Finding underpriced attention in a world disrupted by AI

Executive overview

Every medium starts cheap and gets expensive once marketers pile in. The window to act is the gap between when an audience is there and when the ad world catches up.

Today that gap is Facebook Reels — large scale, almost no competition. AI is the next macro disruption, but most AI investments will fail for the same reason most NFT projects went to zero: the macro is right, the individual picks are mostly wrong.

The edge goes to whoever spots real consumer behaviour before the money does.

Underpriced attention: the core framework

• Attention is always mispriced somewhere — the skill is finding it before the market does
• 1998: email newsletters at 91% open rates; Google AdWords wine terms at $0.05 — both later crowded out
• Right now: Facebook Reels has audience scale but almost no advertiser competition
• Spotify's biddable ad product lets anyone place a recorded ad on the free tier — another overlooked channel
• Remnant outdoor billboards (bought between contracts) deliver strong reach-per-dollar
• Influencer spend can be the best or worst value depending on selection — treat it as a variable, not a category
• The Fortune 500 runs on lagging reports; their slowness is the ongoing arbitrage for everyone else

AI and the next wave of disruption

• AI is a foregone conclusion as infrastructure — treat it like oxygen, not a bet
• The error: writing checks because a technology is happening, not because a specific business is working
• In the MySpace era there were dozens of social networks; almost all failed despite the macro being correct
• AI creative tools are proliferating at the same speed — 20 competitors to Midjourney appeared overnight
• Winners will be determined by operators, not by the technology itself
• Be patient: wait until something is actually working before committing capital or time
• VR and the metaverse are examples of investing in tomorrow too soon — the humans weren't there yet

Social media over the next decade

• Social platforms behave like hot restaurants — each generation needs its own place
• TikTok will eventually be made fun of by Gen Alpha the way Instagram was made fun of by TikTok users
• Expect two or three new major platforms over the next 5–10 years
• The first 24–48 months of any new platform is a disproportionate opportunity window — distribution is cheap, competition is low
• A decentralised social network at meaningful scale is likely within 10 years; Web3 provenance gains relevance as deepfakes spread

On luck and work ethic

• Luck is real — where you're born, your parents, your health are all variables outside your control
• The invalid use of luck: deploying it as a weapon to explain away others' success while producing nothing yourself
• Natural talent alone is insufficient; the finished product reflects compounding hours of work
• Grit without selectivity is a trap — doing everything at 100% leaves no room for choosing the right things