Social Media In Wartime + Betting on the Future + A.I. Passes the Smell Test

Episode Summary

Executive Summary: The episode opens with a comic bit about the app Bonk, a deliberately annoying social tool that sends repeated push notifications, then turns serious: Kevin and Casey argue that X/Twitter is failing as a real-time news platform during the Israel-Hamas war because misinformation, monetized virality, and weakened verification are drowning out reliable reporting. They then discuss whether Threads, Bluesky, and Mastodon can fill the gap, before moving to prediction markets as a way to aggregate information, and finally explore Osmo’s effort to teach AI to smell, including a live odor-testing segment.

Main Topics: Bonk: a deliberately annoying social app (Priority: 3/5): The hosts joke about Bonk, an iOS app that lets users repeatedly 'bonk' friends with push alerts and leaderboard tracking. They frame it as a playful but mildly harassing alternative to messaging—essentially 'Yo' as benign annoyance. X/Twitter’s collapse as a real-time news source during war (Priority: 5/5): Kevin and Casey argue that X is no longer a reliable place to understand fast-moving events like the Israel-Hamas war. They cite false videos, old clips repackaged as new, and the way paid verification and monetization incentivize misinformation. Threads, Bluesky, and the search for a new public square (Priority: 5/5): The discussion shifts to where journalists and users are migrating for live news and commentary. Threads is emerging as a surprisingly active venue for journalists and pundits, even though Meta has signaled it does not want to become a news-first product. Prediction markets as collective forecasting tools (Priority: 4/5): Kevin reports from Manifest, a conference for forecasting and betting enthusiasts. The hosts debate whether prediction markets can improve truth-seeking by forcing people to wager on beliefs, while also worrying about manipulation, insider trading, and gambling incentives. AI that can smell (Priority: 4/5): The final segment features Osmo CEO Alex Wiltschko, who explains his company’s work on an AI smell model and a 'principal odor map.' The hosts try an odor test and discuss potential applications in fragrance, food, and disease detection. The limits of platform control over information (Priority: 5/5): Across the episode, the hosts question whether large platform companies—especially Meta and X—should be trusted to organize public knowledge. They suggest mainstream journalism still matters, but note that people also crave social context and first-person accounts.

Key Arguments: X/Twitter is no longer functioning as the 'best answer to the question of what is going on' because its incentives now reward virality over reliability. Elon Musk’s changes to verification and monetization have created conditions where misinformation can be rewarded financially. Threads is unintentionally attracting journalists and real-time discussion even though Meta has tried to position it as a lighter, less news-heavy app. Prediction markets could improve forecasting by aggregating dispersed information, but they also risk encouraging manipulation and normalizing gambling on public life. AI smell technology may be useful for detecting disease, designing fragrances, and mapping odors more systematically than humans can. Mainstream reporting remains the best starting point for understanding major conflicts, but social platforms still serve a separate need: seeing the conversation, not just the facts.

Data Points: Yo funding round: $1.5 million at a $10 million valuation - Referenced as a comparison to the earlier one-function messaging app Yo. Bonk app availability: TestFlight only; requires iOS 17 - Kevin says he couldn’t install Bonk because his device lacked the required iOS version. Bonk leaderboard example: 5,000+ Bonks in a week - One Bonk user reportedly sent more than 5,000 Bonks that week; Kevin had sent 438. X monetization: $8 per month - Paid verification on X was discussed as a key change affecting content ranking and incentives. Manifold user base: about 50,000 users - Kevin cites this as small relative to major betting or social platforms. Prediction market bet size: 100 mana then 200 mana - Kevin and Casey place bets on whether Linda Yaccarino will still be CEO of X on a future date. Prediction market probability shift: 10% then 8% - The market odds moved sharply after their large no-bet. Odor study sample size: hundreds of thousands of molecules - Osmo trained its model by digitally sniffing a very large molecule database. Odor test panel: 400 molecules - The company selected 400 molecules for human raters in the odor Turing Test. Evaluation benchmark: closer than the median panelist - Osmo says its model usually predicted odors closer to the panel mean than the median human rater.

Pivotal Quotes: "“What if Yo was just a tool for benign harassment?”" — Casey Newton: Summing up the Bonk app’s core concept as playful but annoying social contact. "“There has been no place to actually get good, reliable information about this conflict.”" — Kevin Roose: Opening the discussion of how social platforms are failing during the Israel-Hamas war. "“We have run the experiment where Mark Zuckerberg and Adam Masseri run the global information ecosystem. It did not end well.”" — Kevin Roose: Arguing against trusting Meta to manage news distribution and conflict coverage.

Implications: Listeners are urged to be skeptical of platform-driven news, especially in crises, and to value professional reporting while recognizing social networks’ role in live context. The episode suggests future public discourse may fragment across multiple apps, with prediction markets and sensory AI representing other ways tech is trying to model reality.