# Is the internet really dead? Source: https://www.youtube.com/watch?v=qWK47sqLmJQ Recap page: https://rapidrecap.app/video/qWK47sqLmJQ Generated: 2025-11-24T13:35:24.838+00:00 --- ## Quick Overview The internet is not literally dead, but it feels that way because a significant and growing portion of online activity, potentially over 50% of traffic by 2025 according to one estimate, is driven by sophisticated bots, making it difficult to reliably distinguish between human and automated interactions. **Key Points:** - Global internet traffic is estimated to be over 50% bots by 2025, according to the 2025 Imperva Bad Bot Report (1:04). - The video explores the concept of the 'Dead Internet Theory,' where much of online activity is automated, often for financial gain (2:55, 4:00). - Bot detection algorithms, like Botometer X, analyze features to score accounts between 0 (human) and 5 (bot), though experts disagree on precise estimates (7:01, 9:41). - A highly bot-like account (@Andreah83940354) created in late 2019 scored 4.5/5 on Botometer (7:06). - Politicians' followers were analyzed using follower creation date vs. follower count, showing spikes of non-organic growth, such as for Turkish politician Mustafa Demir (5:58). - The host is working with organizations like New Press to bring transparency to this issue by exposing the data sources and analysis methods (14:30, 14:46). ![Screenshot at 1:04: A stacked area chart from the Imperva 2025 Bad Bot Report shows that bots accounted for slightly over 50% of global internet traffic in 2023, illustrating the scale of automated activity.](https://ss.rapidrecap.app/screens/qWK47sqLmJQ/00-01-04.png) **Context:** The video addresses the 'Dead Internet Theory,' the notion that most online content and interactions are generated by bots rather than real people. The host investigates this claim by looking at research papers, data from bot detection tools like Botometer X, and analyses of high-profile social media accounts to determine the actual prevalence and nature of automated activity online. ## Detailed Analysis The video concludes that while the internet is not entirely dead, the feeling that it is stems from the massive scale of automated activity, with estimates suggesting bots account for over half of all internet traffic by 2025, according to the Imperva 2025 Bad Bot Report (1:04). The host explores the 'Dead Internet Theory,' noting that this high volume of bots is often driven by malicious activity (like spam or scams) or financial incentives (like boosting follower counts for influencers) (2:55, 4:40). The video contrasts this with early social media promises of authentic connection (12:37). Researchers like Onur Varol and Lynnette Ng of Carnegie Mellon University estimate that approximately 20% of social media comments come from bots, while other research estimates range between 9% and 15% of the total population (9:12, 9:22). Tools like Botometer X attempt to quantify this by scoring accounts on a 0-5 scale (human to bot), but these metrics are not universally agreed upon, as demonstrated by the analysis of a highly suspicious account scoring 4.5/5 (7:06). Furthermore, analyses of political figures' followers (like Turkey's Mustafa Demir) show clear patterns of inorganic growth, where large numbers of followers were created in short, suspicious timeframes (5:51, 5:58). The discussion highlights that malicious bots are often distinguished from benign ones, but the sheer volume of bot activity shapes the online experience to feel increasingly inauthentic. ### Introduction to Dead Internet Theory - The host introduces the core question: Is the internet mostly bots? - The sentiment is fueled by headlines suggesting massive bot presence (0:26). - The host plans to investigate the basics of what a bot is and how they work (1:54). ### Evidence of Bot Activity - A phone bot farm setup featuring racks of phones is shown as a physical example of automation (2:01). - Data brokers sell personal information, contributing to data exposure (3:35). - Botometer X is introduced as a tool to score user accounts from 0 (human) to 5 (bot) (7:01). ### Expert Analysis on Bot Prevalence - Lynnette Ng estimates the population of bots on social media is probably about 20% (9:22). - Prof. Darren Linvill notes that distinguishing bots is difficult because they mimic human behavior, and estimates vary wildly (9:43, 10:17). - The danger is that bots can influence public opinion or amplify outrage cycles (2:55, 8:18). ### Visualizing Anomalies - Heatmaps showing follower growth for Turkish accounts like NTV and Mustafa Demir reveal clusters of suspiciously new accounts gaining followers rapidly (4:57, 5:51). - A comparison scale shows a profile scoring 4.5/5 (highly likely bot) versus a known human profile scoring near 0 (7:01). ### Conclusion and Call to Action - The host concludes that the internet feels dead because human interaction is increasingly shaped by sophisticated, non-human forces (13:15). - The video ends by promoting New Press, a new journalism project aiming for transparency in reporting (14:45). ![Screenshot at 1:04: A stacked area chart from the Imperva 2025 Bad Bot Report shows that bots accounted for slightly over 50% of global internet traffic in 2023, illustrating the scale of automated activity.](https://ss.rapidrecap.app/screens/qWK47sqLmJQ/00-01-04.png) ![Screenshot at 2:01: A video clip displays a phone bot farm, showing dozens of mobile devices mounted on racks, all connected to a central computer for automated activity.](https://ss.rapidrecap.app/screens/qWK47sqLmJQ/00-02-01.png) ![Screenshot at 7:01: The Botometer X interface shows a user score of 4.5/5, indicating a high probability of the account being a bot, visualized by the gauge pointing into the red zone.](https://ss.rapidrecap.app/screens/qWK47sqLmJQ/00-07-01.png) ![Screenshot at 10:14: A split screen interview between the host and Darren Linvill, Professor at Clemson University, discussing the difficulty in estimating bot prevalence accurately.](https://ss.rapidrecap.app/screens/qWK47sqLmJQ/00-10-14.png) ![Screenshot at 14:16: The video credits appear on screen, listing key production roles, including Christophe Haubursin for video editing and Saud Mustaneer for animation.](https://ss.rapidrecap.app/screens/qWK47sqLmJQ/00-14-16.png)