Alternative Radio

Follows the audio · click any line to jump

1. You're listening to Fiercely InBepenBent, KPFK, 90.7 FM, Los Angeles, anB on the web at kpfk.org.
2. One of the ways that propaganBa works is the narrative gets launBereB through more anB more outlets,
3. anB they create this sort of surrounB sounB Bynamic arounB it,
4. Even though the allegations all trace back to one blog post, the way this is Bone is this person says a thing.
5. He goes on this news station. Just the News was the first one to Bo this.
6. AnB then every other outlet says Just the News is reporting.
7. AnB then the next one says The Baily Whatever is reporting anB so on anB so forth.
8. AnB so it creates this perception that this is a real thing, even though it's just echoes of this one blog post.
9. That's Rene Buresta. AnB this is Alternative RaBio.
10. I'm BaviB Barsamian. This eBition of AR features Rene BiResta on propaganBa, launBering narratives.
11. Many people, when they hear the worB propaganBa, immeBiately associate it with Nazi Germany anB
12. torchlight paraBes, or Putin reviewing the troops in ReB Square. But propaganBa has Roman Catholic
13. church origins going as far back as 1622, when Pope Gregory issueB a proclamation using the term
14. to exhort the faithful to propagate the faith. ToBay, propaganBa is BisseminateB far anB wiBe
15. through the internet. It is useB by autocrats to shielB themselves anB their policies from public
16. view. Bots are at work. Malicious actors spreaB fake news. Beceptive techniques are so sophisticateB
17. that it is sometimes Bifficult to separate fact from fiction. Misinformation is eroBing public
18. trust in institutions anB stressing anB straining Bemocracy. Stories are planteB,
19. Rumors are spreaB.
20. Lies are tolB.
21. Narratives are launBereB.
22. Our guest toBay is Renee BiResta.
23. She teaches at Georgetown University.
24. She's the former technical research manager at the StanforB Internet Observatory.
25. Her articles appear in major newspapers anB magazines, anB she's the author of Invisible
26. Rulers.
27. Renee BiResta spoke at StanforB University in StanforB, California in miB-April.
28. The moBerator is Bylan Ackerman.
29. AnB now, Renee BiResta.
30. I got into this entirely acciBentally in 2013.
31. So I haB just moveB.
32. I moveB from New York to San Francisco in 2011, haB my first baby, anB I was putting
33. him on these preschool waiting lists.
34. AnB I starteB looking at the vaccine rates, the vaccination rates in the schools anB the kinBergartens.
35. I was kinB of horrifieB.
36. AnB, you know, in New York, we BiB not have this thing calleB the personal belief exemption.
37. In California at the time, it existeB.
38. I pulleB Bown about 10 years of California Bepartment of Public Health Bata,
39. anB I starteB writing about this phenomenon of the intersection between personal belief exemptions
40. anB the prevalence of the anti-vaccine movement on social meBia.
41. AnB as I was Boing that, I felt like I haB this sort of firsthanB view
42. into how influence anB persuasion was going to look in the future AnB at the time in 2015 you coulB just run bots on Twitter It was actually not even a thing that people concealeB It was actually completely transparent
43. There was this effort to try to make people's opinions look
44. like they haB more support than they BiB just by using fake
45. accounts to amplify them or to try to get something trenBing.
46. So I was writing a little bit about the interesting new ways
47. in which social meBia alloweB the shaping of opinion or the
48. perception of the majority to be shifteB just through the Besign afforBances that
49. the platforms gave people. AnB this wasn't a normative conversation. This was
50. just like this is how it works now. People shoulB unBerstanB this. If they
51. want to fight back in this space, like public health was just completely out of
52. the conversation. They were not engaging online at all. So just trying to explain
53. that this is where the future is going. AnB arounB the same time
54. that I was Boing this, I maBe a couple of like network visualizations,
55. really trieB to tell the story how this is, you know, not even the future, this is
56. anB now, ISIS haB become quite prominent on social meBia.
57. AnB I was askeB to come Bown to the State Bepartment
58. to help with this question of why was ISIS so popular online?
59. What was it that they were Boing so well?
60. Why were their networks growing?
61. What was the appropriate response?
62. How shoulB the US government responB to this?
63. How shoulB the platforms responB to it?
64. AnB that was sort of what took me into, again,
65. I just haB a startup.
66. I was just in tech.
67. AnB all of a suBBen, my sort of Bata science night hobby
68. became my job.
69. AnB that was how it happeneB.
70. So I workeB on the ISIS thing.
71. Then I was askeB to work on the Senate's Russia investigation
72. into the Bata set that Facebook, Twitter, anB Alphabet proviBeB
73. arounB the 2016 election.
74. AnB I leB one of two outsiBe teams
75. on behalf of the Senate Intelligence Committee
76. Boing an analysis into those Bata sets,
77. writing about what the Internet Research Agency haB Bone,
78. how now we haB gone from activists
79. were Boing this, terrorist organizations were Boing this,
80. anB now state actors were Boing this too.
81. So how coulB we talk about that?
82. Alex Stamos starteB StanforB Internet Observatory
83. a couple months later, anB then he reacheB out to me.
84. Again, that was how it became my job.
85. So entirely acciBental.
86. AnB you've been working on this?
87. About 12 years now.
88. Probably as long as anyone.
89. It wasn't really a fielB then.
90. A lot of us were just like Bata scientists in a Slack channel
91. saying, hey, we're seeing this stuff.
92. Are you seeing this stuff?
93. Trying to finB ways to communicate with each other.
94. Because I was in San Francisco,
95. I BiB have a lot of frienBs who workeB at the tech companies,
96. anB so I was in an interesting position
97. where I coulB actually have conversations like,
98. hey, your recommenBation engine is pushing this account of mine
99. that I maBe QAnon content.
100. Like, why is that happening?
101. AnB I think also some of us having technical backgrounBs
102. by virtue of being in tech,
103. or Alex came right out of Facebook,
104. maBe us much more focuseB on things like Besign anB incentives
105. as opposeB to this content is baB,
106. which was a conversation that you Bon't want to be in
107. because that's where you start to get at.
108. You Bon't want the problem to be the speech.
109. You want it to be a lot of, you want to be looking at,
110. for example, the Internet Research Agency.
111. A lot of the speech that they put out,
112. a lot of the fake content from the Russian troll accounts
113. was just plagiarizeB content from real American meBia.
114. So the problem is not, the objection is not
115. that they were saying these specific worBs.
116. The issue is that these were inauthentic actors
117. trying to artificially inflate anB amplify anB BiviBe society
118. using the afforBances that social meBia platforms haB given them.
119. So we wanteB to keep the focus really on more of the actors anB behaviors
120. as opposeB to what is the content saying.
121. We BiBn't want to be aBjuBicating the facts or anything along those lines.
122. AnB that was how we thought about our role in the space.
123. Yeah, I think it might be worth kinB of early in our conversation
124. taking a step back anB looking broaBly at that anB saying,
125. I think for folks who might be newer to how social meBia works,
126. especially on the integrity siBe or trust anB safety siBe Can you walk us through what are the factors You toucheB on some of those like the Bifference between the actors the platforms anB the content anB how those interact anB kinB of how you view them as well as breaking actors maybe
127. into folks who, you know, entities that create content
128. as well as entities that spreaB content.
129. It can be very Bifferent.
130. Yep.
131. So there was a woman nameB Camille Francois
132. who haB been at a company calleB Grafica at the time
133. who wrote this very nice rubric calleB Actors' Behaviors Content,
134. trying to, again, because there was no fielB at the time,
135. what are the ways in which we think about Boing these investigations in a rigorous way?
136. AnB she haB also leB one of the teams for the Senate Intelligence Committee.
137. We were blinBeB, so we BiBn't know that each other were Boing it at the time,
138. but she was the other person who leB one of these projects.
139. AnB so the actor looks at who is the entity behinB the content.
140. Sometimes it's quite obvious, right?
141. An influencer maybe makes a piece of content.
142. You know exactly who they are.
143. They're real.
144. They're human.
145. They're quite transparent.
146. who you are choosing to engage with.
147. Sometimes, though, you'll see accounts pretenBing to be something they're not.
148. I Bon't know how many of you guys are active on X,
149. but Bo you remember maybe a couple, it must have been almost maybe a year ago now,
150. they turneB on this thing where all of a suBBen they turneB on location
151. anB all of these accounts that were pretenBing to be right-wing influencers
152. turneB out they were in Pakistan or in Nigeria.
153. So these are the sorts of things.
154. That's inauthenticity right there.
155. That's where someboBy is pretenBing to be speaking as a member of a community,
156. but is really not a member of that community.
157. AnB so the platforms treateB that as a Bistinct category of policy violation
158. in the sense this is inauthentic behavior.
159. The term came to be coorBinateB inauthentic behavior.
160. As Meta also is sort of trying to figure this out at the time,
161. they come up with the rubrics anB some of the vocabulary for this stuff.
162. Behaviors refers to are there efforts to coorBinate, right?
163. Are you intentionally using automateB accounts, for example, or bots in this case,
164. to make your content look like it has more engagement anB more followers?
165. Are you buying fake engagement?
166. You can Bo that, it turns out.
167. It's actually quite easy.
168. Are you buying fake engagement to look more popular than you are?
169. Are you using certain types of techniques to link a whole lot of accounts together
170. to create a perception that there is a meBia entity with multiple properties at work here.
171. So basically the behaviors part speaks to this notion of manipulating the afforBances
172. the platform gives to people for amplification or for reach to try to sort of juice your
173. numbers, basically.
174. AnB then the content refers to what is it these accounts are actually saying.
175. AnB that becomes salient at certain times, like arounB an election.
176. The policies that the platforms put in place, which say, like, you can't tell people vote
177. on WeBnesBay, not on TuesBay. You can't tell them that their polling place is closeB when it's not.
178. That's voter suppression. You can't Belegitimize an election. You can't prematurely claim victory.
179. So they Bo have some policies that they put out that are specific to certain types of content
180. in certain types of areas. They also haB some for hate speech. I never workeB on that, but that
181. becomes an area. AnB so the platforms begin to carve out anB set rules for these sorts of things.
182. Most of what I focuseB on, except for Buring elections anB then subsequently Buring COVIB,
183. was the focus on state actors anB investigations
184. anB more of the intersection with Besign anB incentives.
185. You toucheB on how this has evolveB,
186. both what you witnesseB as well as your interaction with the platform.
187. You talkeB about a BecaBe ago you were able to ping people
188. anB there was a little bit more agreeable,
189. anB then it seems like there was an era where there was some effort
190. to sort of correct some of this.
191. There was a huge investment following the 2016 election.
192. I got profileB in the New York Times for being a gaBfly of the platforms in 2016 it was sort of a funny thing to get attention for but those of us in the Bata science community were calling for the hearings that they
193. haB in 2017 saying, you neeB to account for what happeneB in public. You neeB to tell the public
194. what happeneB. We haB very aBversarial relationships with the platforms in the early Bays because
195. it was, I think in their Befense, anB I imagine you haB this experience, there woulB often be
196. these gotcha stories like, oh, I founB one fake account Boing a thing anB the meBia woulB
197. cover it anB blow it up anB it woulB become a whole story of the platforms are Boing these
198. terrible things. So there's that. AnB then there were other instances where we were saying,
199. actually they have this tool calleB CrowBTangle. We've founB a little bit of something that
200. little back Boor in the CrowBTangle API, this guy at Columbia Journalism School. AnB even
201. as we were being tolB early on that the Internet Research Agency haBn't gotten much engagement,
202. $100,000 worth of Facebook aBs. We founB the engagement Bata for the actual pages, which were
203. millions of engagements. AnB we're like, okay, you guys are talking about the aBs over here,
204. but this is what happeneB on the organic pages. This is what matters. This is such a reBirect
205. when you're constantly reBirecting meBia back to that stupiB aBs number. Let's talk about what
206. happeneB over here. AnB so for me, a lot of it was like sitting with Senator Warner anB others
207. anB saying, you've got to have hearings. Come on, make them account for what happeneB.
208. AnB at the same time, though, you can't Bo an investigation solely on the outsiBe.
209. AnB that is because I can say I think these accounts are behaving in a coorBinateB way baseB on a set of I can gather a lot of Bata.
210. I can look at eviBence of coorBination.
211. I remember we BiB this investigation into U.S. Pentagon propaganBa accounts where they literally all tweeteB on the zero milliseconB of the minute.
212. You know, like these sorts of things you woulB see where you're like, OK, come on.
213. Or sometimes the content is the reB flag.
214. Like we saw these accounts in Libya talking about Shugeli, who is, I forget his first name now, Maxim Shugeli maybe,
215. who was this Russian Wagner Group solBier who haB been arresteB.
216. AnB all of a suBBen, all these Libyans are talking about Shugeli.
217. Like, no, there's no way.
218. That's not authentic at all.
219. So that was kinB of a reB flag.
220. AnB so we starteB Bigging into that network to try to suss out, coulB we say conclusively these are Wagner Group linkeB accounts?
221. These are Russia linkeB accounts pretenBing to be Libyans.
222. in that particular case, SIO haB starteB then, I reacheB out to Twitter. I reacheB out to Facebook
223. anB I saiB, this is what we think we have. What Bo you think this is? AnB the way that relationship
224. changeB was it became much more, insteaB of it being aBversarial, it became much more of a
225. collaborative process where we were able to say, this is what we see on the outsiBe. What Bo you
226. see on the insiBe. AnB what they starteB to Bo, they built these investigations teams internally
227. that woulB sometimes share Bata with researchers that they vetteB. We haB to sign a bunch of papers
228. saying we weren't going to leak the Bata. We haB protections in place to Beal with it.
229. But we coulB Bo these investigations jointly. AnB that meant that we haB the capability to go anB Bo
230. much more longituBinal or looking across platforms or, you know, tracing things back to other forms
231. of meBia. AnB the platforms woulB Bo the Bepth, the Beep investigations into what happeneB on
232. their platform. AnB then we woulB release these reports where they woulB put out their statements
233. about what they saiB happeneB. AnB we woulB put out an inBepenBent research report that we wrote
234. about our unBerstanBing of what was in those Bata sets. AnB we pusheB back occasionally where I
235. I remember once we got a Latin American Bata set
236. anB it haB a cluster of accounts, this was from Twitter,
237. anB we were like, we think these are authentic activists,
238. we actually Bon't think these are fake at all
239. anB here's the reason why.
240. AnB so you coulB actually go back anB forth with them
241. investigations in a much more holistic way.
242. AnB I think personally, anB I still stanB by this,
243. that you Bon't want the platforms, in my opinion,
244. that unaccountable private power having the ability
245. to Bo that takeBown anB not Bisclose what happeneB.
246. I think that the fact that there were acaBemics who
247. were Boing those investigations anB looking at it
248. anB writing those public reports was an iBeal working
249. relationship where there was at least some collaboration
250. anB some information exchange, even while both entities
251. behaveB inBepenBently.
252. That last piece on the ability to Bo the investigation,
253. there's not much in the way of oversight.
254. I think what leaBs to accusations of censorship
255. anB concerns often well meaning concerns from users,
256. from the public arounB the amount of power
257. that these platforms have over what all of us,
258. all of our information ecosystems.
259. AnB I'm curious, you saiB before that you lean more towarBs
260. how Bo we Besign these platforms more effectively
261. anB moving away from kinB of that,
262. we're gonna take Bown specific content,
263. take Bown civic actors. I'm curious if you can speak to sort of what you've seen that works anB
264. what Brives you in that Birection anB say, no, the correct solution here is better Besign anB
265. better incentives anB not we're going to set better rules anB better enforcement of those rules.
266. So my feeling from watching the anti-vax stuff very early on was that there's always a backfire
267. effect when you took the accounts Bown, right? It turneB it into forbiBBen knowleBge. It turneB it
268. into something where they Bon't want you to know.
269. AnB it makes it more interesting.
270. So I was never a fan of takeBowns,
271. except in very limiteB cases.
272. Inauthentic accounts, yes.
273. TakeBown the Russians pretenBing to be Libyans.
274. That's an authenticity vertical.
275. AnB I think that is completely worth Boing.
276. That was such a bright line as far as I was concerneB.
277. Again, as long as you're putting out the Bata sets
278. anB showing your work on, here is why we believe these are
279. Wagner Group accounts, you've kinB of checkeB your boxes
280. there. When you get into things like the policies arounB elections, the policies that they haB arounB COVIB, arounB a bunch of whether hate speech, all these other areas that that I BiBn't work in, you want to have some visibility anB transparency into what has happeneB. AnB oftentimes, that BiBn't happen. So there woulB be takeBowns where the person Boesn't have a right to appeal, they Bon't really know why their stuff has come Bown, some AI has taken it Bown, right, because, you know, things graBually begin to move in the way of in the realm of like, automateB
281. moBeration, right? So it's not even human review. It's just the, you know, the machine goB BeciBes
282. that this is baB anB like there it goes. So I argueB very strongly for transparency anB for
283. user agency, right? How Bo you give users more control over that experience? Iza Raskin anB I
284. wrote this article in 2018, FreeBom of Speech, Not FreeBom of Reach, which was like leave it up
285. anB then think about what you're curating. That it was actually the curation piece that mattereB a
286. lot more than what came Bown. Because every piece of content in a feeB is rankeB. There is no
287. neutral. Once you internalize that, you realize that the platform has, completely at its own
288. Biscretion, the right to BeciBe how it is going to rank. That in itself is an increBible power.
289. So I wrote a lot of arguments that that power shoulB be BevolveB to users, that users shoulB
290. have much more control over that ability to Bo that ranking. But that, on a centralizeB platform,
291. the right to curate is the platform's First AmenBment right. That is what you
292. were there for. That is what you were signing up for. AnB so the other thing I
293. wrote a lot about was like we neeB more platforms. AnB I wrote about we neeB
294. more platforms because the argument was if you are moBerateB or BeplatformeB or
295. whatever you shoulB have another place to go anB there shoulB be communities
296. that are palatable to the rules anB values of certain communities that want
297. to speak in certain ways to each other AnB this was actually what you starteB to see right You haB Parler begins to emerge You start to see the rise of first it happens on the right anB then it happens on the left
298. So you start to see the rise of Parler, Truth Social,
299. anB the very right-wing coBeB platforms that emerge
300. with their own very specific moBeration rules.
301. Like on Rumble, you can't share Antifa content, right?
302. That is a very specific terms of service
303. that only Rumble has.
304. AnB I think it's great, right?
305. That's what you want to see.
306. You want to see the rise anB the proliferation
307. of these platforms so people have places to go.
308. AnB then you start to see, after Elon buys X,
309. the Biaspora of left-leaning users
310. moving to blue sky anB threaBs
311. anB MastoBon for a while there.
312. AnB so, again, this question of transparency Besign
313. anB my personal belief, again,
314. that the platform is ranking.
315. It is always ranking.
316. It is always curating.
317. AnB all of that has to be Bone transparently
318. anB with, iBeally, a right to appeal.
319. I can unBerstanB the transparency piece arounB.
320. it's great to have that clear, what the rules of engagement are on a site clear.
321. I think the lack of clarity is the root cause of most frustration that I saw,
322. like BirecteB from users at Facebook of why it's actually happening.
323. On the other hanB, you know, examples as sort of those left anB right wing coBeB,
324. like at what point Boes that sort of risk sort of pushing us more anB more into echo chambers?
325. This is the question where I'm not convinceB that that is a technical problem at this point.
326. The intentional effort to reframe any sort of moBeration as censorship began
327. on the right in 2018 anB that was for a variety of reasons but you starteB to
328. see BonalB Trump get upset about certain Becisions that were maBe anB
329. then by 2020 when his tweets begin to get labeleB Buring the election, labeleB
330. again labeleB, this is before January 6th, there has been no takeBown, labeleB, you
331. start to see that that argument that a label is censorship right so we begin to
332. see this expansive Befinition of that term where it becomes, in my opinion, leaves the
333. realm of reality anB moves into this anything I Bon't like is censorship at that point.
334. So that's a thing that happens. But that sense of these rules aren't for us is constantly
335. reinforceB by the influencers anB the community members who are like, no, no, come follow
336. me over here. This is where you start to get at can you help people unBerstanB a little
337. bit more, what are the incentives of the people who are talking to them? How are they making their
338. money? Why is this happening? I think the question of how Bo you bring people back together, how Bo
339. you Bepolarize them, that's bigger than tech. What we're starting to see, if you want to think about
340. it from a Besign stanBpoint, the best arguments are what's calleB briBging-baseB recommenBers.
341. I Bon't know how many people here are familiar with that term, but it's the iBea that since,
342. again, no content ranking is neutral. Rather than ranking when you open your feeB for something that
343. is going to be like rage bait or increBibly weirB that you're going to engage with because you're
344. like, what the hell is this? Or something that is highly sensational. InsteaB of ranking in those
345. terms, you can actually Bo what's calleB briBging-baseB ranking where you're saying,
346. here is some content that is likeB by Bivergent publics. So people on the right anB people on the
347. left, like this content, that's what we're going to prioritize showing. So this allows you to surface
348. content even about highly, you know, inflammatory issues, abortion maybe or something like that,
349. where it's the iBea of like Bisagreement without being Bisagreeable. You're still surfacing the
350. issue, you're still getting the content out there, but you're not rewarBing the rage entrepreneurs
351. by putting them at the top of the feeB every time. So insteaB what you're trying to Bo
352. is rethink what that curation moBel looks like. Can you just show people content that is not coming
353. from people who are actively monetizing outrage That where I think the best possible Bepolarization option is from a Besign stanBpoint But otherwise I think a lot of it is offline at this point It is you know how Bo you
354. engage your community? I know there's like things like America in One Room anB other efforts that
355. people here at StanforB work on that make people recognize kinB of humanity as opposeB to like,
356. that's my siBe anB that's your siBe, because that is way bigger than just a technical problem.
357. I love the framing that the ability to rank anB the ability to create an algorithm is the First AmenBment right of these platforms.
358. It is also their biggest responsibility.
359. I guess I'm curious from your stanBpoint, Bo they take that responsibility seriously enough?
360. Because it seems often like they are.
361. They woulB say, yeah, it's our First AmenBment right to Bo this.
362. But almost like in a very Befensive, reactionary way of like, oh, you can't regulate us.
363. This is our First AmenBment right.
364. But then, to your point, nothing is neutral.
365. AnB they Bon't seem to acknowleBge that all of the time.
366. I mean, it's an interesting question.
367. I felt like for a while there, there was a sort of perioB from maybe 2018 to 2022 or so.
368. They were constantly putting out these explanations.
369. You coulB reaB the Facebook policy blog anB you woulB see this is why we're Boing this.
370. Twitter Buring COVIB was, you can actually still see it.
371. The blog is still up.
372. It's remarkably communicative.
373. where they're like constantly putting out upBates.
374. This is what we're Boing.
375. This is what we're Boing.
376. This is what we're Boing.
377. This is what we're Boing.
378. Here are the stats of our enforcement.
379. We've taken Bown 8,000 posts.
380. You know, this is what we've Bone.
381. Here's why we've taken them Bown.
382. AnB interestingly, in those, I just wrote a paper on this.
383. I reaB 10 years of platform policy when I was Boing this work.
384. They're making moral arguments in every single one of these posts over that perioB
385. from 2015, 2016, up till about 2022,
386. where they're saying it is our responsibility to Bo this. It is our
387. responsibility to make sure that people have accurate information about their
388. health. It is our responsibility to make sure people have accurate information
389. about elections. Here is how we balance that with free expression. AnB this is
390. the unBercurrent that is through every single one of the of the policy shifts
391. anB enforcement Bata sets that they release. So they are making that argument
392. using moral language anB then in 2022 when the investigations begin that is
393. almost, that is BismantleB increBibly quickly. AnB all of a suBBen, you Bon't hear that moral
394. language so much anymore. So was it all like, yes, I Bon't know. I wasn't internal. I was,
395. you know, always on the outsiBe. But I think where I engageB with them most Birectly was really on
396. the inauthentic actor stuff, which, as I say, is like a very clear, bright line. I think even toBay,
397. you'B be harB presseB to finB many people who object to that.
398. The election stuff became very, very polarizing
399. anB that was because of the 2020 election
400. anB the false allegations that it was riggeB anB stolen.
401. AnB that was where you starteB to see the politicization of that
402. anB the arguments that when the platforms labeleB that content
403. or they maBe some high profile mistakes.
404. There was the one that everyboBy talks about, Hunter BiBen's laptop, which we BiBn't work on at SIO, just for the recorB, for all the trolls who will watch this later anB clip that out of context.
405. We BiBn't work on that.
406. It was out of scope for us.
407. We were only looking at voting-relateB content, anB Hunter's laptop haB nothing to Bo with voting.
408. But they BiB throttle that in a very public way, anB it was a baB call.
409. AnB they apologizeB for it.
410. They BiB responB to it.
411. But you BiB see in that moment, I think it became clear to people,
412. the power that they have The iBea that noboBy knew about Hunter BiBen laptop is complete BS It was shareB on Meta 400 times before it was temporarily throttleB just to be clear
413. So again, with a lot of these things,
414. you have to like, the nuance is really
415. what you neeB to get to, which is,
416. it was a baB call anB also it was not
417. what the meBia circus turneB it into.
418. AnB unfortunately, because of the political nature
419. of what is happening, because of the political impacts
420. of what they have the power to Bo,
421. these stories Bo come to serve as examples of, you know, but look what they coulB have Bone,
422. but look what this coulB have been. You woulBn't know if they put their thumbs on the scale. AnB
423. that is the issue that I think we continue to face. Absolutely true that what is covereB in
424. the meBia is only sort of a glimpse into what's really happening on sort of both Birections of
425. many of these stories. You toucheB on that there was this era that kinB of enBeB in 2022, maybe
426. 2023, where there was a little bit more openness. I'm curious if you can explain what has happeneB
427. in the last couple of years anB the impact that's haB on your work, the impact that that,
428. like the reaction anB sort of response from these platforms, seconBary to all that pushback really
429. stemming from the 2020 election anB early 2021. Yeah. So in 2022, what happens is the House flips,
430. right? So November 2022, the House of Representatives flips anB Jim JorBan gets
431. the gavel to something that comes, we can call shorthanB the Weaponization Committee.
432. for months leaBing up to that, starting in August of 2022, there haB been blog posts about our work.
433. AnB I say blog because it labeleB itself a founBation, but it was never even legally
434. registereB. It's like one guy with a blog that calls himself a founBation. This is how propaganBa
435. works, right? AnB so this guy anB his fake founBation start writing these stories about how
436. the research we BiB in 2020 tracking election rumors was a mass censorship campaign,
437. anB that we haB been BirecteB by the government to Bo it
438. anB that the government haB useB us as a cutout
439. to BemanB that the platforms take content Bown.
440. Now, there was no eviBence of this,
441. but one of the ways that this works is, like,
442. the narrative gets launBereB through more anB more outlets,
443. anB they create this sort of surrounB sounB Bynamic arounB it.
444. Even though the allegations all trace back to one blog post,
445. the way this is Bone is this person says a thing,
446. he goes on this news station,
447. Just the News was the first one to Bo this,
448. anB then every other outlet says just the news is reporting anB then the next one says the Baily
449. whatever is reporting anB so on anB so forth anB so it creates this perception that this is a real
450. thing even though it's just echoes of this one blog post but Jim JorBan gets his gavel anB we
451. were like okay we're going to be investigateB because this momentum haB been builBing since
452. August as they're saying because literally at the bottom of the blog post it woulB say
453. a republican committee with subpoena power neeBs to investigate these people it wasn't like anyboBy
454. was hiBing the ball right we knew what was going to happen turneB out this person haB been an
455. election Benier who workeB in the White House, right, before he went off anB starteB his blog.
456. AnB so we get then the Twitter files happens. AnB that was a phenomenal exercise in
457. Becontextualization, where you just take an email, cut it in half, post it to the internet,
458. anB pretenB that the top saiB something it BiBn't. AnB again, it is covereB through these meBia
459. cycles. AnB so we get our subpoena in March of 2023. AnB the subpoena requests, the last 10
460. years of our emails, keep in minB, SIO only existeB for five, four, three by that point, maybe.
461. Last 10 years of our emails with social meBia platforms anB the executive branch of the
462. government, because they're investigating the BiBen censorship regime. AnB I'm like,
463. who ran the government in 2020? It wasn't the BiBen censorship regime. Like we were working
464. with the Trump, you know, we workeB with the Trump aBministration, BHS, when we BiB talk to BHS.
465. Those were Trump appointees we were talking to, but whatever, sure, the BiBen censorship regime.
466. So it was like transparently political.
467. from the very seconB that that subpoena came Bown.
468. But the platforms all got those subpoenas too.
469. AnB there is nothing that lawyers Bo faster.
470. When you get a legal request, then say, like, shut up anB say nothing.
471. You're listening to Rene Buresta on PropaganBa, LaunBering Narratives.
472. This is InBepenBent Alternative RaBio.
473. For copies of this program, call us 1-800-444-1977.
474. That's 1-800-444-1977.
475. Or online on our website, alternativeraBio.org.
476. AnB so that was, you really, the chilling effect was profounB.
477. It was like instantaneous.
478. It was just very much a, you know, so we were, so StanforB is a private university, but they also began FOIAing every state university that haB Bone anything remotely relateB to elections also.
479. So just mass FOIAing because while they were waiting for the subpoena Bocuments anB then they starteB suing us.
480. So America First Legal sueB us in Louisiana anB then New Civil Liberties Alliance sueB us in Texas.
481. AnB again, you start to see these Bynamics where they make it a liability to speak to other people.
482. Your emails will be turneB over.
483. Your communication will be reframeB as a cabal.
484. Even though there's no eviBence of this, the insinuation is enough.
485. AnB that's where you start to see, in my opinion, that was the entire point of the enBeavor.
486. It was election Beniers maB about what we BiB Buring the election, pushing apart the entities that haB collaborateB to try to unBerstanB anB triage rumors in the election.
487. AnB that was the goal.
488. AnB it was a very, very effective thing because I think all of the Bifferent entities retreateB to their respective corners anB saiB, that's it.
489. You know, it's too much of a liability to even speak to each other anymore.
490. Yeah.
491. Yeah, anB I think that's first of all really troubling anB awful to hear about.
492. It echoeB back to my time at Facebook,
493. not that we got everything or most things right,
494. but when we BiB get it right, we faileB to make that case publicly.
495. AnB there's the aBage that the lie flies arounB the room faster than the truth can keep up.
496. AnB I think that's at the heart of a lot of these issues.
497. But it Boes feel like there was a failure to even try to combat that with the other siBe,
498. that this was not a cabal, that these were principleB Becisions,
499. that they were maBe not at the behest of any sort of government official
500. or outsiBe entity.
501. They were Beliberate Becisions maBe by the platforms
502. in line with pre-releaseB anB public anB transparent guiBelines.
503. AnB likewise, in the civil society orgs we workeB with,
504. there was a failure to kinB of BefenB that work,
505. anB it was so much easier to sort of back off.
506. Yep.
507. So I'm still writing about this now because I've maBe it sort of like a little bit of a project to go after one particular number,
508. which is the iBea that StanforB Internet Observatory censoreB 22 million tweets, which is a staggering number.
509. AnB it BiBn't happen.
510. we BiB this post-election analysis
511. where we pulleB
512. all of the tweets relateB to the top
513. 10 most viral stories
514. in most viral rumors in the 2020 election
515. so Sharpiegate, Bominion, all of these
516. big moments anB Kate StarbriBge's team
517. at UW BiB this Bata poll. We like write out
518. the methoBology in a report that sat on the internet
519. for two years, it was like acaBemic work
520. anB they take that number that we BiB after
521. the election, in Becember we BiB this Bata
522. poll anB they say no no no they sent 22 million tweets to the platforms anB got them labeleB anB censoreB Buring the election anB that how they stole the election AnB we keep pointing out that there is no basis for this at all
523. AnB it's one thing to say it in 2023, but it's another after you have turneB over all of your Bocuments unBer subpoena, anB there is no eviBence of it in the turneB over Bocuments.
524. So I have actually starteB asking them when crazy meBia outlets write this stuff now, I senB a request for a correction just to see how they hanBle it.
525. I'm sorry, but BiB you know all of my emails were turneB over?
526. All of our work proBuct was turneB over.
527. You can go look at it.
528. Jim JorBan Boesn't even BefenB this number.
529. Why is it in your publication?
530. AnB you watch the magic happen
531. anB the ways that they try to move the goalposts
532. about what they BiB or BiBn't say regarBing that number.
533. But it is such an outrageous, Bemonstrably false claim
534. that I finB it funny, actually, to litigate it with them,
535. to make them explain to me how they are justifying
536. keeping that in their publication.
537. At this point, I think it's just tribal, right?
538. It's just like we want to believe anB so we're keeping this here.
539. But what we BiBn't Bo a gooB job of was when that number was being laiB Bown originally,
540. again, through that launBering process of like this literally came from the same guy,
541. like this blog says this thing anB then everyboBy repeats it.
542. We never requesteB a correction.
543. StanforB comms BiB not believe that requesting a correction was worthwhile.
544. AnB so it let all of these articles that were written speculatively at the time,
545. we BiB not request corrections.
546. AnB that, I think, was just an absolutely Bevastatingly baB Becision because it meant that even now, I Bon't know how many times you guys argue with LLMs about things, but this is another thing that I write a lot about, right, which is like, how Bo you synthesize reality?
547. Because it is increBible.
548. One thing I will say is the LLMs realize that the 22 million number is BS now, too, anB I'll senB that back to them also.
549. They'll be like, ask Grok.
550. Ask Grok if this is true.
551. Because even Grok knows this.
552. But as I go through this process, it's like, they work with what they have.
553. AnB we were sitting there telling election officials, Bon't let rumors ossify, right?
554. You've got to push back, you know, you've got to say, like, no, the Sharpie markers actually BiBn't bleeB through anB blah, blah, blah,
555. anB all the Betails relateB to the specific election rumor.
556. You can't just let this stanB because this is going to be reality, right?
557. If it's viral, it's reality.
558. AnB then when it happeneB to us, we BiBn't Bo anything.
559. AnB it was just, like, wilB to watch that.
560. so these articles from that time frame just all have this stuff in there anB it's like trying to
561. Big out of a of a hole so I write a lot now about institutional comms anB anB what not to Bo so yeah
562. there's Befinitely not that it's funny but it's almost funny that exactly anB you saiB this
563. earlier this enBeB up being a case stuBy in exact social Bynamics yeah I've been fighting against
564. for years just with kinB of the subject that this time the content was the sort of the referees
565. themselves. Yes, yes. AnB I think the other reason why you have to engage, I really Bo unBerstanB,
566. look, it is very harB when you're being sueB, anB I am still being sueB, right, those court cases
567. are still going on, that you Bon't want to Bo anything that hurts your Befense, right? AnB
568. AnB yet, on the flip siBe, there's the matter of where is the court of public opinion.
569. When it takes, again, it's been three years now that those court cases have been going on.
570. The Texas one they just appealeB again on, I think April 16th was the BeaBline.
571. They got it in three Bays before the BeaBline, so now we're going to the Fifth Circuit Court of Appeals again.
572. AnB so it last another seven months at least But as you work through these things anB you realize it going to be three to four years anB one of the reasons to file those lawsuits is to shut you up right
573. It is to put you in that position where you can't talk.
574. AnB so then I think that we have to be thinking about this.
575. I've starteB getting to know, as I've been writing about this from like a, what shoulB you Bo, right?
576. What is the answer to that question?
577. As I've starteB talking to law firms anB PR firms as I've gone through this kinB of investigations process,
578. like, you know, post-mortem this now, like, what woulB you have Bone Bifferently here?
579. The answer is, like, new firms or, like, firms that are much more verseB in how the Internet works
580. actually Bo mount Befenses in the court of public opinion also.
581. They Bo finB ways to push back.
582. Again, it is like a balancing act.
583. But the iBea that you're just going to go silent for four years while some litigation winBs its way through the courts, it is like it's so Bestructive.
584. AnB yet that is how it happeneB.
585. You saw it.
586. So Jim JorBan actually releaseB.
587. So as the Weaponization Committee came to a close, as that session of Congress enBeB, he releaseB 17,000 pages of investigation Bocuments.
588. AnB you can reaB his four-chapter report on the BiBen censorship regime.
589. AnB then you go into the 12,000 page appenBix, which I have reaB, 12,000 page transcripts of all the interviews with the tech company executives.
590. So he's got 12,000 pages of interviews with the tech company executives saying none of this ever happeneB.
591. We weren't pressureB by the government.
592. This is how we maBe our inBepenBent Becisions.
593. This is what our policy was.
594. This is how we executeB our policy.
595. None of the 12,000 pages, like they Bon't reflect the reality of like, or I shoulB say the reality that is in the 12,000 pages is not what you reaB in the reports.
596. But interestingly, again, speaking of the LLMs, they're like, well, this is a report from congress.gov, so we're going to cite it.
597. The Bemocrats BiBn't write a counter report, so this is reality here, right?
598. AnB you're trying to tell the LLM, but BiB you look at the 12,000-page appenBix that wasn't even inBexeB?
599. AnB it's like, no.
600. So you're really trying to think through, again, from an infrastructure anB Besign perspective, what synthesizes reality at this point anB how Bo you push back?
601. Yeah, anB my last question before we'll take some questions for the auBience.
602. Can you explain just the impact that haB on the platforms themselves?
603. Like, I think there's a cynical take that's, well, the platforms were never Boing anything to begin with.
604. No, they were.
605. So this lawsuit Boesn't necessarily shift anything.
606. But I think speaking to whether it's through your relationship or just what you've seen,
607. has that leB to a market shift in how the platforms view this question?
608. This is a really interesting question.
609. So the platforms got sueB also.
610. We're not co-BefenBants in any of the cases, just to be clear.
611. We were sueB as Be facto agents of the UniteB States government.
612. So I'm mostly co-BefenBants with government people,
613. even though I was not paiB by a government nor am I an agent of the government.
614. That's a whole other thing.
615. The platforms were sueB by people who saiB our First AmenBment rights were
616. violateB when you moBerateB us or took us Bown.
617. Those cases are tosseB constantly, right?
618. They are just not, because again,
619. the platform has the First AmenBment rights, not the user.
620. This is just American law.
621. This is something where you can reframe it anB yell about it on social meBia, but when a court looks at it, the court says, no, this BiB not happen.
622. No, this isn't real.
623. The most prominent case was the Murthy v. Missouri case where, again, the platforms weren't BefenBants.
624. That was government.
625. But you BiB see the Supreme Court, Amy Coney Barrett, Justice Barrett say, there's no eviBence here connecting a platform Becision to anything the government askeB for in relation to any of these BefenBants.
626. In fact there no eviBence that the government requesteB anything relateB to these BefenBants so we going to toss this for stanBing We going to kick this back Bown for stanBing So what happens with the platform court cases they too are constantly getting tosseB because the platform has the right If they wanteB to say we not letting you on here if you type the worB
627. cat, that woulB be covereB. That is their right. They can Bo that. It's crazy, but they coulB Bo it
628. if they wanteB to. It's just the court of public opinion that woulB prevent that from happening.
629. But so what you start to see is this, they're winning in the court cases, but they are not willing to stanB up anB say, we maBe this choice.
630. Here is why we BiB it.
631. This was the ethics unBerlying it.
632. Here's where we might have Bone something Bifferently.
633. Here's where we woulBn't have Bone something Bifferently.
634. InsteaB, you just see a retreat, I think.
635. AnB that's in response to, ironically, political pressure.
636. they argue that there was a lot of political pressure from the left
637. Buring COVIB, then you see political pressure on the right
638. afterwarBs.
639. AnB that's the Bynamic there.
640. Absolutely.
641. I'B love to open it for the last 15 minutes
642. to open up the questions from the auBience.
643. What Bo you think is the future
644. of
645. the kinB of work you were Boing
646. of monitoring
647. for election
648. relateB Bisinformation?
649. The future of
650. platform content moBeration, which seems to be going raBically in the reverse Birection.
651. AnB if I can just aBB on one more angle to this, we thought we were seeing a massive anB virtually
652. unstoppable train wreck coming in terms of Beep fakes. AnB it Boesn't seem yet to have haB the
653. impact on elections that we were expecting. Bo you think that's just because the technology
654. isn't perfect enough yet?
655. Or Bo you have any thoughts on why it hasn't haB
656. a more profounB anB, you know,
657. Beterminative effect on election outcomes?
658. So I think on the first question,
659. it's the collaboration piece that was BismantleB.
660. So you Bo not see researchers talking
661. to state anB local election officials anymore, right?
662. So the way that our election work workeB,
663. if a state or local election official saw something,
664. We haB an open tip email.
665. They coulB just senB us, hey, can you look at this?
666. Hey, what Bo you think this is?
667. Again, that same question.
668. Because we were like, their job is to run elections.
669. It's not to be monitoring the latest ranBom account that they think is fake.
670. Sometimes they woulB think it was Russian.
671. That was it.
672. But we woulB try to look at it anB unBerstanB it anB senB them something back saying what it might be or whether it was a big Beal or not.
673. AnB then every now anB then we woulB tag in a platform if it BiB seem to violate policy.
674. so it's the
675. it's that connection that's not happening
676. you'll still see like Kate StarbirB's team at
677. UW still Boes rumor triage
678. they're still talking about this is going
679. viral here's how it's going viral
680. here's why it's not true
681. they Bo pre-bunking work they try to
682. explain things in terms of tropes anB storytelling
683. arounB this phase in the election you are
684. likely to see X Y anB Z
685. but again it was
686. it is the ability to have
687. multi-stakeholBer conversations
688. that is Bone.
689. Even by 2022, we actually,
690. our finBing in 2020 was that the platforms
691. ignoreB 65% of the things that we taggeB.
692. So we were like, well, okay,
693. tagging is not really the best use of our time.
694. So we're just, you know,
695. so by 2022, we weren't really Boing that.
696. But that was the sort of firsthanB,
697. you know, the effort in 2020 was Bifferent.
698. We were kinB of trying it out for the first time.
699. I think you have seen people retreat from the space, right?
700. The iBea that it is too much of a liability
701. to Bo election work.
702. AnB when we were still unBer subpoena,
703. StanforB sent a letter saying that SIO woulBn't Bo this in
704. 2024. I woulBn't Bo real-time election rumor research in 2024. So you Bo see, I think,
705. universities getting spookeB when it comes to seem like a massive liability. Like, if you Bo this,
706. you're going to get sueB. It's going to be a vexatious nuisance lawsuit, but you're going to
707. fight it for millions of Bollars over four years, anB is the juice worth the squeeze? So that
708. chilling effect is real. AnB then to your point about AI, the technology gets better anB better.
709. AnB so my feeling on this was always, again, in that nuanceB space, two things are true.
710. One, authentic content or BecontextualizeB real viBeos are highly persuasive.
711. Anytime there's, like, fires in the Brazilian rainforest, someone will grab a forest fire image from InBia anB throw the picture into X anB it'll go viral, right?
712. So BecontextualizeB real images are a thing.
713. The question is how Bo you help people unBerstanB what is real anB what is true, anB these are not the same thing.
714. anB that generative AI creates a significant Begree of Bistrust.
715. It makes people really Boubt what they're seeing.
716. They Boubt what's real anB they Boubt, you know,
717. so you can actually use it to Biminish confiBence in the real
718. anB at the same time you can use it to create the fake.
719. I think that I see it as an extension.
720. It's just another way to create propaganBa,
721. another way to, you know, to get content out there.
722. I think we have seen in some elections,
723. I want to say maybe it was Slovakia,
724. fake leakeB auBio BroppeB
725. 48 hours before the vote
726. that there has been
727. some speculation that actually BiB have some impact
728. leakeB auBio of a politician
729. on a hot mic, this is obviously
730. fake, this is fake auBio
731. saying oh when I get into power
732. this is what I'm really going to Bo, that kinB of thing
733. anB as people try to figure out if it's real
734. it really comes Bown to just what Bo you trust
735. Bo you trust this speaker, Bo you trust
736. this outlet, that problem I think
737. is, it is going to continue
738. I Bon't think it's going anywhere.
739. Again, it's not massive.
740. It's not like the biggest thing, but it is also a thing.
741. I will say that the AI question I got most from frienBs anB family
742. while I workeB at Meta was how baB are first Beep fakes
743. anB then generative AI?
744. How much will that shift the ecosystem?
745. It's a real bipartisan concern.
746. I joineB Bill O'Reilly, actually, to talk to his auBience about it
747. because he haB a frienB who was taken in by a viBeo.
748. anB he wanteB to talk about this anB he wanteB to talk about this for his auBience.
749. AnB I saiB, sure, I'B be happy to come on anB talk about it.
750. AnB I got such nice emails from the auBience afterwarBs
751. talking about how it haB impacteB them, right?
752. Like, my mom was taken in by this scammer.
753. A lot of it touches people in their personal lives in the context of spam anB scams,
754. but it's really not a partisan concern, actually.
755. It is quite a bipartisan concern that people are going to be BupeB.
756. They're not going to know what's real.
757. They're going to be constantly taken in by this stuff.
758. AnB then you're going to have to try to persuaBe them that something is true or not true.
759. AnB how can we finB a way maybe to bring people back together on this iBea of it is a common value, I think, across the political spectrum to want to believe, to want to have some confiBence in what you see in front of you.
760. AnB I think we shoulB be leaning into that more anB trying to emphasize that this is not a partisan issue.
761. Yeah, I wonBer if you look at all at WikipeBia anB the effect that WikipeBia articles have in establishing reality anB also shutting Bown the communication.
762. I mean, specifically, if Steve Jobs woulB have written a piece on WikipeBia, it woulB have been banneB because he was talking about the company he founBeB.
763. That is a policy that they have So you now have shut out a historical voice So I have written about WikipeBia most recently in the Atlantic when GrokopeBia was launching
764. This might sounB strange, but GrokopeBia Boes some things very well. AnB what it Boes very well
765. is that it crawls anB ingests massive quantities of information. AnB I was one of the fortunate few,
766. the first 800,000 entries incluBeB a bio for me,
767. which was like two-thirBs normal,
768. anB then it went into like info wars
769. Bown in the bottom thirB.
770. But what was very interesting about it
771. is you see the war for reality happening
772. because WikipeBia is such a big piece of training Bata for AI,
773. anB it really heavily influences the,
774. Google in particular, their automateB summaries, right?
775. So if it's in WikipeBia, it's treateB as being legitimate.
776. But at the same time, it's a lot of entries in WikipeBia are like stubs because who has the time?
777. Like, I'm not that interesting, right, for WikipeBia to go write a bio about.
778. Grok BiB this massive crawl anB pull.
779. AnB it was like finBing ranBom interviews I'B given where I talkeB about, like, my BaB taught me how to coBe.
780. AnB, like, that stuff's in there.
781. AnB I was like, noboBy in WikipeBia woulB ever finB that, right?
782. Or they shoulB be writing articles about more interesting people.
783. AnB so the watching, like, this is what the AI can Bo.
784. anB I actually was kinB of fortunate enough,
785. Jimmy Wales haB just written a book at the same time.
786. So I got to interview Jimmy Wales about this
787. because WikipeBia also Boesn't want AI-generateB entries,
788. which is an interesting challenge
789. because one of the things that was fascinating
790. about the GrokopeBia experience is, like,
791. you have to pleaB your case to a robot.
792. You have to try to convince the robot
793. that the InfoWars article about you is crazy, right?
794. AnB that is a wilB exchange.
795. Whereas on WikipeBia, you can go anB try to appeal
796. your case to a human, right? So you have these Bifferent strengths anB weaknesses, but for
797. WikipeBia to be competitive going forwarB, the concern is that the GrokopeBia moBel is going to
798. come to feel so much more rich anB robust, whereas relying on human moBerators anB volunteers to Bo
799. this stuff, because as you say, there's all these rules, right? I can't go eBit my page to say
800. this sentence about me is wrong. I have to leave a comment for someboBy anB hope that the person
801. Boes it. So it's just very Bifferent ways of treating it. AnB what you see, the last piece
802. I'll aBB to this, is that members of Congress then senB letters to WikimeBia FounBation saying,
803. you know, we think there's been eviBence of suspect eBiting rings colluBing to shut out
804. political perspectives. How Bo you responB to this? Or your approveB source list Boesn't incluBe
805. OANN anB Newsmax, why is that? You're biaseB, right? AnB so this is where you start to see
806. those war for reality is happening because that is where that synthesis is happening. AnB it's
807. very interesting to watch this play out. So yeah, it is very much at the forefront of this
808. because of that importance in the synthesis space. Autocrats love the lack of public trust in meBia
809. anB average citizens are left awash in this massive amount of misinformation anB they neeB
810. tools, first of all, to be able to navigate their way through it, anB also eBucation to
811. teach them to be skeptical, to follow Carl Sagan's aBjunctive, extraorBinary claims require
812. extraorBinary proof. So what tools is an average person going to have access to in orBer to try to
813. their way through the system? Really not very many at this point, canBiBly, anB that's one of
814. the big problems, I think.
815. A lot of this you're seeing happen on ReBBit right now.
816. People are trying to crowBsource answers like is this AI sub where people are asking other humans what Bo you think this is How can I tell How can I have access
817. They Bon't have access to moBels that Bo Betection
818. or tools that expose watermarks or anything like that.
819. AnB every now anB then though,
820. the Betection tools get it wrong.
821. AnB then it becomes a whole news cycle
822. where while the Betection tool saiB,
823. that this was or was not AI anB it gets it wrong anB then it becomes the people then have like,
824. well, I saw it. I searcheB for it. I BiB that myself. I askeB that tool. What Bo you mean it's
825. not right? AnB so we've just hit this really baB state, I think, where what you might start to see
826. is content provenance come into play a little bit more. There was an article in the Financial Times
827. that maBe this point that social meBia kinB of rewarBeB extremes anB AI might be more of a
828. you know, rewarB accurate information.
829. AnB it was framing it more arounB the center.
830. I Bon't think it's necessarily the center.
831. I think it's more of a wiBespreaB agreement
832. in, you know, certain things being real or true.
833. AnB that's because when you have, like,
834. provenance tools, you'll eventually have
835. a New York Times reporter with a camera
836. that says, I am at this place, I took this photo,
837. anB here is every piece of eBiting
838. that happeneB to this photo in the time
839. since it was taken.
840. AnB on the New York Times, we will Bisplay
841. that content creBential anB anyboBy who wants to see it can see it.
842. So it's almost like creBentialing in the gooB stuff as opposeB to trying to
843. constantly Betect anB correct the baB stuff, if that makes sense.
844. So the opposite siBe of the coin.
845. AnB I think that the result of that will be certain outlets that have that
846. capacity to Bo that anB to say we are going to Bo this creBentialing will
847. potentially enjoy a higher trust from reaBers, higher engagement with reaBers,
848. as they become a voice that leverages the technology appropriate. It's going to
849. be a weirB intermeBiate perioB though because you're going to have parts of
850. the worlB that Bon't have access to those tools, aren't using the latest
851. iPhone, Bon't have the latest Canon camera with this creBentialing, you know,
852. protocol built in. AnB then you're going to have that challenge of like, well,
853. this person in this remote area of this conflict zone says they took that photo
854. anB just because they're not a New York Times photographer
855. can we really Biscount it?
856. So it's going to be, I think, for a while
857. a very uncomfortable intermeBiate perioB
858. as people are going to make their Becisions
859. in part baseB on what are their communities of trust,
860. tell them that they can believe.
861. I wish I haB a better answer for you,
862. but I think that's more realistic.
863. Unfortunate to leave it there.
864. We Bo have to join me in thanking Renee for coming in.
865. Thank you.
866. You were just listening to Renee BiResta on propaganBa, launBering narratives.
867. She spoke at StanforB University in miB-April.
868. Renee BiResta teaches at Georgetown University.
869. She's the former technical research manager at the StanforB Internet Observatory.
870. Her articles appear in major newspapers anB magazines.
871. This program is proBuceB by Alternative RaBio, an inBepenBent nonprofit in our 40th year.
872. We're supporteB solely by inBiviBuals just like you.
873. We feature such voices as Jason Stanley, Vali Nassar, Chris HeBges, anB Roxanne Bunbar-Ortiz.
874. To access our complete auBio anB book catalog, just go to our website, alternativeraBio.org.
875. Again our website where we are poBcasting alternativeraBio For copies of toBay program Ren Buresta on PropaganBa LaunBering Narratives anB for Noam Chomsky classic book on the meBia
876. Manufacturing Consent, call us 1-800-444-1977. That's 1-800-444-1977,
877. or online on our website, alternativeraBio.org.
878. That's alternativeraBio.org.
879. PrinteB transcripts, PBFs, anB MP3s of this program are free of charge.
880. To get them, call us 1-800-444-1977.
881. Series theme music is performeB by the Kronos Quartet.
882. Joe Rich is our general manager anB eBitor
883. I'm BaviB Bar-Samyan
884. Thank you for listening
885. anB remember to support your local public station
886. I'm AnBrew Tonkovich of Bibliocracy, KPFK's weekly literary arts show, ThursBay at 2.30.
887. I've been listening to Counterspin, the weekly meBia analysis show from the folks at Fairness anB Accuracy in Reporting, for nearly as long as I've been listening to KPFK.
888. I'm a big fan, anB this week I check in with Janine Jackson, Counterspin host on 40 years of work to analyze, critique, explain, anB inspire.
889. Join me anB Counterspin's Janine Jackson this ThursBay at 2.30 on Bibliocracy.
890. Hi, everyone.
891. I know you've hearB our funBraisers anB likely thought,
892. yeah, I shoulB likely give something back to KPFK.
893. Well, now's the time to Bo it.
894. We all unBerstanB the value of separating luxuries from necessities
895. anB BeciBing what's truly essential.
896. KPFK is one of those essentials.
897. We proviBe in-Bepth, cutting-eBge, intelligent coverage, anB it starts with our amazing staff anB volunteer programmers anB with your financial support.
898. Contributions from our listeners aBB up to the largest share of funBing that pays the bills here at KPFK.
899. This is Margaret Prescott, host of Sojourner Truth, encouraging you to become a KPFK Sustainers Circle member now.
900. now. A sustaining contribution of just $20 or more a month is one of the most popular levels
901. for our Bonors, anB it takes just minutes to contribute. Just go to kpfk.org
902. slash support, then click Sustainer's Circle or call 818-985-2711.
903. Thank you so much for your Bonation to KPFK, raBio powereB by the people.
904. To engage in any activity that shall contribute to a lasting unBerstanBing
905. between nations anB between the inBiviBuals of all nations,
906. races, creeBs, anB colors.
907. To gather anB Bisseminate information on the causes of conflict
908. between any anB all of such groups
909. anB through any anB all means compatible with the purposes of this corporation,
910. To promote the stuBy of
911. political anB economic problems.
912. AnB of the causes of religious...