-8,+#) © unpublished tabulation from Corptech Database 62 62 71 75 78 80 82 (,34-# 1 © Samadi, Moradi, Moradinazar, Mostafai, Pasdar: Dietary pattern in relation to the risk of Alzheimer’s disease: a systematic review (https://link.springer.com/article/10.1007/s10072-019-03976-3), „Figure 2: Healthy and unhealthy diet in relation to AD.“ !"# $-#%90'/2 '( '5,2,'2 -#3%-0,23 $-#%9,23 45 ‘$,3 !#+"’ © Statista : Majority of Americans in Favor of Breaking up Big Tech (https://www.statista.com/chart/19440/ survey-responses-breaking-up-big-tech) 5'!#2!,%& ),32%!'-,#), ),32%!'-,#), %20 5%-!,#) !' !"# %,,$ © L.tak (https://commons.wikimedia.org/wiki/File:Asian_Infrastructure_Investment_Bank_map.svg), https://creativecommons.org/licenses/by-sa/4.0/legalcode +",2%’) '23',23 !#--,!'-,%& 0,)54!#)] © O4ce of the Secretary of Defense (https://commons.wikimedia.org/wiki/File:China’s_Disputed_Territories. PNG), „China’s Disputed Territories“, marked as public domain, more details on Wikimedia Commons: https:// commons.wikimedia.org/wiki/Template:PD-US 4) 0#5%-!*#2! '( 64)!,+# %*'23 '!"#-) %22'42+#0 23 +-,*,2%& +"%-3#) © U.S. Department of Justice (https://commons.wikimedia.org/wiki/File:U.S._Commerce_Secretary_Wilbur_ Ross,_Acting_Attorney_General_Matthew_Whitaker,_Homeland_Security_Secretary_Kirstjen_Nielsen,_FBI_ Director_Christopher_Wray_Announces_23_Criminal_Charges_Against_China’s_Huawei_&_Wanzhou_ Meng.jpg), „U.S. Commerce Secretary Wilbur Ross, Acting Attorney General Matthew Whitaker, Homeland Security Secretary Kirstjen Nielsen, FBI Director Christopher Wray Announces 23 Criminal Charges Against China’s Huawei & Wanzhou Meng“, marked as public domain, more details on Wikimedia Commons: https:// commons.wikimedia.org/wiki/Template:PD-US %*#-,+% 3&'$%& !-%0# 0#(,+,! © Statista: U.S. Trade Deficit in Goods Reaches Record High, (https://www.statista.com/chart/17281/us-trade- balance). 98 +"%-&#) +'43"&,2, &#%0#- '( !"# +"-,)!,%2 (-'2! © Craine, Detroit (https://commons.wikimedia.org/wiki/File:CharlesCouglinCraineDetroitPortrait.jpg), „CharlesCouglinCraineDetroitPortrait“, marked as public domain, more details on Wikimedia Commons: https://commons.wikimedia.org/wiki/Template:PD-US 100 3#-*%2 %*#-,+%2 $420 -%&&., 2#/ .'-9, *%0,)'2 )74%-# 3%-#2, (#$ 1939 © Department of Defense. Department of the Army. O4ce of the Chief Signal O4cer. (09/18/1947 - 02/28/1964); ARC Identifier 36068 / Local Identifier 111-OF-2 1943 (https://commons.wikimedia.org/wiki/ File:Nazi_Strike_2.jpg), „Nazi Strike 2“, marked as public domain, more details on Wikimedia Commons: https://commons.wikimedia.org/wiki/Template:PD-US