Jan 23, 2019 · 33m · y-combinator
Tracking Political Manipulation Through Social Media - Samantha Bradshaw · Y Combinator
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this interview, Oxford Internet Institute researcher Samantha Bradshaw analyzes the evolution of automated social media bots, foreign election interference, and digital manipulation tactics. She explores corporate accountability, user psychology, and regulatory frameworks required to protect democratic resilience in an era of computational propaganda.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. How this is scored →
The partners as informed peer 4.3 Guest teaching 5.7 Guest disagreement 0.1 The partners pushing back 0.3
speaking balance: gold is the partners, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 15:01 Direct critique of government content moderation policies
The guest firmly rejects government content takedown legislation like NetzDG, arguing that targeting content rather than underlying algorithms creates collateral censorship and aids authoritarian regimes.
Hardest push from the partners ▶ 17:54 Challenging the premise of purely technical fixesThe host pushes back against the assumption that cleaning up fake news fixes political manipulation by questioning how platforms can overcome human selective exposure bias.
Biggest teaching moment ▶ 25:38 Explaining Oxford's junk news classification and midterm findingsThe guest details Oxford's 5-point empirical criteria for junk news and shares quantitative findings revealing that low-quality news shares actually increased from 1:1 in 2016 to 1.3:1 in 2018.
The partners hold their own ▶ 11:24 Citing the researchers' fifty posts per day thresholdThe host shows direct knowledge of the research group's exact methodology by citing their 50 posts per day heuristic for automated account classification.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The partners as informed peer | Guest teaching | Guest disagreement | The partners pushing back | Why |
|---|---|---|---|---|---|---|
| Defining Social Media Bots and Automation | 3 | 5 | 0 | 0 | The host sets up the framing by asking the guest to define what bots actually do. The guest provides a structured educational breakdown comparing useful scrapers like Google Search with social amplification bots and interactive chatbots. | |
| History and Evolution of Bot Manipulation Tactics | 4 | 5 | 0 | 0 | The host asks about historical timelines and connects bot activity to gaming algorithmic recommendation signals. The guest explains how bot operations evolved from crude automated retweets into sophisticated cross-platform search engine optimization tactics. | |
| Content Creation and Manipulation on Messaging Apps | 4 | 6 | 0 | 0 | The host draws parallels between WhatsApp content spread and Reddit meme culture from 2016. The guest educates on the unique methodological difficulties of studying closed messaging ecosystems like WhatsApp in India and Brazil. | |
| Infiltration Mechanics Across Twitter and Facebook | 5 | 5 | 0 | 0 | The host demonstrates familiarity with the guest's specific research metrics, citing their 50-tweets-per-day automation threshold. The guest elaborates on account verification differences between Twitter and Facebook. | |
| Corporate Accountability and Legislative Regulation | 5 | 7 | 1 | 1 | The host frames platform business incentives around active user counts versus advertiser ROI and references German regulatory models. The guest delivers an in-depth analysis showing how content-policing laws like NetzDG induce collateral censorship and give authoritarians dangerous precedents. | |
| Human Psychology, Platform Monopolies, and Personal Privacy | 4 | 5 | 0 | 1 | The host questions whether eliminating fake news would even solve echo chambers given innate human confirmation bias. The guest affirms the selection effect and discusses how platform monopoly lock-in worsens democratic discourse. | |
| Data Privacy, Mueller Report, and US Election Research | 5 | 7 | 0 | 0 | The host asks targeted questions regarding statistical controls and geographic swing state targeting. The guest explains Oxford's research methodology and surprising findings showing junk news sharing ratios rose in the 2018 midterms compared to 2016. |
Statements from this episode (16)
Sophisticated bots use natural language technology to reply in comment threads
“Some of the more sophisticated bots might actually interact with real people so they blend, like, chatbot bot technology and, you know, we all have been to those customer service pages where the thing pops up at the bottom and says, hi, can I help you today? T…”
Authoritarian governments have used social bots since early platform days
“From some of the research that I've done here at the OII and on the Computational Propaganda Project, we know that these techniques have been experimented with by governments for a long time. We've had evidence of them going back to, you know, even the early d…”
Political actors use traditional SEO tactics to game Google and YouTube
“We're seeing a lot more gaming of the algorithms as well, and more of these kind of sophisticated techniques. So not just liking and sharing, but using specific keywords to get content trending, to try to get things at the top of Google's search algorithm, try…”
Disinformation actors target locally dominant platforms like WhatsApp in India
“The people who want to manipulate public opinion, they're gonna go to where the people are, and the platforms that the people are using. So in the US, we tend to see a lot of these campaigns on Twitter and Facebook, because that's the platforms that the majori…”
Oxford researchers found widespread meme-based disinformation in Brazilian WhatsApp groups
“We looked at WhatsApp in Brazil, for example, and in that study we joined a bunch of different groups Were looking at the kinds of images and the kinds of conversations that were being shared. And there were a lot of memes that were being used to public, to pu…”
Twitter's open API and anonymity allow more automated fake accounts
“Twitter tends to have a lot more of these fake accounts that use some kind of automation. There are also a bunch of tools that allow you to, you know, automate your activity on Twitter, things like Hootsuite and whatnot.”
Fake Facebook accounts hold more influence because users expect authenticity
“Because it's so hard to create fake accounts on Facebook, the accounts that are fake might actually be a little bit more powerful because people don't expect there to be as many fake people on Facebook as they do Twitter, and so they might actually have more o…”
Platforms ignored fake accounts because active user metrics drove market valuations
“I think for a long time there hasn't been an incentive for them to, because the more active accounts there are on these platforms The more they're valued on the market, right? Because all of a sudden there's this huge user base of people, people that these pla…”
Direct content regulation fails to fix the underlying drivers of disinformation
“When governments go after the content of what's being shared I think that's a mistake. I don't think that's getting to the underlying problem that's sort of fueling the fake content or the disinformation to spread or to go viral in the first place.”
Authoritarian regimes weaponize Germany's content moderation laws to silence dissent
“We're already also seeing authoritarian governments adopt this law into their own into their own legal systems to silence dissent and to go after journalists who are publishing so-called fake news and whatnot.”
Americans shared junk news at a one-to-one ratio in 2016
“In 2016 we found that Americans on average were sharing about a one to one ratio of junk news to professionally produced information.”
The junk news sharing ratio worsened to 1.3:1 during 2018 midterms
“In 2018, when we redid this analysis during the midterm, the ratio of junk news actually went up. And so it went to like 1.2 or 1.3 to one.”
Social media users in 2016 swing states shared more junk news
“And we also in 2016 did a study on the swing states. And you know, in actual swing state, States people were sharing more junk news compared to uncontested ones.”
US users share vastly more junk news than European democracies
“During the UK elections, and Germany, and France, and Sweden, Mexico, all of those countries have much lower levels of junk news to professionally produced news shares than the US.”
Social media platforms will fail to curb manipulation before 2020 elections
“The fact that we saw an increase already in 2018 you know, I don't think the platforms are going to be able to get their act together in time for 2020.”
Deepfakes are overhyped and will not become a major political threat
“I think, you know, there's a lot of hype about deep fake right now. I don't know how, how real it's actually going to be. And, you know, we're already seeing, you know, a lot of the research agencies like DARPA and things like that work on being able to detect…”