1 00:00:01,000 --> 00:00:06,001 I'm Kathy Kiley. I am an editor at Bloomberg Politics. I am 2 00:00:06,001 --> 00:00:08,000 also a member of the National 3 00:00:08,000 --> 00:00:13,000 Press Club, and I am a director of the National Press Club's Institute 4 00:00:13,000 --> 00:00:14,001 for Journalism. Now, 5 00:00:14,001 --> 00:00:17,000 I think a lot of people I know for our TV audience 6 00:00:17,000 --> 00:00:19,000 have heard of the National Press 7 00:00:19,000 --> 00:00:23,001 Club. You might think of it as a place where you hear very interesting luncheon 8 00:00:23,001 --> 00:00:29,000 speakers or a place where journalists gather after deadline to exercise our 9 00:00:29,000 --> 00:00:35,000 elbows. And those are both very venerable traditions that I long may they reign, 10 00:00:35,001 --> 00:00:41,001 but it's also a place where journalists gather to help each other sharpen our 11 00:00:41,001 --> 00:00:44,001 skills and add to our toolboxes. 12 00:00:45,000 --> 00:00:50,000 And that is where the National Press Club's Institute for Journalism comes in. We 13 00:00:50,000 --> 00:00:56,000 are the nonprofit wing of the National Press Club, and we put on programs like 14 00:00:56,000 --> 00:01:01,000 this. We do depend on the kindness of friends and strangers, so when you're 15 00:01:01,000 --> 00:01:05,001 thinking about making a charitable donation, you might add us to our list. Even 16 00:01:05,001 --> 00:01:10,000 if you don't know a journalist or are not a journalist, you should care because 17 00:01:10,000 --> 00:01:15,000 we think that a free press supports a free society. 18 00:01:15,000 --> 00:01:22,000 And on that note, I'd really like to celebrate and share, for 19 00:01:22,000 --> 00:01:27,000 those of you who don't already know, the news that Jason Rezaian is headed back 20 00:01:27,000 --> 00:01:32,001 to the United States. And that is a goal that many people at the National Press 21 00:01:32,001 --> 00:01:38,001 Club helped to press for, and so we're very delighted, especially our 22 00:01:38,001 --> 00:01:43,001 Committee for the Freedom of the Press, which is also a part of the National 23 00:01:43,001 --> 00:01:47,000 Institute, the National Press Club's Institute for Journalism. 24 00:01:48,001 --> 00:01:54,000 Now today's program, which is part of a wonderful journalism education series 25 00:01:54,000 --> 00:01:59,001 that Julie Hsu, our director, puts together, is really interesting to me as a 26 00:01:59,001 --> 00:02:05,001 political junkie. It is, we live in a society, as I think we all know, where 27 00:02:05,001 --> 00:02:12,000 an awful lot of money is spent to influence the outcomes of elections. 28 00:02:13,000 --> 00:02:19,000 And much of that money goes to mass communications of various sorts. This project 29 00:02:19,000 --> 00:02:25,001 helps people to decode those messages and, in some cases, to 30 00:02:25,001 --> 00:02:31,001 trace the messages back to the messengers, which, as those of us in this room, 31 00:02:32,000 --> 00:02:36,001 this is a very distinguished panel of political journalists and investigators, 32 00:02:37,000 --> 00:02:43,000 and we all know that sometimes that's not so obvious. So this is also a wonderful 33 00:02:43,000 --> 00:02:49,000 collaboration of journalists and technologists, journalists 34 00:02:49,000 --> 00:02:56,000 and engineers, to help to combine talents and resources 35 00:02:56,000 --> 00:03:03,000 to take these messages and turn them into data. And when you can turn 36 00:03:03,000 --> 00:03:10,000 propaganda into data, you can sort it, you can manipulate it, 37 00:03:10,000 --> 00:03:15,001 and you can help people see patterns that they might not have otherwise seen. So 38 00:03:15,001 --> 00:03:21,000 this is really important work. It's very arduous work, and it's being done by 39 00:03:21,000 --> 00:03:25,000 some really, really smart people. And as a journalist, I'm really humbled by the 40 00:03:25,000 --> 00:03:30,000 technologists who have joined in this effort. So with that, I'll turn it over to 41 00:03:30,000 --> 00:03:36,000 Roger from the Internet Archive, which, like the National Press Club, does a lot 42 00:03:36,000 --> 00:03:40,000 more things than what it's best known for, which is probably the way back machine 43 00:03:40,000 --> 00:03:41,001 and putting the Grateful Dead online. 44 00:03:43,001 --> 00:03:49,001 [...] Ebert Thank you very much. Great to be here with you. And as Kathy said 45 00:03:49,001 --> 00:03:56,000 about Humble, we are librarians. We are technologists, and we are humbled to be 46 00:03:56,000 --> 00:04:01,000 here with you and our journalists and other investigative partners. 47 00:04:02,001 --> 00:04:06,001 We're at the Internet Archive. It's the largest public digital library in 48 00:04:06,001 --> 00:04:08,000 the world at archive.org. 49 00:04:08,001 --> 00:04:14,000 We archive the web and have billions of pages available to look back in time. We 50 00:04:14,000 --> 00:04:19,001 have software, we have books, and we have television. We're recording a great 51 00:04:19,001 --> 00:04:23,000 deal of television, foreign and domestic, and about three and a half years opened 52 00:04:23,000 --> 00:04:25,001 up a television news research library. 53 00:04:26,000 --> 00:04:31,000 We repurposed the closed captioning to use as a search index to allow people to 54 00:04:31,000 --> 00:04:37,000 compare and contrast and quote and borrow television news. In order to 55 00:04:37,000 --> 00:04:38,001 explore television news, 56 00:04:41,001 --> 00:04:46,000 we've opened up an additional element of television for 57 00:04:46,000 --> 00:04:48,000 inspection, and that are political ads. 58 00:04:49,000 --> 00:04:53,000 We do live in this era where there's an enormous amount of money being poured 59 00:04:53,000 --> 00:05:00,000 into political advertising. The public and journalists and scholars and civic 60 00:05:00,000 --> 00:05:05,000 organizations that serve the public having to make these hard choices have a hard 61 00:05:05,000 --> 00:05:11,000 time being able to understand and convey what the larger meaning is of this 62 00:05:11,000 --> 00:05:13,001 messaging and the money behind the messaging. 63 00:05:14,000 --> 00:05:20,000 And as librarians, we're hoping to provide some opportunity to treat media as 64 00:05:20,000 --> 00:05:27,000 data and study at scale and in the individual moments 65 00:05:27,000 --> 00:05:33,001 these kinds of media. In this election year, we're particularly amazed to 66 00:05:33,001 --> 00:05:39,001 be joining you all in being amazed at how so much of the equation of the 67 00:05:39,001 --> 00:05:44,000 relationship between paid media, political advertising, and earned media, the 68 00:05:44,000 --> 00:05:51,000 coverage that's given in all forms of media, has been upended. Where 69 00:05:51,000 --> 00:05:57,000 there's an enormous amount of ... Trump has gotten enormous amount of press 70 00:05:57,000 --> 00:06:02,000 attention without spending any money on political advertising, television 71 00:06:02,000 --> 00:06:08,000 political advertising. Now that he has in the last two and a half weeks, we can 72 00:06:08,000 --> 00:06:10,000 track those effects as well. 73 00:06:11,001 --> 00:06:18,001 So today, we're here to flip the switch, to open up politicaladdarchive.org. It 74 00:06:18,001 --> 00:06:24,001 is a project of the Internet Archive. We are tracking political ads in key early 75 00:06:24,001 --> 00:06:30,001 primary states and key markets in those states. We utilize audio fingerprinting 76 00:06:30,001 --> 00:06:37,000 to identify political ads and look for them throughout all of TV. 77 00:06:37,001 --> 00:06:42,000 We surface television news and you can see the context in which political ads 78 00:06:42,000 --> 00:06:49,000 occur in television news through our site. We also track political ads in 79 00:06:49,000 --> 00:06:55,001 entertainment and other programming. We don't surface that for inspection out of 80 00:06:55,001 --> 00:07:00,001 respect for the broadcasters, whereas we think news is very tightly coupled to 81 00:07:00,001 --> 00:07:04,000 open public interest access. 82 00:07:11,001 --> 00:07:17,000 This politicaladdarchive, we track individual ads. We 83 00:07:17,000 --> 00:07:21,001 give you information about the [... ] action, about who's the sponsor of the ad. 84 00:07:22,000 --> 00:07:28,000 We dynamically link to organizations like Open Secrets that does fantastic 85 00:07:28,000 --> 00:07:33,000 research on who's behind, who's behind, who's behind the political action 86 00:07:33,000 --> 00:07:39,000 committee and the various other flavors of sponsors who wrap, who's behind, who's 87 00:07:39,000 --> 00:07:42,000 behind in multiple layers of obfuscation. 88 00:07:42,001 --> 00:07:49,001 One of the reasons why we're using audio fingerprinting is 89 00:07:49,001 --> 00:07:55,000 because most, it seems, most of the political ads are not captioned. We did an 90 00:07:55,000 --> 00:07:59,000 experimental study in the Philadelphia Television Market area during the fall 91 00:07:59,000 --> 00:08:06,000 elections in 2014 and found 70% of the political ads there were not captioned, 92 00:08:06,000 --> 00:08:12,001 which is a curious byproduct of the old school media, the media mentality 93 00:08:12,001 --> 00:08:17,001 of people who are creating advertising because obviously there's no politician in 94 00:08:17,001 --> 00:08:21,001 the world who would say, look, I'm going to spend a lot of money on this, but I 95 00:08:21,001 --> 00:08:27,000 want to make sure to exclude those who have difficult time hearing. And we're 96 00:08:27,000 --> 00:08:31,000 going to exclude people in health clubs and in bars. Can we do that? Yeah, you 97 00:08:31,000 --> 00:08:36,000 can. So it's amazing that we've reached this juncture and there's a wonderful 98 00:08:36,000 --> 00:08:40,000 public policy opportunity for the parties to get together and go, you know, guys, 99 00:08:40,000 --> 00:08:44,001 we're going to make all our messaging accessible, but it's not. So we're using 100 00:08:44,001 --> 00:08:51,001 audio fingerprinting to detect these ads. It detects the ads as they are played 101 00:08:51,001 --> 00:08:58,001 in paid advertising slots. It also detects ads that are, that 102 00:08:58,001 --> 00:09:04,001 appear in news. And once again, we have this opportunity to provide tools for 103 00:09:04,001 --> 00:09:11,001 researchers to study this earned media where their messaging is 104 00:09:11,001 --> 00:09:15,001 inserted within the news program, or I mean, is part of the stories within news 105 00:09:15,001 --> 00:09:20,000 programs, which we're seeing a lot of, which is fairly exceptional. We're 106 00:09:20,000 --> 00:09:24,000 supported by the Knight Foundation, a Knight News Challenge, a collaboration 107 00:09:24,000 --> 00:09:29,000 between the Knight Foundation, the Democracy Fund, Rita Allen Foundation, and the 108 00:09:29,000 --> 00:09:35,001 Hewlett Foundation. The Democracy Fund has also funded outreach and training for 109 00:09:35,001 --> 00:09:42,000 journalists with our partners. And rather than going over in a great meal more 110 00:09:42,000 --> 00:09:48,000 detail than I want to do, I just wanted to say that we are, as I said at the 111 00:09:48,000 --> 00:09:53,001 outset, we're librarians and engineers working together to serve the public. So 112 00:09:53,001 --> 00:10:00,001 here today and ongoing, we need to hear from you all about what we can do to 113 00:10:00,001 --> 00:10:04,000 serve your interests and to serve your imaginations. Because what we're doing 114 00:10:04,000 --> 00:10:10,000 today, having flipped this switch on politicaladsarchive.org, is open a door, 115 00:10:10,000 --> 00:10:15,000 open a door to imagination and thoughtful inquiry and research. 116 00:10:16,001 --> 00:10:20,000 Thanks, Roger, for the overview. We're going to turn, we have a wonderful group 117 00:10:20,000 --> 00:10:24,001 of partners here who are working with us to make this project, give it a lot 118 00:10:24,001 --> 00:10:26,000 more depth and resonance. 119 00:10:26,001 --> 00:10:30,000 And they're all going to talk a little bit about what they're observing in the 120 00:10:30,000 --> 00:10:33,001 election and what we hope to document on the new website, which is 121 00:10:33,001 --> 00:10:37,000 politicaladsarchive.org, live today. 122 00:10:37,001 --> 00:10:43,000 And first up is Robert McGuire. He has the title of Political Nonprofit 123 00:10:43,000 --> 00:10:47,001 Investigator at the Center for Responsive Politics. The Center for Responsive 124 00:10:47,001 --> 00:10:52,000 Politics is a wonderful, wonderful resource, opensecrets.org, where they gather 125 00:10:52,000 --> 00:10:57,000 together all kinds of information about who's contributing to campaigns and 126 00:10:57,000 --> 00:11:00,001 investigate the connections between those groups. So Robert, if you'd like to 127 00:11:00,001 --> 00:11:03,000 talk a little bit about the work you're doing, that would be great. 128 00:11:04,000 --> 00:11:08,000 First of all, thank you so much to the Internet Archive for having us here today. 129 00:11:08,001 --> 00:11:14,000 This is really exciting. We are particularly excited about partnering with the 130 00:11:14,000 --> 00:11:18,001 Internet Archive on this because finding out who is running ads, when they're 131 00:11:18,001 --> 00:11:25,000 running ads, how often they're running ads is such a crucial part, not only of, 132 00:11:25,000 --> 00:11:30,000 you know, understanding the election as a whole, but in particular the area that 133 00:11:30,000 --> 00:11:35,000 I work in, which is dark money. So the Center for Responsive Politics, for those 134 00:11:35,000 --> 00:11:41,000 who aren't familiar, we're a nonpartisan, partisan nonprofit organization, who we 135 00:11:41,000 --> 00:11:47,000 deal mostly in data, so lobbying data, campaign finance data. I'm going to focus 136 00:11:47,000 --> 00:11:54,000 specifically on dark money, and this is the way we refer to group 137 00:11:54,000 --> 00:11:58,001 501c organizations, particularly social welfare organizations and trade 138 00:11:58,001 --> 00:12:04,000 associations, spending an increasing amount of money in elections without 139 00:12:04,000 --> 00:12:05,001 disclosing the sources of the funding. 140 00:12:07,000 --> 00:12:13,000 So for example, in 2006 we had about $5 million, $5.5 million spent by these 141 00:12:13,000 --> 00:12:17,001 kinds of organizations that grew to more than $300 million in the last 142 00:12:17,001 --> 00:12:23,000 presidential election. And then on top of that you have an additional hundred 143 00:12:23,000 --> 00:12:29,000 million or so, hundreds of millions, that is spent outside of windows that have 144 00:12:29,000 --> 00:12:34,001 to be reported to the Federal Election Commission. And that is where this kind of 145 00:12:34,001 --> 00:12:40,000 information is absolutely crucial. To that end, we work with the Wesley Media 146 00:12:40,000 --> 00:12:46,001 Project on reports about who's spending when, how many ads they're running. We 147 00:12:46,001 --> 00:12:52,000 work with the Internet Archive to try and do the same thing. We are processing 148 00:12:52,000 --> 00:12:58,001 Federal Communications Commission data to process and link 149 00:12:58,001 --> 00:13:05,001 filings up to the particular groups, standardize those groups so that people can, 150 00:13:06,000 --> 00:13:09,000 you know, we get calls all the time saying, this group is running ads on my TV, 151 00:13:10,000 --> 00:13:13,001 they're telling me to vote for somebody, or they're insinuating that I should 152 00:13:13,001 --> 00:13:16,001 vote for someone. But I don't see them on your 153 00:13:16,001 --> 00:13:19,000 website. We're trying to make that possible. 154 00:13:19,000 --> 00:13:23,001 We're basically trying to do everything we possibly can to give people access to 155 00:13:23,001 --> 00:13:29,000 the information they need. And just in the last few minutes I'm going to 156 00:13:29,000 --> 00:13:35,001 highlight two stories where ads were crucial. In October we reported on a group 157 00:13:35,001 --> 00:13:41,001 called Carolina Rising. They spent effectively 100 percent of their money 158 00:13:41,001 --> 00:13:47,001 supporting a single candidate, Tom Tillis, who ended up unseating the incumbent 159 00:13:47,001 --> 00:13:54,000 Democrat in North Carolina. The group did so without reporting any of its donors. 160 00:13:55,000 --> 00:14:00,000 And we were able to use both FEC, Federal Election Commission data, and Federal 161 00:14:00,000 --> 00:14:04,001 Communications data to show, you know, this is the ad, these are the ads that 162 00:14:04,001 --> 00:14:09,000 they reported. This is the total amount spent. But we can point to these ads and 163 00:14:09,000 --> 00:14:14,000 show that they were quite active before they started reporting to the commission. 164 00:14:15,001 --> 00:14:19,000 The same is true for a report we put out yesterday about a group called 165 00:14:19,000 --> 00:14:24,000 Oklahomans for a Conservative Future. This group was run by political operatives. 166 00:14:26,000 --> 00:14:32,000 It was started weeks before the candidate it supported was – made his 167 00:14:32,000 --> 00:14:39,000 announcement that he was going to run for office. It spent between 1.3 and 1.6 168 00:14:39,000 --> 00:14:45,001 million trying to get this guy elected. He failed. And then they essentially sent 169 00:14:45,001 --> 00:14:51,000 an IOU to the IRS saying, OK, we know we didn't – we spent a lot of money on 170 00:14:51,000 --> 00:14:56,000 politics, but we promise we'll do better next year, which isn't actually a legal 171 00:14:56,000 --> 00:15:01,001 argument and doesn't hold water. The question now is whether the FEC or IRS will 172 00:15:01,001 --> 00:15:07,001 ever do anything about that, and that is very much an open question. So these two 173 00:15:07,001 --> 00:15:13,001 examples are showing a trend not only of, you know, we have increasing political 174 00:15:13,001 --> 00:15:18,001 activity from groups that don't disclose their donors, but these groups are 175 00:15:18,001 --> 00:15:23,000 becoming a part of the campaigns themselves. This is what's particularly 176 00:15:23,000 --> 00:15:27,000 worrisome, is that these are supposed to be apolitical organizations, but they're 177 00:15:27,000 --> 00:15:33,000 actually a sort of non-disclosing extension of the campaign. So, for example, 178 00:15:35,000 --> 00:15:39,000 Conservative Solutions Project in this cycle has spent about $8 million 179 00:15:39,000 --> 00:15:43,000 supporting Marco Rubio. Of course, they haven't reported any of that to the 180 00:15:43,000 --> 00:15:48,001 Federal Election Commission. What they've done is they've framed all of the ads 181 00:15:48,001 --> 00:15:52,000 as educational ads. They want to talk about the Iran 182 00:15:52,000 --> 00:15:53,001 deal. They want to talk about taxes. 183 00:15:53,001 --> 00:15:58,001 But they only want to talk about it as it relates to Marco Rubio and how stellar 184 00:15:58,001 --> 00:16:04,000 his decision-making has been on all of these issues. Or, in my favorite example, 185 00:16:04,000 --> 00:16:08,000 they talk about Marco Rubio and Mike Lee, and at the end of the ad it says 186 00:16:08,000 --> 00:16:13,000 something along the lines of support Marco Rubio and big letters, and then 187 00:16:13,000 --> 00:16:18,000 underneath in small letters it says, and Mike Lee, and gives their phone numbers. 188 00:16:19,000 --> 00:16:25,000 So this is something that is going to be a moving target the whole cycle. It's 189 00:16:25,000 --> 00:16:30,001 going to be really difficult for us to... We're going to get the data to people, 190 00:16:30,001 --> 00:16:37,000 but it's not easy. But we're hoping that this partnership with Internet Archive 191 00:16:37,000 --> 00:16:39,001 is going to make that a lot easier. 192 00:16:40,000 --> 00:16:47,000 Thank you. We are, too. We all are kind of hoping that the archive project will 193 00:16:47,000 --> 00:16:53,001 surface an ad by a cure-to-for unheard-of organization, which then we can tell 194 00:16:53,001 --> 00:16:57,001 Open Secrets about, and they'll run out and investigate it. And we'll get that 195 00:16:57,001 --> 00:17:00,001 information up there, as will they and all of us, really. 196 00:17:01,000 --> 00:17:05,000 So next I'd like to introduce Dave Leventhal. He's a senior political reporter 197 00:17:05,000 --> 00:17:11,001 for the Center for Public Integrity, a Pulitzer Prize-winning organization where 198 00:17:11,001 --> 00:17:15,000 they investigate many things among the money and politics, and they have a really 199 00:17:15,000 --> 00:17:20,000 great program this election called Source Check. And I believe 200 00:17:20,000 --> 00:17:22,000 you're going to talk a little bit more about that. 201 00:17:22,000 --> 00:17:29,000 Absolutely. Well, thanks for being here. And of course, you might be out there, a 202 00:17:29,000 --> 00:17:33,000 voter in Iowa, or perhaps a voter in New Hampshire, where the first caucus and 203 00:17:33,000 --> 00:17:37,001 primary are showing up, or South Carolina, Nevada, some of the early states. And 204 00:17:37,001 --> 00:17:42,001 you sit down tonight, you turn on Wheel of Fortune with the Colbert Show or Jimmy 205 00:17:42,001 --> 00:17:47,000 Fallon, and then it goes to a commercial break, and you see another political 206 00:17:47,000 --> 00:17:52,001 advertisement. And it has the same hallmarks as all the rest, ominous music and 207 00:17:52,001 --> 00:17:57,001 dark shadowy figures. And at the end of it, it lists a name that you don't 208 00:17:57,001 --> 00:18:02,000 recognize. And it might be a name that's only slightly less ridiculous than 209 00:18:02,000 --> 00:18:08,000 Americans for Hamburgers or Citizens for Unicorns. And you're wondering, well, 210 00:18:08,000 --> 00:18:12,000 what is this group? So you go, like anyone would, to Google, and you type in the 211 00:18:12,000 --> 00:18:15,001 name of the organization, if you can even catch it in the midst of the 212 00:18:15,001 --> 00:18:20,001 advertisement, and you don't find much about them. Maybe a name of the 213 00:18:20,001 --> 00:18:26,000 organization, and a couple of news articles about what they've done here in a 214 00:18:26,000 --> 00:18:30,000 certain television market or there in another television market, but not a whole 215 00:18:30,000 --> 00:18:34,001 lot about who's behind them, or what they want. But you know that certainly in 216 00:18:34,001 --> 00:18:40,000 their advertisements that you're seeing day in and day out, that they're very, 217 00:18:40,000 --> 00:18:45,000 very much against Hillary Clinton, or against Jeb Bush, or against Donald Trump, 218 00:18:45,000 --> 00:18:49,000 or any of the other candidates who are running in the presidential race. And then 219 00:18:49,000 --> 00:18:53,000 you wonder, well, all right, what do they want? What's their point? 220 00:18:53,001 --> 00:18:55,000 What are they trying to get at? 221 00:18:55,001 --> 00:19:00,000 Why are they trying to influence this election that I'm soon going to be voting 222 00:19:00,000 --> 00:19:05,001 in that could very much change the trajectory of the entire presidential race? So 223 00:19:05,001 --> 00:19:09,001 at the Center for Public Integrity, we're a nonpartisan, nonprofit investigative 224 00:19:09,001 --> 00:19:14,000 news organization. We've launched a project for 2016 called Buying of the 225 00:19:14,000 --> 00:19:20,000 President 2016. And as Nancy mentioned, one key element of this project is called 226 00:19:20,000 --> 00:19:24,000 Source Check, which is going to be an effort to look at all of these different 227 00:19:24,000 --> 00:19:28,001 advertisements to the extent that we can and dig deep on them. Try to find out 228 00:19:28,001 --> 00:19:33,000 truly who are the people behind these advertisements, where the money is coming 229 00:19:33,000 --> 00:19:36,001 from behind these advertisements, and ultimately try to answer the question, 230 00:19:37,000 --> 00:19:40,001 well, what interests do these groups, which you have so little information about, 231 00:19:41,000 --> 00:19:46,000 have in trying to support one candidate over another, or certainly, as many, many 232 00:19:46,000 --> 00:19:51,000 do, attack the heck out of certain candidates in the race, who are going to be 233 00:19:51,000 --> 00:19:54,001 running against everyone else and competing tooth and nail in all of these early 234 00:19:54,001 --> 00:19:59,001 contests, and then forward from that onto some of the later contests. So as part 235 00:19:59,001 --> 00:20:04,001 of this, we're relying heavily on the Internet Archive, in addition to many, many 236 00:20:04,001 --> 00:20:09,000 other data sources, some of which Robert mentioned, federal sources, Cantor 237 00:20:09,000 --> 00:20:13,000 Media, CMAG, which is an advertising tracking firm, and using this and 238 00:20:13,000 --> 00:20:17,001 amalgamating it in a way that's going to be able to provide real, useful, 239 00:20:18,000 --> 00:20:22,001 accessible journalism to tell those voters, to tell people why this stuff should 240 00:20:22,001 --> 00:20:26,001 matter to them, and truly who these people are behind the ads. We're in an era 241 00:20:26,001 --> 00:20:32,001 right now where oftentimes it's extremely difficult to find out who are behind 242 00:20:32,001 --> 00:20:37,001 these organizations that just simply aren't run by the candidates themselves. 243 00:20:38,000 --> 00:20:41,000 Sure, you have a lot of candidate committees, as they always have in every 244 00:20:41,000 --> 00:20:46,000 election, putting up television advertisements. But ever since a seminal Citizens 245 00:20:46,000 --> 00:20:51,000 United versus Federal Election Commission decision, which celebrated, if you can 246 00:20:51,000 --> 00:20:55,001 call it a celebration somewhat, others wouldn't, its sixth anniversary yesterday, 247 00:20:56,000 --> 00:21:01,000 it has freed in a major way various types of organizations, be them political 248 00:21:01,000 --> 00:21:07,000 groups, or in many cases, nonprofit organizations that shouldn't have a primary 249 00:21:07,000 --> 00:21:12,000 purpose by law of engaging in political activity. It's given them the ability to 250 00:21:12,000 --> 00:21:17,000 nevertheless engage in politics in a major way with millions, even tens of 251 00:21:17,000 --> 00:21:22,001 millions of dollars behind each individual group. So by pouring into these 252 00:21:22,001 --> 00:21:28,000 groups, or boring into these groups, we want to basically educate the public in a 253 00:21:28,000 --> 00:21:32,001 way that, again, is going to be useful when they go and cast their ballot at the 254 00:21:32,001 --> 00:21:37,000 ballot booth or go and caucus in their state. And Robert mentioned one of the 255 00:21:37,000 --> 00:21:40,001 groups that has been very active in the presidential race, a group that's 256 00:21:40,001 --> 00:21:45,001 supporting Marco Rubio, is part of our source check project where we do sort of a 257 00:21:45,001 --> 00:21:50,001 census or a deep dive on the activities of individual groups. We looked at an 258 00:21:50,001 --> 00:21:56,000 environmental group that accepts anonymous contributions, but yet is up with an 259 00:21:56,000 --> 00:22:02,000 ad campaign supporting Bernie Sanders in the presidential race and tried to find 260 00:22:02,000 --> 00:22:06,001 out exactly what they were all about. And more broadly speaking, even going 261 00:22:06,001 --> 00:22:11,000 beyond just television advertisements, you have many organizations, even super 262 00:22:11,000 --> 00:22:15,000 PACs, these groups that can raise and spend unlimited amounts of money to 263 00:22:15,000 --> 00:22:19,001 advocate for and against politicians in a race. They're supposed to disclose all 264 00:22:19,001 --> 00:22:23,001 their donors. One of my colleagues who's here today, Michael Beckle, wrote just 265 00:22:23,001 --> 00:22:28,001 an absolutely outrageous piece several months ago about a super PAC, also 266 00:22:28,001 --> 00:22:33,001 purporting to support Bernie Sanders, and yet isn't filing any of its required 267 00:22:33,001 --> 00:22:39,001 paperwork. And it's run by somebody who we revealed had a criminal history and a 268 00:22:39,001 --> 00:22:41,001 very, very, very curious background. 269 00:22:41,001 --> 00:22:47,000 Would encourage you to read this article. And ultimately was able to convince 270 00:22:47,000 --> 00:22:52,000 Daniel Craig, who you may know as the actor who plays James Bond, to make about a 271 00:22:52,000 --> 00:22:55,000 $50,000 donation to this group. 272 00:22:55,000 --> 00:23:00,000 And his publicist was perhaps a little bit surprised to hear all the details 273 00:23:00,000 --> 00:23:05,000 about this group when it came out. So the moral of the story is, as a citizen, as 274 00:23:05,000 --> 00:23:10,000 a voter, as somebody who cares about politics, you need to pay attention to this 275 00:23:10,000 --> 00:23:15,000 stuff, but oftentimes don't have the requisite tools and information to go ahead 276 00:23:15,000 --> 00:23:20,000 and do it. So we want to make sure that you do ultimately do that. And I would 277 00:23:20,000 --> 00:23:25,000 point out one other feature of our reporting this year, which will range from 278 00:23:25,000 --> 00:23:29,000 deep dive investigative stories to these quick hit source check items. We have a 279 00:23:29,000 --> 00:23:34,000 data visualization that was pioneered by another colleague, Chris Zubach, skis 280 00:23:34,000 --> 00:23:38,000 that went up yesterday, relies very heavily on internet archive material in 281 00:23:38,000 --> 00:23:43,001 addition to other sources, and is a very accessible visualization of where these 282 00:23:43,001 --> 00:23:48,000 ads are, who they're sponsored by, where they're going, the information will be 283 00:23:48,000 --> 00:23:54,000 updated constantly every week, and again, provides a way just another way, visual 284 00:23:54,000 --> 00:24:00,000 way, as opposed to a textual way to get this information to absorb it and to act 285 00:24:00,000 --> 00:24:06,000 on it, if you choose to do so. Thanks, Dave. It's all really, really fascinating 286 00:24:06,000 --> 00:24:11,001 stuff. And so we've heard from two organizations that specialize in following the 287 00:24:11,001 --> 00:24:16,001 money trail. And now we have a host of folks who check the facts. So, you know, 288 00:24:16,001 --> 00:24:19,001 one thing of interest that we're trying to gather on the news website, again, 289 00:24:19,001 --> 00:24:25,001 that's political ad archive. org launching today. We are gathering together 290 00:24:25,001 --> 00:24:32,000 information about the facts in the ads, you know, our partner groups who 291 00:24:32,000 --> 00:24:37,001 specialize in is this correct? Is it not correct? Is it kind of correct? Is it 292 00:24:37,001 --> 00:24:42,000 medium correct? And so I'm going to introduce Lori Robertson. She's the managing 293 00:24:42,000 --> 00:24:47,001 editor of fact check.org. And I believe she has something to show you that we'll 294 00:24:47,001 --> 00:24:52,001 start with. Right. Seven out of 10 Americans take prescription drugs. But in the 295 00:24:52,001 --> 00:24:57,000 last seven years, drug prices have doubled. Hillary's going to take on the drug 296 00:24:57,000 --> 00:25:02,000 companies require Medicare to negotiate lower drug prices. Let people buy their 297 00:25:02,000 --> 00:25:06,001 prescription drugs from countries like Canada at half the price and cap monthly 298 00:25:06,001 --> 00:25:10,001 prescription costs for every American. The drug companies have been overcharging 299 00:25:10,001 --> 00:25:15,000 long enough. It's time to fight back. I'm Hillary Clinton and I approve this 300 00:25:15,000 --> 00:25:22,000 message. Great. So this is one of the ads that we've fact checked thus far 301 00:25:22,000 --> 00:25:27,001 in this campaign cycle, obviously from the Clinton campaign. And we looked at the 302 00:25:27,001 --> 00:25:32,001 claim in there that drug prices have doubled in the last seven years. The Clinton 303 00:25:32,001 --> 00:25:38,000 campaign pointed to a report by a group called Express Scripts. They're one of 304 00:25:38,000 --> 00:25:43,000 the largest pharmacy benefit managers in the country. When we looked at the 305 00:25:43,000 --> 00:25:49,000 report, it didn't really say what the ad says. It did say that brand name drugs 306 00:25:49,000 --> 00:25:54,000 had more than doubled in the last seven years. But if you continue reading the 307 00:25:54,000 --> 00:25:58,000 report, it also says that more than 80 percent of filled prescriptions are 308 00:25:58,000 --> 00:26:04,001 actually generics and that generic drug prices had declined by about 63 percent 309 00:26:04,001 --> 00:26:06,000 over the same time period. 310 00:26:07,000 --> 00:26:12,001 Unfortunately, the group did not come up with a number that kind of fused the 311 00:26:12,001 --> 00:26:19,001 generic drug prices and the brand name drug prices, at least in terms of how much 312 00:26:19,001 --> 00:26:26,000 consumers are buying. And there in fact is no data that shows overall drug 313 00:26:26,000 --> 00:26:30,000 prices what has happened over the last seven years. 314 00:26:30,001 --> 00:26:37,000 We spoke with an expert as well at the University of Chicago about this. 315 00:26:38,000 --> 00:26:44,000 And we just don't have data on what has happened with overall drug prices. But 316 00:26:44,000 --> 00:26:48,000 obviously, if brand names have gone up more than double and generics have 317 00:26:48,000 --> 00:26:54,001 declined by 63 percent and 80 percent of filled prescriptions are 318 00:26:54,001 --> 00:27:00,001 generics, it was not more than doubling over seven years as the ad says. 319 00:27:01,001 --> 00:27:08,000 In fact, there is some other data on national spending on prescription drugs. The 320 00:27:08,000 --> 00:27:12,001 Centers for Medicare and Medicaid Services puts out annually a report on national 321 00:27:12,001 --> 00:27:18,000 health care expenditures. And that the drug spending has continued to go up, but 322 00:27:18,000 --> 00:27:23,001 they've found low rates of that growth in spending in recent years. And they've 323 00:27:23,001 --> 00:27:30,001 specifically attributed that to a switch to generic drugs. I think what we 324 00:27:30,001 --> 00:27:35,000 missed the very beginning of that ad where the campaign mentioned 325 00:27:35,000 --> 00:27:37,000 some specific conditions. 326 00:27:37,001 --> 00:27:44,001 They say heart disease, asthma, diabetes, and the report that they point to again 327 00:27:44,001 --> 00:27:50,000 does look a little more specifically at drugs, top 10 drugs 328 00:27:50,000 --> 00:27:54,001 for those specific conditions year over year. 329 00:27:55,001 --> 00:27:59,001 So in looking and that combines both the generics and the brand name drugs. So 330 00:27:59,001 --> 00:28:05,000 looking at those specific conditions year over year for 2014, diabetes was the 331 00:28:05,000 --> 00:28:11,000 only one that saw an increase in overall drug prices and heart disease and asthma 332 00:28:11,000 --> 00:28:17,001 saw a decline. So you know this type of claim is something that we see a lot of 333 00:28:17,001 --> 00:28:22,000 where you know a campaign, I guess a little bit like cherry picking, but I think 334 00:28:22,000 --> 00:28:25,001 the Clinton campaign goes a little beyond cherry picking here and cherry picks 335 00:28:25,001 --> 00:28:29,001 their statistic and then kind of leaves off that this was about brand name drugs. 336 00:28:31,000 --> 00:28:36,001 But we see this a lot where a campaign points to a report as supposed backup for 337 00:28:36,001 --> 00:28:40,001 the claim, but when you go in and actually read the report you find that it did 338 00:28:40,001 --> 00:28:47,000 not really back up what they're saying or had additional information that would 339 00:28:47,000 --> 00:28:53,001 make the claim misleading or add some kind of 340 00:28:53,001 --> 00:28:59,000 information that makes the claim not what it appears to be in the ad. 341 00:29:01,001 --> 00:29:08,000 Thanks Laurie. That's really interesting and yeah it's so true that if you can 342 00:29:08,000 --> 00:29:12,001 link to something people don't necessarily read it right so you get to point out 343 00:29:12,001 --> 00:29:17,000 what's actually in there. So next we're going to have John Greenberg who is staff 344 00:29:17,000 --> 00:29:22,001 writer at PolitiFact and PolitiFact also a Pulitzer Prize 345 00:29:22,001 --> 00:29:29,001 winning organization also sleuthing out the facts and you also have a 346 00:29:29,001 --> 00:29:35,001 video we're going to show. Yeah well let's go for it. This was Trump's first 347 00:29:35,001 --> 00:29:39,000 television ad. I mean Donald Trump right you know him. 348 00:29:40,001 --> 00:29:45,001 I'm Donald Trump and I approve this message. The politicians can pretend it's 349 00:29:45,001 --> 00:29:50,001 something else but Donald Trump calls it radical Islamic terrorism. That's why 350 00:29:50,001 --> 00:29:54,001 he's calling for a temporary shutdown of Muslims entering the United States until 351 00:29:54,001 --> 00:29:59,000 we can figure out what's going on. He'll quickly cut the head off ISIS and take 352 00:29:59,000 --> 00:30:03,000 their oil and he'll stop illegal immigration by building a wall on our southern 353 00:30:03,000 --> 00:30:10,000 border that Mexico will pay for. We will make America great again. 354 00:30:12,000 --> 00:30:19,000 All right that was quintessential Trump and the part of that ad pretty 355 00:30:19,000 --> 00:30:25,001 scary stuff are those shots of people scurrying over the border 356 00:30:25,001 --> 00:30:31,001 and that's the southern border that Trump says he's going to protect. 357 00:30:32,001 --> 00:30:38,001 So there's just small problem. Small problem. That image comes from Morocco 358 00:30:38,001 --> 00:30:45,000 and it was a number of people crossing the border into a Spanish 359 00:30:45,000 --> 00:30:51,001 enclave on Morocco's border. So you anyone watching this would 360 00:30:51,001 --> 00:30:55,000 presume this is all about America's southern border. 361 00:30:56,000 --> 00:31:03,000 What we found is actually sort of very typical of the kind of ways that 362 00:31:03,000 --> 00:31:09,000 the Trump organization uses information. I won't necessarily call it facts. 363 00:31:10,001 --> 00:31:15,000 This video actually came from 2014. 364 00:31:15,001 --> 00:31:22,001 It was shot or appeared on an Italian broadcast network and 365 00:31:22,001 --> 00:31:29,000 at the time it actually had a location and a time stamp. But like a 366 00:31:29,000 --> 00:31:36,000 lot of information it got detached from those details so that by the time the ad 367 00:31:36,000 --> 00:31:43,000 drifts through YouTube and shows up in the Trump ad it is devoid 368 00:31:43,000 --> 00:31:49,001 of any time stamp or any location. It's just simply there to be used. So we 369 00:31:49,001 --> 00:31:55,001 tracked it down. We pointed this out and rated the whole ad or this part 370 00:31:55,001 --> 00:31:57,000 of the ad pants on fire. 371 00:31:57,001 --> 00:32:03,001 The interesting thing is that the response of the Trump organization at first was 372 00:32:03,001 --> 00:32:08,000 to say well we don't really know where video comes from and then the next day 373 00:32:08,000 --> 00:32:13,001 campaign manager doubles down and says this was one thousand percent on purpose. 374 00:32:14,000 --> 00:32:19,000 We don't care that we were talking about America's southern border and this is a 375 00:32:19,000 --> 00:32:25,000 shot from Morocco. It doesn't matter. We're just trying to make the point. So 376 00:32:25,000 --> 00:32:31,000 we'll give him points for honesty on that point. But what we think is so 377 00:32:31,000 --> 00:32:37,000 important about the whole project now with the Internet Archive and tracking 378 00:32:37,000 --> 00:32:43,001 these political ads is we all know you know the old line from Mark Twain about 379 00:32:43,001 --> 00:32:50,000 you know rumor is halfway around the world before the truth gets its boots on. So 380 00:32:50,000 --> 00:32:57,000 a political ad particularly in the general campaign may show up in multiple 381 00:32:57,000 --> 00:33:03,001 states. And what we are interested in doing is working through 382 00:33:03,001 --> 00:33:10,000 our partnerships. Our partnerships with NBC News with 383 00:33:10,000 --> 00:33:16,001 Scripps television stations with a number of outfits in Arizona Ohio 384 00:33:16,001 --> 00:33:23,001 Pennsylvania Virginia. These are our affiliates so that we can if we 385 00:33:23,001 --> 00:33:30,001 identify or if one of them identifies a political ad that has a factual error all 386 00:33:30,001 --> 00:33:37,000 of them can be alerted because that ad may show up in their markets and the fact 387 00:33:37,000 --> 00:33:44,000 check would be news to their audience. This is a multiplier effect that allows 388 00:33:44,000 --> 00:33:50,001 us to close the gap between the speed with which a campaign can put an ad out 389 00:33:50,001 --> 00:33:56,001 there and get it to millions of people and then our ability to turn around and 390 00:33:56,001 --> 00:34:01,000 get a corrective message out to many people. We're not going to necessarily reach 391 00:34:01,000 --> 00:34:05,000 millions but we'll be faster and we'll be reaching a lot of them through the 392 00:34:05,000 --> 00:34:09,001 whole network of affiliates and that's why there's a real multiplier effect that 393 00:34:09,001 --> 00:34:15,000 can come through having this resource. Thanks very much. So next we're going to 394 00:34:15,000 --> 00:34:18,001 hear from Michelle Lee who's a reporter with the Washington Post fact checker 395 00:34:18,001 --> 00:34:23,000 also has a video to show. Should I cue that up? Barack Obama released terrorists 396 00:34:23,000 --> 00:34:27,001 from Guantanamo and now they're plotting to attack us. Instead of fighting to 397 00:34:27,001 --> 00:34:32,000 fund our troops he fights to fund Planned Parenthood. He spies on Israel and cut 398 00:34:32,000 --> 00:34:36,001 a deal with Iran. His plan after the attack in San Bernardino take away our guns 399 00:34:36,001 --> 00:34:41,000 and while ISIS is beheading people and burning them in cages he says climate 400 00:34:41,000 --> 00:34:45,001 change is our greatest threat. I'm Marco Rubio. I approved this message because 401 00:34:45,001 --> 00:34:50,001 America needs a real commander in chief and a president who will keep us safe. 402 00:34:54,000 --> 00:34:58,001 So apparently the world is falling apart with the ominous music in the 403 00:34:58,001 --> 00:35:05,001 background. We chose this ad to fact check. I mean we're trying to fact check 404 00:35:05,001 --> 00:35:11,001 as many ads as possible especially right now because you know this is a candidate 405 00:35:11,001 --> 00:35:17,001 speaking directly to the audience with just really ominous music and it's filled 406 00:35:17,001 --> 00:35:23,000 with rhetoric and this is an example of you know a very challenging ad because 407 00:35:23,000 --> 00:35:29,000 it's a series of claim a series of claims with very varying degrees of veracity. 408 00:35:29,001 --> 00:35:36,001 So you know for example yes the US does spy on Israel but the US also spies 409 00:35:36,001 --> 00:35:42,001 on Iran and he didn't say that. So we chose two statements out of that ad to 410 00:35:42,001 --> 00:35:47,001 check that were very fact checkable and are very common GOP talking points and we 411 00:35:47,001 --> 00:35:53,001 wanted to unpack that. So we chose the Guantanamo Bay one where Barack Obama is 412 00:35:53,001 --> 00:35:57,001 releasing terrorists and they're all plotting to attack us and then the other one 413 00:35:57,001 --> 00:36:02,000 where Obama wants to take away our guns and that particularly stood out because 414 00:36:02,000 --> 00:36:08,000 this ad came out either the day of or the day after Obama announced 415 00:36:08,000 --> 00:36:10,000 his executive action on guns. 416 00:36:10,000 --> 00:36:15,000 So it made it sound like that was his plan to take away our guns. So to unpack 417 00:36:15,000 --> 00:36:21,001 those two points the Rubio campaign pointed to statistics on ex Guantanamo 418 00:36:21,001 --> 00:36:27,001 Bay detainees who were confirmed or suspected of re-engaging in terrorist 419 00:36:27,001 --> 00:36:33,000 activities but only six percent of them happened under Obama's term. The rest 420 00:36:33,000 --> 00:36:40,000 were under Bush's term so that was pretty misleading and we a 421 00:36:40,000 --> 00:36:46,000 lot of people point to the Australian buyback program that happened where you 422 00:36:46,000 --> 00:36:52,001 know they took back guns, assault rifles mostly, but the Rubio campaign didn't 423 00:36:52,001 --> 00:36:56,001 actually point to that. They just pointed to Obama's announcement on the 424 00:36:56,001 --> 00:37:01,001 executive action which didn't mention Australian buyback at all. It actually 425 00:37:01,001 --> 00:37:08,001 didn't have a buyback plan and so we noted that while Obama has praised the 426 00:37:08,001 --> 00:37:13,001 Australian buyback program that's not a part of his executive action at all and 427 00:37:13,001 --> 00:37:18,001 this idea that Obama wants to take away our guns you know it's it's rhetoric but 428 00:37:18,001 --> 00:37:25,000 it's not based on actual facts in his plan. So you may know that we rate our 429 00:37:25,000 --> 00:37:29,000 claims on a Pinocchio scale. I think there's a Pinocchio there so it's from zero 430 00:37:29,000 --> 00:37:35,001 to four Pinocchio's four being false and in this case the guns claim was clearly 431 00:37:35,001 --> 00:37:41,000 a four Pinocchio claim. The Guantanamo one was mostly like a two which is like 432 00:37:41,000 --> 00:37:46,000 you know lacks context. You can think of it as half true if you're using the 433 00:37:46,000 --> 00:37:51,001 PolitiFact scale and so we you know evened it out with the three Pinocchio's 434 00:37:51,001 --> 00:37:57,001 which is mostly false. But we're just really excited to partner on this project 435 00:37:57,001 --> 00:38:03,001 because as John said you know it's so important to know about these ads that are 436 00:38:03,001 --> 00:38:08,000 being aired in markets that we're not even aware of. You know at the Fact Checker 437 00:38:08,000 --> 00:38:13,000 it's me and Glenn Kessler. Our entire team is two people and we can't be 438 00:38:13,000 --> 00:38:18,001 everywhere so we rely on our readers a lot. Reader submissions, people sending us 439 00:38:18,001 --> 00:38:23,000 photos of ads that they're getting in their mail, sending us tips about ads that 440 00:38:23,000 --> 00:38:28,000 were airing in their markets and this just helps us be in more places at once. 441 00:38:29,001 --> 00:38:33,000 Thanks Michelle and so I just wanted to flashback. 442 00:38:33,001 --> 00:38:38,000 We have a number of really great partners who are joining us in this effort for 443 00:38:38,000 --> 00:38:42,000 the website we're launching today the Political TV Ad Archive political ad 444 00:38:42,000 --> 00:38:48,001 archive.org and we've heard from some of them some of them aren't here. We newly 445 00:38:48,001 --> 00:38:53,000 joined our efforts are the Duke Reporters Lab out of North Carolina. We have a 446 00:38:53,000 --> 00:38:57,000 bunch of students at the ready who will be looking at the ads we surface and 447 00:38:57,000 --> 00:39:01,000 we'll be fact checking them like crazy. We'll get their information up there and 448 00:39:01,000 --> 00:39:05,000 the American Press Institute who we're also partnering with to do some trainings 449 00:39:05,000 --> 00:39:09,001 thanks to the Democracy Fund and then again you've heard from OpenSecrets.org, 450 00:39:10,001 --> 00:39:14,000 the Center for Public Integrity, PolitiFact, FactCheck.org and the Washington 451 00:39:14,000 --> 00:39:18,000 Post Fact Checker. And now we're going to get kind of more into the nuts and 452 00:39:18,000 --> 00:39:22,000 bolts of how we built this project, how we built this site. And there's a small 453 00:39:22,000 --> 00:39:27,001 army of people who are not present in this room who help make this possible and 454 00:39:27,001 --> 00:39:31,001 but I would say at the head of it are a couple people in this room, one of whom 455 00:39:31,001 --> 00:39:35,001 is in the audience but will be around to answer questions and she is Tracy 456 00:39:35,001 --> 00:39:42,001 Jaquith who's the architect of the TV archive at the Internet Archive and unlike 457 00:39:42,001 --> 00:39:49,000 me probably had AV Club and many other things in her college application. And 458 00:39:49,000 --> 00:39:53,000 so she will be available also to speak and then I'm going to introduce Dan Schulz 459 00:39:53,000 --> 00:39:58,001 who is a senior software engineer with us and he has built the back end of this 460 00:39:58,001 --> 00:40:01,001 thing and he's going to talk a little bit more about how he did that 461 00:40:01,001 --> 00:40:04,000 in terms that we can all understand. 462 00:40:04,000 --> 00:40:10,001 I hope so. Thanks, Nancy. So today we've been talking a lot about the content and 463 00:40:10,001 --> 00:40:14,001 the data side of the political ad archive but I want to draw attention to 464 00:40:14,001 --> 00:40:19,000 something that might not be as transparent which is that everything powering this 465 00:40:19,000 --> 00:40:23,000 data set, everything we built and everything we're using is free and open source. 466 00:40:23,001 --> 00:40:27,000 And so we're hoping that a lot of people out there who are excited by the project 467 00:40:27,000 --> 00:40:32,000 will not only use our data but might actually use some of our code and maybe even 468 00:40:32,000 --> 00:40:37,001 contribute back to it because we're continuing to improve it. We built a system 469 00:40:37,001 --> 00:40:44,000 to find copies of known political ads and it's called the dupletron and from 470 00:40:44,000 --> 00:40:48,001 the perspective of a computer there's actually nothing special at all about a 471 00:40:48,001 --> 00:40:52,001 political ad. It's just a piece of video that can be found inside bigger pieces 472 00:40:52,001 --> 00:40:58,000 of video and video is a lot of data so it's kind of slow to find matches. 473 00:41:00,001 --> 00:41:05,000 So this is a process called audio fingerprinting and audio fingerprinting 474 00:41:05,000 --> 00:41:10,001 involves taking a sound, breaking it into thousands of tiny pieces like less than 475 00:41:10,001 --> 00:41:16,000 a second long and then summarizing those pieces. And in the summary process the 476 00:41:16,000 --> 00:41:20,001 trick is that only the most noteworthy and important pieces of that little 477 00:41:20,001 --> 00:41:25,001 snippet get kept. A lot of the noise gets thrown away. And so the reason we do 478 00:41:25,001 --> 00:41:30,001 this is because it's a lot faster and kind of ironically a lot more accurate to 479 00:41:30,001 --> 00:41:35,001 compare summaries of sound than it is to compare the full-blown sound. And that's 480 00:41:35,001 --> 00:41:40,001 what allows us to process millions of minutes of video at a time. So the 481 00:41:40,001 --> 00:41:45,000 collection of all these little summarized pieces is called an audio fingerprint 482 00:41:45,000 --> 00:41:50,001 and we're using a free and open tool called ODD-F Print to do the generation and 483 00:41:50,001 --> 00:41:56,001 to also do some of the exploration. So in practice we get an hour of TV content 484 00:41:56,001 --> 00:42:03,000 from Tracy's pipeline which we then the first thing we do is summarize it into an 485 00:42:03,000 --> 00:42:08,001 audio fingerprint. Then we compare that fingerprint for that hour with all the 486 00:42:08,001 --> 00:42:12,001 fingerprints of the political ads we've seen and registered in our system. And 487 00:42:12,001 --> 00:42:18,001 anytime we find an overlap if that is long enough it gets flagged as a copy of a 488 00:42:18,001 --> 00:42:23,000 political ad and pretty soon that shows up on the political ad archive as an 489 00:42:23,000 --> 00:42:27,000 increased count. And when you download the data you can see the channel, the 490 00:42:27,000 --> 00:42:33,000 markets, the time of airing and the length of the airing as well. As we're moving 491 00:42:33,000 --> 00:42:36,000 forward in this project we're working really hard to make the Duplatron faster 492 00:42:36,000 --> 00:42:40,001 which means we'll be able to process even more content because we're regularly 493 00:42:40,001 --> 00:42:47,000 adding new channels to what this system is watching. And we're also trying to 494 00:42:47,000 --> 00:42:52,000 do a couple things that are pushing the limits of this system which we're hoping 495 00:42:52,000 --> 00:42:59,000 to be able to show off in the coming months. For instance, in 496 00:42:59,000 --> 00:43:04,000 addition to looking for copies of known political ads, we're also capable of 497 00:43:04,000 --> 00:43:10,000 flagging copies of any repeated content whether it's been tagged as special or 498 00:43:10,000 --> 00:43:15,000 not. So this means whenever we see a new political ad that we've never seen 499 00:43:15,000 --> 00:43:21,000 before it's possible to be able to actually surface that to a human and have the 500 00:43:21,000 --> 00:43:27,000 human review it as a potential brand new airing. It's also possible to keep track 501 00:43:27,000 --> 00:43:31,000 of things like sound bites from interviews or sound bites from political speeches 502 00:43:31,000 --> 00:43:35,001 and see how those things are disseminated across the media across the country 503 00:43:35,001 --> 00:43:38,000 right as they're popping out of the woodwork. 504 00:43:39,001 --> 00:43:43,000 So all this is to say please keep an eye on this project in addition to using the 505 00:43:43,000 --> 00:43:47,001 data over the next couple months because we're going to be releasing new features 506 00:43:47,001 --> 00:43:52,000 and if you or anybody you know is interested in collaborating reach out. Thanks. 507 00:43:52,001 --> 00:43:58,000 Thanks Dan. So now I'm going to talk a little bit. I'm Nancy Watzman and I am the 508 00:43:58,000 --> 00:44:04,001 managing editor of this new website politicaladarchive.org. I can be reached 509 00:44:04,001 --> 00:44:11,000 by twitter at at nwatzman or you can reach me by email at nancyw at archive 510 00:44:11,000 --> 00:44:14,001 .org and I'm saying that because I am the person you should bug if you have any 511 00:44:14,001 --> 00:44:18,000 particular questions about the project. I'm going to get into a little bit of the 512 00:44:18,000 --> 00:44:23,000 nuts and bolts of you know how it works and what it looks like. But first back to 513 00:44:23,000 --> 00:44:29,001 the internet archive. Roger spoke a little bit about who we are. Internet Archive 514 00:44:29,001 --> 00:44:34,000 is based in San Francisco in this lovely building which I'm told was bought to 515 00:44:34,000 --> 00:44:39,001 match the logo. It looks like a library. It's a former Christian scientist church 516 00:44:39,001 --> 00:44:43,001 and it's it's actually if you're ever in the neighborhood you should drop by and 517 00:44:43,001 --> 00:44:50,001 visit because they give tours and they have swag you can buy with the logo and 518 00:44:50,001 --> 00:44:52,000 it's a very interesting place to visit. 519 00:44:52,000 --> 00:44:57,000 Has the very modest goal that our organization of archiving all of human 520 00:44:57,000 --> 00:45:04,000 knowledge online for free accessible to all and we're probably best known for the 521 00:45:04,000 --> 00:45:08,000 Wayback Machine. Many journalists are familiar with the Wayback Machine. Lots of 522 00:45:08,000 --> 00:45:13,001 other folks in the general publicist as well. This is a really amazing tool where 523 00:45:13,001 --> 00:45:18,001 you can post in paste in a URL of any website and see a snapshot of the way it 524 00:45:18,001 --> 00:45:24,000 used to be and on a particular date. So I wonder how many people in this room are 525 00:45:24,000 --> 00:45:28,000 you know we're around when the White House dot gov site looked like 526 00:45:28,000 --> 00:45:30,000 this. I do remember it actually. 527 00:45:30,001 --> 00:45:35,001 That was from 1997. So things have changed a bit but we preserve it for all time 528 00:45:35,001 --> 00:45:42,001 on the internet archive. There are so many resources. I mean we actually it's 529 00:45:42,001 --> 00:45:49,001 25 petabytes of data that we have and I'm told that's 15 zeros after the 25. 530 00:45:50,001 --> 00:45:54,001 So it's a lot of stuff and you could spend a lot of time. 531 00:45:55,001 --> 00:46:01,000 One of the really really great resources we have is the TV news broadcast library 532 00:46:01,001 --> 00:46:07,001 and there you can search through closed captioning clips of any news broadcast 533 00:46:07,001 --> 00:46:12,001 back to 2009 and when I first used the site I went immediately to my favorite 534 00:46:12,001 --> 00:46:17,001 quote of all time which is Senator John Corzine former Senator John Corzine who 535 00:46:17,001 --> 00:46:21,001 said when he was called before Congress to talk about MF Global he said mistakes 536 00:46:21,001 --> 00:46:26,001 will continue to happen in the course of human events which is just a wonderful 537 00:46:26,001 --> 00:46:30,000 you know thing you can use around the house as an excuse. 538 00:46:32,001 --> 00:46:36,000 So you can find this clip and what's really wonderful you can't really see it 539 00:46:36,000 --> 00:46:42,000 here it's a little small but you can actually edit this clip you can embed it you 540 00:46:42,000 --> 00:46:48,000 can share it on social media so whether you're a journalist and you want to embed 541 00:46:48,000 --> 00:46:53,001 a quote that you're referring to from a misfigure in your story or you just want 542 00:46:53,001 --> 00:46:58,000 to get something out on Twitter we're there for you and you can find all kinds of 543 00:46:58,000 --> 00:47:04,000 interesting stuff there. So we have a number of things that we are doing for the 544 00:47:04,000 --> 00:47:10,000 2016 primary elections of which the political ad archive is one but there are a 545 00:47:10,000 --> 00:47:16,001 few other things too but well we'll talk about the political ad archive 546 00:47:16,001 --> 00:47:19,001 first so it's data fueled analysis. 547 00:47:20,000 --> 00:47:23,001 Dan talked a little bit about the metadata that we're naturally gathering in the 548 00:47:23,001 --> 00:47:29,001 process of sucking in these tv ads and counting them. We are going to eight 549 00:47:29,001 --> 00:47:35,001 primary states that are key and we are recording 20 markets sifting through 20 550 00:47:35,001 --> 00:47:40,000 markets we've already gone through more than a hundred thousand hours of time to 551 00:47:40,000 --> 00:47:43,000 find these ads and you know that number is just going to get bigger and bigger 552 00:47:43,000 --> 00:47:49,000 and bigger. We're of course starting with Iowa and New Hampshire where we're 553 00:47:49,000 --> 00:47:52,001 gathering ads and we're going to we have this is a graphic on the new website 554 00:47:52,001 --> 00:47:59,000 which will be continually updated to show where the ads are coming from. This is 555 00:47:59,000 --> 00:48:03,001 the way the new website will look when you open it up political ad archive 556 00:48:03,001 --> 00:48:05,000 .org if I haven't said that already. 557 00:48:07,001 --> 00:48:10,001 And when you get in there one thing we are doing it's called we 558 00:48:10,001 --> 00:48:12,000 call it a canonical ad page. 559 00:48:12,001 --> 00:48:17,001 For every ad we discover it's going to get its own web page and you will be able 560 00:48:17,001 --> 00:48:21,001 to see on that web page the video of the ad similar to the tv news broadcast 561 00:48:21,001 --> 00:48:26,001 library it'll be the whole clip and you can edit it embed it do as you wish with 562 00:48:26,001 --> 00:48:31,000 it so you can actually see the ad and then we will be linking to all the great 563 00:48:31,000 --> 00:48:35,000 fact checking and follow the money information that our partners produce on that 564 00:48:35,000 --> 00:48:41,001 web page in one place. But even more we have these data that we get to collect 565 00:48:41,001 --> 00:48:47,001 along the way the actual counts so you know you'll be able to see under the ad 566 00:48:47,001 --> 00:48:52,000 how many times it's aired where and when so it's you know if you go back to the 567 00:48:52,000 --> 00:48:55,001 classic reporting who what why where when we're going to give you who aired the 568 00:48:55,001 --> 00:49:00,000 ad who's targeted in the ad in other words which candidates are mentioned and who 569 00:49:00,000 --> 00:49:04,000 sponsored the ad the name of the group the committee to save puppies or whatever 570 00:49:04,000 --> 00:49:09,001 it is what the ad itself is you can view it and embed it where the ad aired on 571 00:49:09,001 --> 00:49:15,001 which station and in which town or city and when exactly that aired how often it 572 00:49:15,001 --> 00:49:21,000 aired and during what shows and we actually are also using PolitiFact's subject 573 00:49:21,000 --> 00:49:25,001 index to track the subjects covered in the ad and also we're recording the 574 00:49:25,001 --> 00:49:30,001 message of the ad whether it's it's generally a pro ad just entirely pro the 575 00:49:30,001 --> 00:49:36,000 candidate or an entirely against the candidate or whether it's mixed message like 576 00:49:36,000 --> 00:49:40,000 it's saying this candidate is horrible and I'm great or I'm great and this 577 00:49:40,000 --> 00:49:44,001 candidate is horrible or those five candidates are really bad but I'm the best so 578 00:49:44,001 --> 00:49:48,000 we'll have that information too and we're going to be continually adding to that 579 00:49:48,000 --> 00:49:53,001 list of data we're collecting and it's all downloadable in a spreadsheet a CSV 580 00:49:53,001 --> 00:49:59,001 spreadsheet so that you can see again all this information organized sorry or an 581 00:49:59,001 --> 00:50:06,001 API so but what we are we're the library so we are 582 00:50:06,001 --> 00:50:12,000 not interpreting what we're seeing we are presenting it and it's up to you the 583 00:50:12,000 --> 00:50:16,001 public the journalists the researchers to explain what it's about and our 584 00:50:16,001 --> 00:50:21,000 partners of course so that's what we hope when we bring all this information 585 00:50:21,000 --> 00:50:24,001 together we hope that all kinds of things will happen with it that we can't even 586 00:50:24,001 --> 00:50:30,000 anticipate that will bring what we like to say is light to the heat of the 587 00:50:30,000 --> 00:50:37,000 campaign so again it's political ad archive. org launches today it's a growing 588 00:50:37,000 --> 00:50:43,001 resource it's we have more than 30,000 errings of ads now 589 00:50:43,001 --> 00:50:48,001 documented but again it's going to grow and grow it's a living breathing thing so 590 00:50:48,001 --> 00:50:52,001 keep checking in and it's going to get better and better but we also have a few 591 00:50:52,001 --> 00:50:58,001 other tools that we we have that are really useful in analyzing this coming 592 00:50:58,001 --> 00:51:05,000 election and Roger touched on a few of these briefly but we have something 593 00:51:05,000 --> 00:51:11,001 called the um well Caleb Leatrew who is in the audience today 594 00:51:11,001 --> 00:51:17,001 um we work with closely an engineer and a social scientist um and he has 595 00:51:17,001 --> 00:51:21,000 developed something called the candidate name tracker so what this is is it's 596 00:51:21,000 --> 00:51:26,000 using our tv news broadcast data to figure out well how many times does someone 597 00:51:26,000 --> 00:51:32,001 say the word Donald Trump in a tv news show and Hillary Clinton and you know Ben 598 00:51:32,001 --> 00:51:38,000 Carson and so it quantifies that and if you go into this link which you know we 599 00:51:38,000 --> 00:51:42,001 can provide um you can actually you know change it around so you can see you can 600 00:51:42,001 --> 00:51:46,001 see whether it's all data look at the 100 days seven days um and see how their 601 00:51:46,001 --> 00:51:52,000 mentions go up and down with the news um you know this is kind of a good measure 602 00:51:52,000 --> 00:51:57,000 of earned media when a candidate gets a lot of attention in the news for whatever 603 00:51:57,000 --> 00:52:03,000 reason um and it can be really useful like we did this chart right after the 604 00:52:03,000 --> 00:52:08,001 first Trump ad well actually I correct myself the Trump ad on immigration had not 605 00:52:08,001 --> 00:52:12,001 yet aired Trump had simply announced that he was going to spend two million a 606 00:52:12,001 --> 00:52:18,001 week on campaign ads and you could see how his blue line here shot up um just you 607 00:52:18,001 --> 00:52:23,000 know he's a master in getting the earned media so this is a way that you can 608 00:52:23,000 --> 00:52:29,001 track that kind of thing so that can be really useful another thing that Caleb 609 00:52:29,001 --> 00:52:36,000 built and which is wonderful and working with us is um the debate tracker it's 610 00:52:36,000 --> 00:52:41,001 it's um a way to um see which phrases from a particular political debate 611 00:52:42,001 --> 00:52:47,001 were repeated the most on tv news so this is a slide from that first democratic 612 00:52:47,001 --> 00:52:52,000 debate where Bernie Sanders talked about those damn emails and uh you could see 613 00:52:52,000 --> 00:52:56,001 that that shot up that's what that big line is and so that's another way again to 614 00:52:56,001 --> 00:53:02,001 see well not just what was said in the debates but how is that reflected um in 615 00:53:02,001 --> 00:53:09,001 how we consume the news so uh that's the project again this is 616 00:53:09,001 --> 00:53:14,001 my contact information uh political ad archive. org is the website and we hope you 617 00:53:14,001 --> 00:53:19,001 will visit early and often and keep coming back we're also going to have a blog 618 00:53:19,001 --> 00:53:24,001 there where we track a lot of the you know great journalism that comes out about 619 00:53:24,001 --> 00:53:30,000 the ads and we'll get that up there along with our partner information so um i am 620 00:53:30,000 --> 00:53:33,001 now going to turn it back over to Roger we're going to see if anybody has any 621 00:53:33,001 --> 00:53:40,001 questions. Thank you so we have a a 622 00:53:40,001 --> 00:53:47,001 mic and if you've got any questions hold the mic close to your mouth and let us 623 00:53:47,001 --> 00:53:52,000 know your name and affiliation and what's the question and we'll have any of 624 00:53:52,000 --> 00:53:53,001 us do our best to answer it. 625 00:53:59,000 --> 00:54:05,000 Hi Craig Holman with Public Citizen. I'm trying to figure out how extensive the 626 00:54:05,000 --> 00:54:11,001 political ad tv database is uh you know from CMAG they've got a they use 627 00:54:11,001 --> 00:54:18,000 satellite technology so they pretty well capture all campaign ads that air on 628 00:54:18,000 --> 00:54:25,000 television how is it you decide which ads to capture how extensive is 629 00:54:25,000 --> 00:54:31,000 your campaign ad database does that make sense? Oh makes a great deal of sense 630 00:54:31,000 --> 00:54:38,000 absolutely um so CMAG a section of cantar media is a commercial 631 00:54:38,000 --> 00:54:44,000 organization who just not in advertising but from our data perspective produces 632 00:54:44,000 --> 00:54:50,001 what is currently probably the most comprehensive um they do a great job 633 00:54:50,001 --> 00:54:55,001 tracking ads and other kinds of elements of of television we're not a competitor 634 00:54:55,001 --> 00:55:02,001 to them by any means we are hacking in an open way at treating media as 635 00:55:02,001 --> 00:55:07,000 data looking at ads and other kinds of attributes allowing people to 636 00:55:07,000 --> 00:55:12,001 experimentally see how they can explore those and visualize those and interpret 637 00:55:12,001 --> 00:55:19,000 those in an open library so if we were to look at comprehensiveness and 638 00:55:19,000 --> 00:55:26,000 widespread coverage CMAG is from my personal perspective is clearly the best 639 00:55:26,000 --> 00:55:32,000 we're a new experiment an experiment that's trying to work with you just like 640 00:55:32,000 --> 00:55:38,001 we're a public library we're a public library for you on this issue could I 641 00:55:38,001 --> 00:55:45,001 refine my question yes please I mean like CMAG uses uh through their satellite 642 00:55:45,001 --> 00:55:51,000 technology regular television stations will send a signal saying okay here comes 643 00:55:51,000 --> 00:55:56,000 an ad and then it picks that up so it picks up just about all the ads that way 644 00:55:56,000 --> 00:56:02,000 how do you identify which ads you uh will put in your database do you have people 645 00:56:02,000 --> 00:56:07,001 like watching television in the markets I'd like to to bump this to Dan but 646 00:56:07,001 --> 00:56:12,000 there's somebody from CMAG in the audience who also might address some of the 647 00:56:12,000 --> 00:56:17,001 magic behind what they do which is not exactly as you assume it is a it's even 648 00:56:17,001 --> 00:56:23,000 more complex and challenging than that would it were that that uh that it was as 649 00:56:23,000 --> 00:56:29,001 easy as that in fact there's lots of folk hacking at how you reliably 650 00:56:29,001 --> 00:56:36,000 algorithmically have computers and machines doing this work but allow me to bump 651 00:56:36,000 --> 00:56:42,001 back to our senior engineer Dan Schultz sure so right now with what we're 652 00:56:42,001 --> 00:56:48,000 launching today finding the first copy of an ad is a is a pretty manual process 653 00:56:48,000 --> 00:56:52,000 and that was what I was getting at about over the next couple months where 654 00:56:52,000 --> 00:56:59,000 basically the tool has the ability to find any repeated content whether we 655 00:56:59,000 --> 00:57:03,000 flagged it or not and then surface that to a human when it's been repeated enough 656 00:57:03,000 --> 00:57:07,000 times to be oh this is probably a political ad or this is probably an ad in 657 00:57:07,000 --> 00:57:14,000 general and so this the concept of discovery in in a semi-automated way is 658 00:57:14,000 --> 00:57:20,001 coming it's not there yet but the the the pathway there would be as 659 00:57:20,001 --> 00:57:26,000 we're seeing pieces of content that show up again and again and again all that 660 00:57:26,000 --> 00:57:31,000 would take is a human saying that is a political ad that is a Kraft macaroni and 661 00:57:31,000 --> 00:57:36,000 cheese ad right but the once you've picked out a political ad whether it was you 662 00:57:36,000 --> 00:57:42,000 watched tv and you saw it and flagged it or the dupletron discovered it and 663 00:57:42,000 --> 00:57:46,001 exposed it once it's in the system as an ad it gets added to the political ad 664 00:57:46,001 --> 00:57:50,001 archive does does that clarify and make sense and you go through the back line 665 00:57:50,001 --> 00:57:57,000 say that oh and that yes that process we have 100 you know we have 100 666 00:57:57,000 --> 00:58:03,001 000 videos in there now and so that repetition discovery process would go through 667 00:58:03,001 --> 00:58:10,001 that backlog as well hi my name is Michael Beckle and I'm a reporter at 668 00:58:10,001 --> 00:58:15,000 the Center for Public Integrity I wanted to sort of follow up on Craig's question 669 00:58:15,000 --> 00:58:22,000 if the process is working well and the ads are identified it looks like you 670 00:58:22,000 --> 00:58:29,000 know the 20 major markets that you all are focused on and just to make sure 671 00:58:29,000 --> 00:58:33,001 I'm understanding that so if you know the ad airs in Iowa it's going to be 672 00:58:33,001 --> 00:58:38,001 captured but if the same ad aired in Minnesota it might not be is that basically 673 00:58:38,001 --> 00:58:44,001 that's true yeah if it's and and so then within within those states you know 674 00:58:44,001 --> 00:58:50,000 these key battleground states like Iowa, New Hampshire, Florida, Ohio do you have 675 00:58:50,000 --> 00:58:55,001 a sense of you know have you captured you know 98 percent of ads that you think 676 00:58:55,001 --> 00:59:00,001 are airing or 90 percent or you know the extent to which these markets that you 677 00:59:00,001 --> 00:59:07,000 pick do relate to the viewers in those states receiving those messages if that 678 00:59:07,000 --> 00:59:14,000 makes sense we don't know yet we've just flown open the doors and first time 679 00:59:14,000 --> 00:59:18,001 this has ever been done in this kind of public fashion so we'll be looking at the 680 00:59:18,001 --> 00:59:25,000 data to see and exposing it all to get a sense of how comprehensive we are what 681 00:59:25,000 --> 00:59:30,001 we've missed we have noted that you know there's a couple versions of of ads one 682 00:59:30,001 --> 00:59:36,001 where it may be the same voice actor with a lower timber or it's a different 683 00:59:36,001 --> 00:59:42,001 voice actor our system is accurate enough to pick up that this is a this is a 684 00:59:42,001 --> 00:59:48,000 this is different so that was a surprise to us and we're we expect and as you can 685 00:59:48,000 --> 00:59:51,001 imagine we welcome those sorts of surprises 686 00:59:59,000 --> 01:00:05,000 Hi Tom Broan with Newsday here are you going to go back and double check against 687 01:00:05,000 --> 01:00:11,000 various filings with the FEC and the FCC to see if you've missed any ads or to 688 01:00:11,000 --> 01:00:17,000 double check your coverage so we're relying on our partners who are the experts 689 01:00:17,000 --> 01:00:22,000 in doing that to get that kind of information we're the as part of the project we 690 01:00:22,000 --> 01:00:26,001 are not yet anyway integrating the reports to the federal communications 691 01:00:26,001 --> 01:00:31,001 commission on ad contract buys the the you know actual dollar amounts being spent 692 01:00:31,001 --> 01:00:36,001 where we're going to start worked first with just quantifying these counts and 693 01:00:36,001 --> 01:00:41,001 but I know that there are various projects underway at some of our partners to 694 01:00:41,001 --> 01:00:46,001 try to do that and as that comes online we hope to integrate things like that I 695 01:00:46,001 --> 01:00:52,001 don't know if any of them want to address that perhaps Robert so we presently are 696 01:00:52,001 --> 01:00:59,000 at a point where we have a full database of I believe it's about approaching 400 697 01:00:59,000 --> 01:01:06,000 ,000 records from the federal communications commission we have set up a procedure 698 01:01:06,000 --> 01:01:12,001 to standardize the records we have scraped entire pdfs for I don't have an 699 01:01:12,001 --> 01:01:18,001 exact count of it but it's in the very large tens of thousands so that those 700 01:01:18,001 --> 01:01:24,001 documents are searchable um and the aim here is to be able to 701 01:01:24,001 --> 01:01:31,000 update that data on a daily basis assign it a group assign it an id 702 01:01:31,000 --> 01:01:36,000 whether it's an employee identification number for a 501c organization or a 703 01:01:36,000 --> 01:01:43,000 committee id for a federal election commission reporting entity and that way 704 01:01:43,000 --> 01:01:50,000 that the aim there is to be able to track with the ads that we're seeing in 705 01:01:50,000 --> 01:01:55,001 the political ad tracker but also to see how that meshes with federal election 706 01:01:55,001 --> 01:02:01,001 commission data and so we're at a point where this is is very close to being at 707 01:02:01,001 --> 01:02:07,001 least something that we can update internally on a daily basis and and provide in 708 01:02:07,001 --> 01:02:14,000 some manner on the site whether that's at least just going to you know if you see 709 01:02:14,000 --> 01:02:19,001 an ad by patriot majority that doesn't have a federal election commission report 710 01:02:19,001 --> 01:02:25,001 you can at least come to us and say here by date by station are the reports that 711 01:02:25,001 --> 01:02:31,000 patriot majority or cautious gps or any of these groups have filed with the fcc 712 01:02:31,001 --> 01:02:38,001 i'd like to this is a great question and one of the great opportunities that we 713 01:02:38,001 --> 01:02:44,000 have is to provide some ground truth on just what the stations are reporting in 714 01:02:44,000 --> 01:02:49,000 their so-called political files the public inspection files which they're now 715 01:02:49,000 --> 01:02:54,001 required to upload to the fcc alas they can upload them as pdfs doesn't even have 716 01:02:54,001 --> 01:03:00,001 to be text they can be pictures they can be um askew they can be fuzzy um but 717 01:03:00,001 --> 01:03:06,000 this is the one of the first chances that a a bunch of people up here and 718 01:03:06,000 --> 01:03:09,000 throughout the that are represented on the stage and throughout the country and 719 01:03:09,000 --> 01:03:15,000 in universities and other places are working to crack those pdfs open make them 720 01:03:15,000 --> 01:03:19,000 structured data join them with the open data that we're creating and 721 01:03:19,000 --> 01:03:24,000 interestingly enough in this room two people to your right is gray gallon who is 722 01:03:24,000 --> 01:03:30,001 the the fcc's first chief data officer who implemented um just that that database 723 01:03:30,001 --> 01:03:37,000 so i'm sure he's available for more on uh on how it works and what its challenges 724 01:03:37,000 --> 01:03:42,001 are and what its opportunities are for the future could i add a little to that as 725 01:03:42,001 --> 01:03:48,001 well um one really um i think that is exciting to me because i am one of those 726 01:03:48,001 --> 01:03:53,001 people who's wrestled with those data and know how interesting but also 727 01:03:53,001 --> 01:03:57,001 frustrating they can be is that we are capturing these ads as they actually air 728 01:03:57,001 --> 01:04:01,001 as roger says so not everything gets reported to the fcc not everything gets 729 01:04:01,001 --> 01:04:06,001 reported to the fcc um you know there's a lot of rules about what does what 730 01:04:06,001 --> 01:04:11,001 doesn't um i've been in the position where political you know a local broadcast 731 01:04:11,001 --> 01:04:15,001 station will say well i don't have to post those files because this wasn't an 732 01:04:15,001 --> 01:04:20,000 issue of quote unquote national importance so there's a lot of loopholes so it's 733 01:04:20,000 --> 01:04:24,001 great to look at those things but what we're going to be doing is unearthing what 734 01:04:24,001 --> 01:04:29,000 actually airs and helping folks identify those and then you guys can start 735 01:04:29,000 --> 01:04:32,000 running with that and saying wait a minute they didn't even report this to the 736 01:04:32,000 --> 01:04:37,000 fcc or the fcc or whatever government agencies theoretically should have records 737 01:04:37,000 --> 01:04:44,000 or conversely we have to say we can also people can also say they did report and 738 01:04:44,000 --> 01:04:49,000 they were crazy accurate and how wonderful is that to not only bring critique 739 01:04:49,000 --> 01:04:55,000 against those who don't but bring approval and and and highlight those who are 740 01:04:55,000 --> 01:04:59,001 responsibly doing a great job because there are people out there who do and we 741 01:04:59,001 --> 01:05:04,000 think that they deserve attention if only to credit what they've done but also to 742 01:05:04,000 --> 01:05:08,001 create a positive feedback loop so others go maybe we better too because not only 743 01:05:08,001 --> 01:05:15,000 might we get uh a slight spanking and look bad but we look good and i want to we 744 01:05:15,000 --> 01:05:19,000 want to join these stations that also look good i just wanted to add one more 745 01:05:19,000 --> 01:05:24,000 note about the fcc we submitted our comments to the fcc encouraging them they'll 746 01:05:24,000 --> 01:05:30,000 have a vote next week i believe it is um to possibly expand the the politic the 747 01:05:30,000 --> 01:05:36,001 public file to satellite radio and cable so we'd be getting a lot bigger chunk of 748 01:05:36,001 --> 01:05:42,000 the data if that goes through um and i'm gonna pass so i actually have a question 749 01:05:42,000 --> 01:05:48,000 for you guys um and it's about uh the the two-way street opportunities for the 750 01:05:48,000 --> 01:05:55,000 public so say you know i live in ohio i see an ad uh can you walk me 751 01:05:55,000 --> 01:06:02,000 through the process whereby um i could go to your website see 752 01:06:02,000 --> 01:06:06,001 if the ad's there and if it isn't there is there an opportunity for me to flag it 753 01:06:06,001 --> 01:06:13,000 for you absolutely we want that and thanks for bringing that up um 754 01:06:13,000 --> 01:06:17,001 yes you can reach us in a number of ways um depending on how comfortable you are 755 01:06:17,001 --> 01:06:24,000 with things you can email us at the i had the well i'm nancy w at archive.org if 756 01:06:24,000 --> 01:06:29,000 you spot something um if you see something in our tv broadcast news library i 757 01:06:29,000 --> 01:06:34,000 think i'll ask dan how best someone can mark that so that we know it's there do 758 01:06:34,000 --> 01:06:39,000 you sure i don't know if um i think probably for now the best that would be to to 759 01:06:39,000 --> 01:06:44,000 tweet at us and let us know yeah okay where you were and and and what time you 760 01:06:44,000 --> 01:06:47,000 were was like the general time you were watching maybe what channel if you do 761 01:06:47,000 --> 01:06:53,001 that we'll find it um it's at pellet ad archive.org or i'm also 762 01:06:53,001 --> 01:07:00,001 at at n watsman w a t z m a n uh yes calif yeah so just i just want 763 01:07:00,001 --> 01:07:04,000 to make a quick comment like the questions about like comprehensiveness for me as 764 01:07:04,000 --> 01:07:08,001 a you know as an outsider as a data scientist um you know the way i kind of look 765 01:07:08,001 --> 01:07:12,000 at is as any library it's it's not going to get absolutely everything that's out 766 01:07:12,000 --> 01:07:16,001 there but because what they're doing is open data um for someone like me it's 767 01:07:16,001 --> 01:07:19,001 really exciting because like the question of well how could someone flag that you 768 01:07:19,001 --> 01:07:22,000 know to me it's like well you know someone else would come along say hey let's 769 01:07:22,000 --> 01:07:25,001 make a little shazam like app where you see a political ad you know you could run 770 01:07:25,001 --> 01:07:28,001 it and and with the apis that they're providing and the way in which they're 771 01:07:28,001 --> 01:07:32,001 exposing that data some a developer like me could come along build an app like 772 01:07:32,001 --> 01:07:35,001 that and make that available to the public um or you know type of stuff i do 773 01:07:35,001 --> 01:07:39,000 which involves you know deep learning image recognition the you know the 774 01:07:39,000 --> 01:07:42,001 opportunities to be able to take these ads and process them and say well you know 775 01:07:42,001 --> 01:07:46,001 what ads feature people smiling or frowning or you know doing uh you know 776 01:07:46,001 --> 01:07:50,000 different types of analysis and music so to someone like me it's like yeah it may 777 01:07:50,000 --> 01:07:53,000 not be as comprehensive it may not get everything that's out there by by a long 778 01:07:53,000 --> 01:07:57,000 shot but you know again to someone like me i see this is just starting uh and so 779 01:07:57,000 --> 01:08:02,000 the opportunities i think i think to me it's really about taking the the world of 780 01:08:02,000 --> 01:08:05,001 media into that data world that people might like me live in and creating those 781 01:08:05,001 --> 01:08:11,000 bridges so that we can you know process and do incredible stuff with it before 782 01:08:11,000 --> 01:08:16,001 you before you hand the mic over kalov could you introduce yourself and what's 783 01:08:16,001 --> 01:08:22,000 your filly i'm uh kalovli taru um and i do a lot of big data big data work 784 01:08:28,000 --> 01:08:32,001 actually just two quick questions what's the turnaround when an ad first appears 785 01:08:32,001 --> 01:08:36,001 and when it'll show up on your site and the second thing is is that did i hear 786 01:08:36,001 --> 01:08:41,001 this right you're not including cable you're just doing uh public uh or or 787 01:08:41,001 --> 01:08:48,000 commercial airwaves or whatever um so we are our our goal is to get ads up at 788 01:08:48,000 --> 01:08:53,001 least 24 hours after they air um we as roger said we're just flipping the switch 789 01:08:53,001 --> 01:08:58,000 today um we expect that we'll get better as time goes on we'd like to be quicker 790 01:08:58,000 --> 01:09:04,000 than that we may not hit that 24-hour goal immediately on everything but that's 791 01:09:04,000 --> 01:09:07,001 the goal and hopefully we'll do better and yes right now we are not tracking 792 01:09:07,001 --> 01:09:12,001 cable um you know as as with everything we would love to do everything we're 793 01:09:12,001 --> 01:09:19,000 starting where we can start and how far back do you have now um so we're we are 794 01:09:19,000 --> 01:09:23,001 in each of the states we're tracking and there's a list of those eight states and 795 01:09:23,001 --> 01:09:29,001 the materials um we're trying to get two months of television proceeding where 796 01:09:29,001 --> 01:09:33,001 whether it's the caucus or an election so we've been recording in iowa new 797 01:09:33,001 --> 01:09:37,000 hampshire for quite some time and we're moving on to other states such as 798 01:09:37,000 --> 01:09:43,000 colorado nevada south carolina north carolina florida ohio uh 799 01:09:43,000 --> 01:09:48,000 virginia i think that might be i think that's the list yes tracy or 800 01:09:51,001 --> 01:09:56,000 uh i just wanted to point out that we are tracking about 20 to 30 uh cable 801 01:09:56,000 --> 01:10:00,001 channels in san francisco so if it did have national coverage an ad we would 802 01:10:00,001 --> 01:10:05,000 actually end up uh tracking and fingerprinting it but but you're but the answer's 803 01:10:05,000 --> 01:10:09,001 generally right for the markets we're tracking um also this is sort of a a lead 804 01:10:09,001 --> 01:10:14,001 but nancy could you explain a little bit about um how we're manually finding ads 805 01:10:14,001 --> 01:10:18,001 now by maybe subscribing to feeds and things like that and looking at uh things 806 01:10:18,001 --> 01:10:23,000 that are getting pushed from campaigns because yes i mean right now we're doing a 807 01:10:23,000 --> 01:10:28,001 combination as dan said we we have both the human and uh the duplet ron process 808 01:10:28,001 --> 01:10:32,001 right so the human processes again i mentioned there's an army back in san 809 01:10:32,001 --> 01:10:39,000 francisco and in other places um so folks are actually scouring the uh public 810 01:10:39,000 --> 01:10:44,001 channels on youtube and other spots where candidates post ads or super PACs post 811 01:10:44,001 --> 01:10:48,001 ads etc so when we see them there we're loading them into our site so we can 812 01:10:48,001 --> 01:10:52,001 create those fingerprints and once we create a fingerprint again we're golden 813 01:10:52,001 --> 01:10:58,001 except when in those very interesting situations where we don't find matches 814 01:10:58,001 --> 01:11:03,001 because the ad as aired is a little bit different in sound quality than the ad as 815 01:11:03,001 --> 01:11:07,001 posted but those in themselves can become stories uh the one that really brought 816 01:11:07,001 --> 01:11:13,000 that to our attention was actually the trump immigration ad um the voice uh that 817 01:11:13,000 --> 01:11:18,000 we captured as we like to say in the wild from where it aired on tv uh the the 818 01:11:18,000 --> 01:11:22,000 voice was i believe lower the narrator's voice was substantially lower at least 819 01:11:22,000 --> 01:11:27,000 enough lower that that it did not match to the youtube copy we had but then we 820 01:11:27,000 --> 01:11:30,000 were able to find a copy i mean because we had the copy in the wild we can create 821 01:11:30,000 --> 01:11:34,000 that fingerprint now we're hoping what that will lead to is really great things 822 01:11:34,000 --> 01:11:39,000 where we can distinguish between ads that are earned media and ads that are paid 823 01:11:39,000 --> 01:11:43,001 media because when broadcast news errors those ads often they'll use the youtube 824 01:11:43,001 --> 01:11:50,001 copy so um that way we're not quite there yet but hopefully we'll be able to 825 01:11:50,001 --> 01:11:56,000 distinguish between that and tell folks oh you know this ad um as a paid ad aired 826 01:11:56,000 --> 01:12:02,001 this many times and it was also repeated on this many news broadcasts just to 827 01:12:02,001 --> 01:12:07,000 clarify um a little quotidian point but we're librarians and you're journalists 828 01:12:07,000 --> 01:12:13,001 so we're all comfortable here um if the uh the broadcast in the wild ad was a 829 01:12:13,001 --> 01:12:20,000 higher timber and the uh the youtube ad as posted the first uh was 830 01:12:20,000 --> 01:12:23,000 a um a lower register 831 01:12:27,001 --> 01:12:34,000 uh is there any more questions yeah michael bechol again 832 01:12:34,000 --> 01:12:38,000 from the center for public integrity as a reporter you know wanting to stay on 833 01:12:38,000 --> 01:12:43,001 top of what's new and uh you know whether you know i'm following a particular 834 01:12:43,001 --> 01:12:48,000 campaign or trying to just focus on ads in a particular state what tools are 835 01:12:48,000 --> 01:12:52,001 available through this new website you know the sunlight foundation for instance 836 01:12:52,001 --> 01:12:56,001 has different alerts and things that i can subscribe to are there going to be 837 01:12:56,001 --> 01:13:00,000 features like that or is there you know one landing page that you know i should 838 01:13:00,000 --> 01:13:04,000 go back to over and over that kind of thing so the landing page political ad 839 01:13:04,000 --> 01:13:08,000 archive.org our twitter feed are good to monitor for now um as one of the 840 01:13:08,000 --> 01:13:11,001 features we hoped out in the future we do hope to build an alert service of some 841 01:13:11,001 --> 01:13:16,000 sort you know whether it's something you subscribe to or some sort of feed that 842 01:13:16,000 --> 01:13:20,000 that you get to see the latest ads aired but um for right now the best place to 843 01:13:20,000 --> 01:13:24,000 do that is to go the website dan do you have anything to add all i'll say is 844 01:13:24,000 --> 01:13:30,000 reach i mean when i mentioned if anybody wants to collaborate or wants to use 845 01:13:30,000 --> 01:13:36,001 this stuff the data or the code reach out to me uh at slifty on twitter 846 01:13:36,001 --> 01:13:41,001 um reach out to nancy on her handle or the or the archive the point is this is a 847 01:13:41,001 --> 01:13:45,001 living breathing project so when you have something that would be useful to you 848 01:13:46,001 --> 01:13:49,001 we need i we need to know about it because then we can 849 01:13:49,001 --> 01:13:51,001 build something around it so 850 01:13:56,001 --> 01:14:01,001 have have any of the candidates reacted to what you're doing or do they even know 851 01:14:01,001 --> 01:14:08,000 what you're doing not yet probably as i said we've just thrown open the doors 852 01:14:08,000 --> 01:14:15,000 today okay i do have one more quick question the downloadable database 853 01:14:15,000 --> 01:14:21,000 does that include the frequency of the airings and the markets yeah that is one 854 01:14:21,000 --> 01:14:27,001 line per airing so every if you download the csv every row in that in that 855 01:14:27,001 --> 01:14:33,001 table corresponds to a single airing on a single channel so you can know the time 856 01:14:33,001 --> 01:14:37,000 that that airing happened the channel that happened the location and the and the 857 01:14:37,000 --> 01:14:41,001 market among other things about the ad in general if you go to the website there 858 01:14:41,001 --> 01:14:46,000 is a tab that says data and it has the list of the data we provide and 859 01:14:46,000 --> 01:14:52,001 definitions of what should be in the csv hi thank you very much for doing this 860 01:14:52,001 --> 01:14:58,000 i'm don kaley i'm a columnist for a newspaper in ugine oregon my question is to 861 01:14:58,000 --> 01:15:04,001 the fact checkers how long will it be before those that put up the ads no longer 862 01:15:04,001 --> 01:15:11,001 volunteer the data that they're drawing from and just shrug at 863 01:15:11,001 --> 01:15:15,001 you and say we're not going to tell you where we got that number you go figure it 864 01:15:15,001 --> 01:15:22,000 out well i mean if the past is any explanation 865 01:15:22,000 --> 01:15:28,001 or the the general response that we've seen before 866 01:15:29,000 --> 01:15:35,000 is that the campaigns actually designate somebody to respond to our 867 01:15:35,000 --> 01:15:41,001 background information requests now that could change but 868 01:15:41,001 --> 01:15:46,001 in general they have adapted to us 869 01:15:46,001 --> 01:15:53,000 normally the campaigns in the past have been pretty good at putting 870 01:15:53,000 --> 01:15:59,000 together the the backup and you know it'll be kind of like a chart thing where 871 01:15:59,000 --> 01:16:03,000 it's got the the quote from the ad on the left and then on the right are 872 01:16:03,000 --> 01:16:10,000 citations of what backs that up although i would have to say not so much on ads 873 01:16:10,000 --> 01:16:16,001 but on claims in general we we have some campaigns that are very very responsive 874 01:16:16,001 --> 01:16:22,000 to us and some campaigns that are not responsive at all and then some that are 875 01:16:22,000 --> 01:16:26,000 somewhere in the middle sometimes responding sometimes not and we're really just 876 01:16:26,000 --> 01:16:33,000 now starting to get into the thick of the ads so it'll be interesting to see how 877 01:16:33,000 --> 01:16:39,000 well they are doing those backup reports and and sending them to us yes i totally 878 01:16:39,000 --> 01:16:44,000 agree and for example the rubio ad that i talked about one of the claims the one 879 01:16:44,000 --> 01:16:49,000 about the guns they only pointed to the executive action announcement but nothing 880 01:16:49,000 --> 01:16:54,001 else and it was up to us to give the fuller picture so we work with the backup 881 01:16:54,001 --> 01:17:00,000 that they send and then do the rest of the reporting to complement that just uh 882 01:17:00,000 --> 01:17:03,001 just to be clear there's nothing that they're they're not required to give you 883 01:17:03,001 --> 01:17:08,001 any backup i mean there's no regulations or anything like that i mean at some 884 01:17:08,001 --> 01:17:14,000 point they could just shrug and say go find it they already do you know there's a 885 01:17:14,000 --> 01:17:19,000 campaign that once told me it's not up to the speaker to prove that we're right 886 01:17:19,000 --> 01:17:24,001 and i said no it actually is i'd rather not right now 887 01:17:24,001 --> 01:17:31,001 in in in that vein i thought the campaigns did have a self-interest because 888 01:17:31,001 --> 01:17:38,000 the tv stations can be challenged about false advertising and if they're not 889 01:17:39,000 --> 01:17:45,000 airing true ads or true enough ads that they can be pulled uh is that not a 890 01:17:45,000 --> 01:17:50,000 regulation it is but it yeah it's very difficult for you know so many things are 891 01:17:50,000 --> 01:17:55,001 misleading and that so it kind of becomes a little bit of a gray area and it you 892 01:17:55,001 --> 01:18:00,001 know it's yeah i don't i mean things that are qualified it has to be really 893 01:18:00,001 --> 01:18:06,000 really false i guess is what i'm saying to see a local station pull the ad and it 894 01:18:06,000 --> 01:18:09,001 does i don't know what the data is on that maybe somebody else knows but it does 895 01:18:09,001 --> 01:18:14,001 not happen very often can 896 01:18:14,001 --> 01:18:21,000 i ask a question of the fact checkers one thing that 897 01:18:21,000 --> 01:18:26,001 i found interesting looking at nonprofits that aren't actually non-profits is 898 01:18:26,001 --> 01:18:32,000 that when you look at they're often saying you know our purpose is to educate the 899 01:18:32,000 --> 01:18:38,000 public we are we're an educational organization and we've on 900 01:18:38,000 --> 01:18:44,000 many occasions used the fact checkers reports to say you know this well-funded 901 01:18:44,000 --> 01:18:49,000 organization that's running thousands of ads is running false ads repeatedly and 902 01:18:49,000 --> 01:18:52,000 we're pointing to the research that y'all are doing i'm wondering what their 903 01:18:52,000 --> 01:18:56,000 response to you because the response from an outside group and it doesn't have to 904 01:18:56,000 --> 01:19:03,000 be just a five one c but a super pack or you know any other outside group they're 905 01:19:03,000 --> 01:19:08,000 not technically the candidate even though they you know it's it's sort of uh it's 906 01:19:08,000 --> 01:19:12,000 an extension of the campaigns but do they respond in a different way when you 907 01:19:12,000 --> 01:19:19,000 call them then the campaign itself does it all depends on the group i mean 908 01:19:19,000 --> 01:19:25,001 some of them are large they they want to have a presence and so 909 01:19:25,001 --> 01:19:32,001 for whatever reason they will be responsive but others know you can sometimes try 910 01:19:32,001 --> 01:19:38,001 to reach somebody and get absolutely nowhere that generally your experience guys 911 01:19:38,001 --> 01:19:42,001 i mean some of the groups are even hard to get a hold of and i mean i'm sure 912 01:19:42,001 --> 01:19:45,001 you've found that too you go to their website and it's like a p.o box and no 913 01:19:45,001 --> 01:19:51,001 phone number but others are emailing us and we have a contact person and you know 914 01:19:51,001 --> 01:19:55,001 they're pretty aggressive in terms of being in touch with media 915 01:19:55,001 --> 01:20:00,001 thanks are there more questions 916 01:20:03,000 --> 01:20:08,000 one super super quick i realized there was a question earlier about if somebody 917 01:20:08,000 --> 01:20:13,000 sees an ad in the wild how can they find it in our site and i don't we we answer 918 01:20:13,000 --> 01:20:16,000 the question of how they could report one but the way you would find it is 919 01:20:16,000 --> 01:20:19,000 there's a search bar at the very top and you could just type in either the name 920 01:20:19,000 --> 01:20:23,001 of the candidate or the name of the sponsor if you happen to catch it at the end 921 01:20:23,001 --> 01:20:28,001 and you would you should and we should say the url for that it's archive.org 922 01:20:28,001 --> 01:20:34,000 slash tv and you'd want to say you'd want to search for something that was 923 01:20:34,000 --> 01:20:40,000 captioned you said actually i'm referring to the search on political ad arc oh 924 01:20:40,000 --> 01:20:46,001 okay but but that too yeah right i wanted to offer just 925 01:20:46,001 --> 01:20:52,000 a an interesting sort of little meta meta commentary you know marshall mcclellan 926 01:20:52,000 --> 01:20:58,000 who said the medium is the message well one of the interesting things in the 927 01:20:58,000 --> 01:21:03,001 areas that get inundated with political ads and it used to be they bought mostly 928 01:21:03,001 --> 01:21:09,000 in news slots and now they're buying without we did an experiment in the 929 01:21:09,000 --> 01:21:15,001 philadelphia television market in the um uh during the fall 2014 elections and 930 01:21:15,001 --> 01:21:21,001 there was a bunch of as a governor of pennsylvania um representatives members of 931 01:21:21,001 --> 01:21:26,001 the house of representatives from mariuson in in pennsylvania and new jersey they 932 01:21:26,001 --> 01:21:30,001 were up and there's a lot of dark money flowing around and we were collaborating 933 01:21:30,001 --> 01:21:34,001 with the university of delaware and they were asking an interesting question 934 01:21:34,001 --> 01:21:41,000 which was so local news there's is making money hand over fist and airing in a 935 01:21:41,000 --> 01:21:47,001 great deal of ads sometimes up to i think the maximum was like 13 uh 30 second 936 01:21:47,001 --> 01:21:54,001 ads over a single 30 minute program uh 30 minute local news program well what 937 01:21:54,001 --> 01:21:58,000 they discovered is something that's kind of disturbing and that is instead of the 938 01:21:58,000 --> 01:22:02,001 medium becoming the message the message started becoming the medium itself and 939 01:22:02,001 --> 01:22:09,001 they looked to see well did the local news programs and reflect uh editorially on 940 01:22:09,001 --> 01:22:14,001 any of the content of the ads that they were airing back to back with all the 941 01:22:14,001 --> 01:22:20,001 news that they were airing and sadly for our republics their their number was the 942 01:22:20,001 --> 01:22:25,000 ratio between substantive political reporting on the ads or just about the 943 01:22:25,000 --> 01:22:30,001 campaign that wasn't any campaign in the market area there wasn't a political 944 01:22:30,001 --> 01:22:35,000 horse race substantive political ads of substantive political reporting against 945 01:22:35,000 --> 01:22:41,001 the amount of time devoted to political advertising within those news programs 40 946 01:22:41,001 --> 01:22:48,000 to 1 40 minutes of political advertising for every one minute of substantive 947 01:22:48,000 --> 01:22:54,000 political reporting on the elections in the region it's because we offer this 948 01:22:54,000 --> 01:23:00,001 service um this that um this scholarly group was able to 949 01:23:00,001 --> 01:23:06,000 ask this question answer it soundly and because it's an open archive anybody can 950 01:23:06,000 --> 01:23:08,001 go back and reflect and do their own sort of accounting 951 01:23:12,000 --> 01:23:15,001 thanks are there any other questions we have a few more minutes before we break 952 01:23:17,001 --> 01:23:22,001 yeah i mean if not i might ask a few um yeah i'm just interested again to hear 953 01:23:22,001 --> 01:23:29,000 from the fact checkers uh have if you've done other elections if this is not your 954 01:23:29,000 --> 01:23:34,000 you know first time doing it how whether you're seeing any big a differences in 955 01:23:34,000 --> 01:23:40,000 uh in the ways that candidates are saying things or i'd also be interested in 956 01:23:40,000 --> 01:23:44,000 kind of a another question do you notice a lot of differences between the ads 957 01:23:44,000 --> 01:23:47,001 that are sponsored by the candidates themselves or the ones that are sponsored by 958 01:23:47,001 --> 01:23:54,000 outside groups yeah well i'll take the second part um i mean you it's kind of a 959 01:23:54,000 --> 01:24:00,001 generalization but um we do see um more negative or 960 01:24:00,001 --> 01:24:05,000 harder hitting attacks um however you want to define that from the outside groups 961 01:24:05,000 --> 01:24:11,000 than we do from the candidates themselves and i mean it's kind of like a um the 962 01:24:11,000 --> 01:24:16,000 candidate it's easier i guess for those groups to make um more controversial 963 01:24:16,000 --> 01:24:22,001 claims perhaps or um you know more a more negative claim because 964 01:24:22,001 --> 01:24:28,000 they're just you know some name of some group whereas the candidate stand is 965 01:24:28,000 --> 01:24:33,001 standing you know directly behind that ad and that claim so that would be you 966 01:24:33,001 --> 01:24:39,001 know one kind of difference that we that we see there i think back to eight years 967 01:24:39,001 --> 01:24:45,001 ago um i was in iowa right around this time a little earlier covering the caucus 968 01:24:45,001 --> 01:24:51,001 for the dallas morning news where i worked and the notion of a super PAC or or a 969 01:24:51,001 --> 01:24:57,000 non-profit coming into the iowa caucuses and spending millions upon millions of 970 01:24:57,000 --> 01:25:03,000 dollars on tv ads or digital ads or even canvassing or some of the core functions 971 01:25:03,000 --> 01:25:08,000 of a campaign would have been nonsensical in part because super PACs did not 972 01:25:08,000 --> 01:25:14,000 exist and non-profits did not have the types of powers that that they do quite in 973 01:25:14,000 --> 01:25:19,001 the way that that they have now so on a very very basic level it just shows that 974 01:25:19,001 --> 01:25:25,001 in less than a decade the campaign environment the battleground if you will has 975 01:25:25,001 --> 01:25:31,001 shifted uh so markedly not to be unrecognizable but but in a way that that's 976 01:25:31,001 --> 01:25:36,001 quite profound and in a way where where you can't make really apples to apples 977 01:25:36,001 --> 01:25:42,001 comparisons in a meaningful way to the election that barack obama was running 978 01:25:42,001 --> 01:25:46,000 against hillary clinton at that time and the one that hillary clinton is now 979 01:25:46,000 --> 01:25:49,001 running against bernie sanders or for that matter any of the republicans eight 980 01:25:49,001 --> 01:25:55,000 years ago to the ones today and you know i think oftentimes where we're so uh 981 01:25:55,000 --> 01:25:59,001 enmeshed in the moment and we're focused on everything that's happening now to 982 01:25:59,001 --> 01:26:03,001 take a step back just for a moment and and look at what was just happening in 983 01:26:03,001 --> 01:26:10,000 years ago and how different it is really shows uh how 2016 is a 984 01:26:10,000 --> 01:26:16,000 unique political animal unto itself and and that we've we've moved in a for 985 01:26:16,000 --> 01:26:20,000 better or for worse in a very different direction than we were going in uh eight 986 01:26:20,000 --> 01:26:23,001 years ago yeah and on that note it might be a good point when i started on this 987 01:26:23,001 --> 01:26:28,000 project one of the things i did was i went to archive.org and i found old 988 01:26:28,000 --> 01:26:32,001 political ads you know from the 60s from the 70s and what struck me was that a 989 01:26:32,001 --> 01:26:36,000 lot of those ads back then were really honestly they were the candidate talking 990 01:26:36,000 --> 01:26:40,000 to the camera about what i'm going to do about energy policy it was amazing how 991 01:26:40,000 --> 01:26:44,001 just kind of straightforward and non-slick they were and then of course along 992 01:26:44,001 --> 01:26:48,001 came the daisy ad the famous daisy ad that is on the archive you can find it and 993 01:26:48,001 --> 01:26:53,000 that was the ad that the johnson campaign did that famously showed a little girl 994 01:26:53,000 --> 01:26:58,001 with a daisy and then there's a you know kind of a nuclear explosion. Well on the 995 01:26:58,001 --> 01:27:03,001 republican side in particular the vast majority of ads uh it our our data shows 996 01:27:03,001 --> 01:27:09,001 that the vast majority of ads at least on the republican presidential side uh are 997 01:27:09,001 --> 01:27:14,000 being sponsored by so-called outside organizations the super PACs and non-profits 998 01:27:14,000 --> 01:27:19,000 so you have a situation there in particular where it's not the candidates talking 999 01:27:19,000 --> 01:27:23,000 to the voters it's not the candidates committee that's the sponsor of the ad you 1000 01:27:23,000 --> 01:27:26,001 have a proxy war that's taken place where you have all of these other different 1001 01:27:26,001 --> 01:27:31,001 types of groups who are supporting Marco Rubio or Ted Cruz or Donald Trump or 1002 01:27:31,001 --> 01:27:35,000 John Kasich or Chris Christie or any of the others and they're the ones who are 1003 01:27:35,000 --> 01:27:41,000 providing the message so that the candidate can can in a way have plausible 1004 01:27:41,000 --> 01:27:47,000 deniability or implausible deniability some might suggest that that oh you know 1005 01:27:47,000 --> 01:27:50,000 we're not the ones who are behind the attack ads we're not the ones who are being 1006 01:27:50,000 --> 01:27:54,000 negative it's that independent outside group that that's a step or two or ten 1007 01:27:54,000 --> 01:27:59,000 away from my campaign and and we see less of that on the democratic side although 1008 01:27:59,000 --> 01:28:03,000 there are definitely democratic super PACs Hillary Clinton has quite an outside 1009 01:28:03,000 --> 01:28:07,001 machine unto herself and even Bernie Sanders who says I hate super PACs and 1010 01:28:07,001 --> 01:28:11,000 outside money is horrible and I would reform the whole system and that's the 1011 01:28:11,000 --> 01:28:17,000 basis of his campaign still even does involuntarily have some outside groups to a 1012 01:28:17,000 --> 01:28:21,001 degree supporting his campaign so everyone is affected and everyone is involved 1013 01:28:21,001 --> 01:28:27,001 at least most everyone so we're about finished with time do we want to say one 1014 01:28:27,001 --> 01:28:31,000 thing quickly I just wanted to say one quick thing that you don't even have to go 1015 01:28:31,000 --> 01:28:36,000 back to 2008 to see how far things have come I mean in 2012 it was sort of 1016 01:28:36,000 --> 01:28:40,001 scandalous that candidates were setting up super PACs that were you know 1017 01:28:40,001 --> 01:28:46,001 extensions of their campaigns now they're doing that openly and most candidates 1018 01:28:46,001 --> 01:28:51,001 over at least you know half a dozen candidates at this point have a 501c4 that is 1019 01:28:51,001 --> 01:28:56,001 run by the same people that run their super PACs it was established to raise 1020 01:28:56,001 --> 01:29:01,001 money from people who didn't want to be named these are you know not only do you 1021 01:29:01,001 --> 01:29:05,001 have the unlimited extension of the of the campaign but you have the unlimited 1022 01:29:05,001 --> 01:29:11,000 undisclosed extension of the campaign which did not exist in 2012 this is the 1023 01:29:11,000 --> 01:29:15,000 first presidential cycle we've ever seen that so the changes are coming 1024 01:29:15,000 --> 01:29:20,000 incredibly quickly so Roger would you like to say a few concluding words 1025 01:29:20,000 --> 01:29:26,001 yes from a librarians and archivists perspective so 1026 01:29:26,001 --> 01:29:33,000 we're trying to create this resource that's available to help citizens help make 1027 01:29:33,000 --> 01:29:38,001 sound choices when confronted by an enormous amount of money and an inundation of 1028 01:29:38,001 --> 01:29:45,001 of messaging and making these choices this year that affect themselves their 1029 01:29:45,001 --> 01:29:52,000 families their communities the world around us and will affect posterity 1030 01:29:52,000 --> 01:29:58,000 for decades to come from a archivist and librarian perspective more looking down 1031 01:29:58,000 --> 01:30:02,001 that timeline one of the great things about capturing as much as we can now on 1032 01:30:02,001 --> 01:30:07,001 the web on tv and so many other different media types is that the the greatest 1033 01:30:07,001 --> 01:30:14,000 insight will come via the hindsight five ten twenty fifty years from now so 1034 01:30:14,000 --> 01:30:21,000 we are trying to invest in the future of of the capability of insight for 1035 01:30:21,000 --> 01:30:27,000 generations to come in an era when so much is digital and so much is being lost 1036 01:30:27,000 --> 01:30:34,000 at every moment we're trying to strive to to meet the needs of people 1037 01:30:34,000 --> 01:30:41,000 now and help people reflect back on what we did and it's not all bad what did we 1038 01:30:41,000 --> 01:30:46,000 do good where was their great success and where did we go so wrong 1039 01:30:46,000 --> 01:30:50,001 thank you thanks for coming out