The Next Wave of Social Media | Instagram Founders Kevin Systrom & Mike Krieger

Core Strategic Shift in Social Media

• The industry is undergoing a fundamental transition from "social graph" directed discovery (content from people you follow) to machine learning-driven content discovery (TikTok/Reels style).
• Mike suggests the "complete abandonment" of the social graph is too binary, but notes that relying solely on friend networks is a "great mistake" made by social networks over the last 20 years.
• Social networks were originally "great filters," operating on the thesis that friends share similar interests; however, algorithms now reveal that friends (e.g., Mike and Harry) often have divergent content preferences despite strong social overlap.
• Current algorithms prioritize "median voter theory," serving middle-of-the-road content (e.g., tech news, generic ChatGPT topics) rather than the "amazing texture" of individual user interests.
• TikTok (and ByteDance) are credited with unlocking the potential of user behavior as a primary signal, demonstrating that content relevance can be decoupled from social connections.

The Role of Algorithms and User Experience

• The speaker advocates for "learned association" and "learned affinity" through behavior to replace social graphs as the core discovery signal.
• There is a stated goal to ensure algorithms work for the user rather than for the company, aiming for a long-term improvement in user experience.
• Acknowledged risks in this shift include the potential for "filter bubbles" and the distribution of misinformation by high-reach users, similar to existing issues with traditional social networks.
• The speaker draws a parallel between the 2007-2008 mobile shift (where early Facebook apps were non-native HTML) and the current ML shift, suggesting the market has only "half-stepped" the move to native ML.
• While discovery may become unconnected, the "desire to talk" about content remains; discovery is inverted from a "social" gatekeeper to a conversation starter.
• Example: Finding a housing policy article on Artifact triggers immediate sharing to a DM chat or WhatsApp group to start a conversation.
• Shared interests (e.g., Golden State Warriors rumors) are identified as a mechanism to strengthen relationships by providing new conversation topics, even if the initial content source is not a friend.

Community Building and Verticals

• Future community formation is envisioned around specific interests or taste profiles rather than pre-existing social connections (e.g., people meeting via "World Cup" interest groups who have never met in person).
• Artifact has already observed "glimpses" of this in beta groups where users connected purely through shared interests like sports or events.
• The strategy does not necessarily require integrating tools like Discord or building verticalized community management infrastructure immediately; the focus remains on content-driven connection.
• The speaker dismisses the notion that investing in "verticals" or "spaces" is superior, labeling it as "BS" and a passing trend.
• Success is attributed 100% to "team plus the area," emphasizing that the right team can make "unsexy" areas (like photos in 2010 or cars today) viable and profitable.
• Facebook is highlighted as a professional incubator that taught the "scientific method of growth" (A/B testing, multivariate testing, statistical power) to Product Managers.
• The Instagram team succeeded not through pre-existing insight into the "unsexy" photo market, but by leveraging learned growth skills and luck, challenging industry data that suggested photos were a low-value category compared to video.

Market Outlook and Investment Thesis

• The speaker argues against the narrative that "social is a bad place to invest," noting that no truly new consumer social network has emerged in the last 10 years.
• Concerns regarding the defensibility of "positive-only" social networks in Europe are met with skepticism, citing the difficulty of copying teams with deep execution history and growth skills.
• The speaker implies that "feature richness" or specific mechanics are less defensible than the underlying ability to execute growth through data-driven methods.
• The market trend is described as a cycle where "generative AI" and "sharing economy" verticals come and go, but the core challenge remains team execution in broad social spaces.
• The speaker expresses confidence in pursuing "unsexy" opportunities (like photos vs. video) where others see patterns of failure, framing them as significant market opportunities.