Trevor Thompson 2.2 8 ideas

The group discusses the "Dead Internet" theory, noting that "60 or 70% of the replies... are slop." Thompson mentions seeing Community Notes appearing on Instagram, indicating the bot/misinformation problem is spreading beyond X. If a significant percentage of engagement on social platforms is non-human, advertising metrics (CPM/CPC) are effectively fraudulent or highly inflated. As advertisers realize they are paying to show ads to "agents on a Mac Mini" rather than humans, ad revenue for open social platforms (like Meta) faces a structural de-rating risk. Watch for deterioration in ad-efficiency metrics as the "bot problem" dilutes the value of user bases. Meta successfully implements "human-only" filters or biometric verification to restore advertiser trust.

Thompson explicitly contrasts his decentralized solution with the institutional standard: "How does a business solve it like Coinbase? They KYC... You have to give them government documents... scan your face. That is the extreme degree of knowing who is on the other end." As the "open" internet becomes flooded with 70% bot traffic and "slop," trust becomes the premium asset. While Thompson critiques KYC for privacy reasons, he acknowledges it is the standard for businesses. As AI agents degrade trust in open networks, capital and high-value interactions will flee to "walled gardens" like Coinbase where identity is cryptographically and legally assured. Long Coinbase as the "Fortress of Identity" in a "Dead Internet" environment. Regulatory overreach or privacy-focused users migrating to decentralized identity solutions (like Ethos) over time.

Thompson notes the ease of bot creation: "Anyone with a little bit of prompting to Claude can go write their own bot... You have a little agent on a Mac Mini running." The "Slop Super Cycle" is driven by the commoditization of Large Language Models (LLMs). Every bot, agent, and automated reply requires inference compute. "Claude" (Anthropic) is heavily backed by Amazon and Google. The explosion of agents—even the spammy ones—drives massive revenue for the cloud/model infrastructure providers. Long the "Pick and Shovel" providers of the AI agent explosion. Regulatory crackdowns on AI agents or platform-level bans (e.g., X blocking API access) reducing inference volume.

"How does a business solve it like Coinbase? They KYC, right? They they have to know their customer... You have to give them government documents... scan your face." As the open internet becomes flooded with indistinguishable AI agents and "slop," the value of "walled gardens" where human identity is cryptographically and legally verified increases. Coinbase's strict KYC compliance—often seen as a friction point—is flipping into a competitive moat. It creates a high-trust environment for transactions and interactions that social platforms cannot match. Long COIN as a "Trust Infrastructure" play. Regulatory crackdowns or high compliance costs slowing user growth.

"Anyone with a little bit of prompting to Claude can go write their own bot... the number of agents... can basically scale way more than humans could scale." The "slop super cycle" described by the guest implies an exponential increase in API calls and inference compute. Every one of the millions of bots spamming social media requires processing power from the underlying LLM providers. Microsoft (OpenAI), Amazon (Anthropic/Claude), and Google (Gemini) are the "arms dealers" selling the compute required to flood the internet with agents. Long the Hyperscalers/AI Model providers. Regulatory intervention on AI agents or platforms blocking API access (e.g., X blocking scrapers).