a16z Podcast | The Meaning of Emoji

Network Effects: History, Taxonomy, and Defensibility

• Origin and Definition: The term "network effects" was first coined in an AT&T 1907 annual report, noting that customer value increased as more people joined the telephone network, making new connections desirable for existing users.
• Metcalfe's Law: Formulated by George Gilder (and later associated with Robert Metcalfe) during the Ethernet era, this principle posits that a network's value grows exponentially (specifically $n^2$) as the number of users increases, a concept that moved from hardware to software dominance in the late 20th century.
• Evolution to Software: Network effects shifted from hardware-based connectivity (telephone lines, Ethernet cards) to software platforms (Facebook, Microsoft OS), becoming a primary source of defensibility in the digital age where traditional barriers like geography and physical assets have eroded.
• Four Core Defensibilities: In the modern software landscape, businesses rely on brand, scale, embedding, and network effects; network effects are considered the most "native" to technology because silicon-based systems inherently seek connectivity.
• Direct Network Effects: Defined as value derived directly from connections between users, this category includes:
  • Personal Direct: Driven by human relationships (e.g., Facebook), though pure social utility may have a shorter lifecycle than utility-driven networks.
  • Personal Utility Direct: Enhanced when personal connections are combined with functional utility (e.g., WhatsApp for communication, WeChat for payments and services), resulting in higher engagement and longer retention.
  • Market Networks: A sub-type of personal utility where financial transactions occur within the network (e.g., marketplaces), creating a stronger barrier to exit as users derive income from the platform.
• Two-Sided Network Effects: Occur when two distinct user groups (e.g., developers and users, buyers and sellers) must be balanced, including:
  • Platform Two-Sided: Where a core product attracts third-party developers or partners who add value to the ecosystem (e.g., Microsoft in the 1990s).
  • Asymptoting Two-Sided: Where the network reaches a saturation point of utility, after which competitors can effectively enter with lower pricing or niche features.
• Bandwagon Effects: Driven by social conformity and popularity rather than intrinsic utility (e.g., Slack adoption among tech workers), these effects can be powerful but are often distinct from genuine product-market fit defensibility.
• Distinction from Virality: Viral growth focuses on acquiring new users (lowering CAC), whereas network effects focus on retention and defensibility; a business can have viral loops without strong network effects, but network effects are required for long-term moats.
• Failure Cases:
  • Myspace: Failed to establish a strong personal network effect because it allowed pseudonyms and lacked real-name identity, turning into an "entertainment" platform rather than a utility with high retention.
  • Path: A niche network capped at 150 users failed to achieve scale due to intense competition from superior alternatives like WhatsApp and Facebook Mobile, rather than a lack of product-market fit within its niche.
• The Tipping Point: Identifying the moment a network effect becomes self-sustaining is difficult; success relies on continuous improvement in product-market fit rather than a single measurable metric, with early adoption in specific dense clusters (e.g., Harvard for Facebook, Silicon Valley for LinkedIn) often serving as the catalyst.

Emoji: Standardization, Politics, and Cultural Evolution

• Regulatory Body: Emoji are governed by the Unicode Consortium, a nonprofit based in Mountain View, California, with 12 full voting members (including tech giants like Apple, Google, Microsoft, and Samsung, plus SAP, Huawei, and the Government of Oman) that meet quarterly to approve new characters.
• Proposal Criteria: To be approved, a proposal must demonstrate unique meaning, ambiguity for contextual flexibility, and utility, while adhering to a ban on specific categories like celebrities, deities, and logos (though exceptions like the Easter Island Moai exist due to specific cultural usage).
• The Dumpling Emoji: Jenny Lee led a successful campaign to add dumpling, takeout box, chopsticks, and fortune cookies to Unicode 10 (announced 2017), arguing for their global ubiquity and symbolic completeness (e.g., the takeout box representing both cuisine and delivery).
• Platform Implementation Variance: While Unicode defines the code point, individual OS vendors (Apple, Android, Microsoft) control the graphical rendering, leading to significant interpretive differences; for example, the "bunny ear" emoji is a party scene on iOS but a Playboy-style profile shot on Android.
• Political and Social Constraints: The finite palette of emoji creates friction regarding representation, leading to:
  • Gender Roles: Women are historically limited to roles like bride, princess, or dancer, whereas men have diverse professional roles (police, doctor, Santa).
  • Weaponry: Proposals for a rifle emoji were suppressed by the Unicode consortium due to safety and political concerns, despite the existence of a generic gun emoji.
  • Flags: Flags are implemented as composite characters (country codes) rather than single images, offloading the political decision of representation to handset manufacturers (e.g., Microsoft displays the text code rather than a flag).
• Inclusivity and Skin Tones: To address lack of representation, Unicode introduced the Fitzpatrick scale for skin tones, allowing users to customize emoji colors; this shift was empowering for people of color but highlighted the initial default bias toward Caucasian representation.
• Emoji vs. Stickers:
  • Emoji: Standardized text characters (Unicode code points) that can be used in email subject lines, are screen-reader accessible, and ensure cross-platform compatibility.
  • Stickers: Proprietary, inline images (e.g., Bitmoji, Kimoji) controlled by specific apps, offering higher customization but lacking standardization and accessibility; they function as a separate, non-universal layer of communication.
• Machine-Readable Sentiment: The expansion of emoji and reactions (e.g., Facebook's "Like" variations) provides high-quality, labeled data for sentiment analysis, creating a "mood graph" that allows platforms to codify and analyze human emotion at scale.
• Emojidick Project: Fred Benenson crowdsourced a translation of Herman Melville's Moby Dick into emoji via Amazon Mechanical Turk to test the feasibility of emoji as a language, resulting in the Library of Congress acquiring the work as a cultural artifact despite the linguistic ambiguity.
• Future Outlook: Unicode acknowledges that adding 60 new emoji annually is unsustainable, signaling a potential future shift toward inline images and customizable stickers, though standardization remains critical for accessibility and cross-device communication.
• Upcoming Events: The first-ever "EmoCon" is planned for November in San Francisco, featuring panels, a film festival, and a hackathon to open discussions on emoji policy and culture beyond the closed meetings of the Unicode Consortium.