AI summary by TwinMind
Emojli: Behind the Scenes and Why You Should Never Build An App
Tom Scott
30m 58s
App Origin Story
- Matt and Tom at Westfield Stratford shopping center outside Jamie's Italian
- Tom suggests emoji-only messenger idea during casual conversation
- Both had independently thought of concept, surprised no one else had done it
- Unicode had just announced new emoji releases (Spock hand, reversed victory sign)
- Yo app had recently launched, social networks existed for everything
- Reserved domain em.li on phone using tube Wi-Fi on overground journey home
- Initially thought it would be "really easy build" - instant messenger with minimal backend/frontend
- Estimated completion in one week
Early Development Challenges
- Week later, hadn't built anything
- Created simple registration form: email address + emoji username
- Sent confirmation emails saying "thanks, we'll get to you soon"
- Expected no attention, planned to call it joke if ignored
- Released explanatory video as usual Tom/Matt content
Explosive Growth
- 70,000 usernames reserved within couple days
- Unlike normal social media, no one bitter about unavailable usernames
- Emoji usernames created unique problem: can't type your name
- Creative solutions emerged:
- Adam: used blood type symbols (A, B, O) plus chart symbols
- "Boobs" possible using dollar sign as S
- Three boobs, cool boobs variations
- No actual offensive words successfully created
Support Nightmare
- Hundreds of emails first day
- Common complaints:
- "Tried to register smiley face, not working"
- Only 500 old emoji available, need multiple characters
- Dozens asking for taken usernames
- At least 5 emails asking for non-emoji usernames
- Turned into technical support job for spare-time project
- Had to register under Data Protection Act (cost £35)
- Privacy policy required listing possible stored data including sexuality details (due to emoji like two men holding hands)
Technical Implementation
Backend
- PHP and MySQL (audience booed)
- Matt: professional web developer for "us versus them"
- Unicode handling: base64 encoding to prevent munging
- Tom: broadcast engineer, doesn't normally program
Frontend
- HTML, CSS, JavaScript (not native app)
- Cordova/PhoneGap for app packaging
- Chose app over web due to browser emoji support issues:
- iOS: excellent emoji support
- Android: no support
- Windows Phone: runs Internet Explorer, no support
- Emoji treated as two characters in JavaScript
Launch Process
- Took month+ in spare time (not week as estimated)
- Apple approval: 11 minutes review time
- Submitted week beforehand, approved 10 minutes before midnight on launch date
- First time app maker, approved first submission
- Launched day before EMF talk
Media Attention and Startup Assumptions
- Time Magazine and other publications covered it
- Journalists assumed they were startup seeking investment
- "Two white middle class males" automatically perceived as "Tech Bros"
- Investment fund contacted them
- Sent 70,000 emails to confirmed users from EMF festival (some from phone)
- No adult rating needed despite usernames like "OBI no under 18s"
Competition and Copycats
- Two other emoji messengers launched during their development month
- One called "Emojli" (same name)
- Competitor registered domain 3 weeks after their launch
- Twitter handle was their name + extra letters
- Competitor claimed prior planning despite timeline evidence
- 21 five-star reviews in "pigeon English" (later disappeared)
- Getting support requests for competitor's app
- Filed complaint with Apple (ignored)
Technical Challenges Continued
Push Notifications
- Multiple failed attempts with garbled characters
- Required numerous Apple certificates:
- Server certificate
- User certificate
- App certificate
- Developer certificate
- Sandbox certificate (only worked on Tom's phone)
- Testing required physical proximity due to certificate setup
- Screenshot/message debugging process between team members
Cross-Platform Issues
- iOS 8 compatibility problems discovered post-launch
- iOS 6 support requests (tested, didn't work, not supported)
- Android version requested but not developed:
- Don't own Android device
- Already invested significant money
- Android push notifications more complex
- Code not directly portable
Current Status and Future Plans
- 70,000 registered users
- Works reliably for current user base
- 100+ tech support requests handled morning of talk
- 90% resolved by automatic vacation response
- No intention to continue significant development
- Would consider selling to interested parties
- Not interested in becoming startup founders
- Both have fulfilling day jobs they want to keep
Startup Culture Commentary
- Internet startup culture "gone beyond parody" since Yo received $1.2M funding
- Yo later "pivoted" to notification service
- App store review: "finally found something worse than Yo"
- Used negative review as marketing quote
- Discussed app bubble similarities to previous tech bubbles
Technical Constraints and Decisions
- No trademark (costs money, not investing more)
- Apple developer account and SSL certificates already cost money
- Hosting sponsored by Bytemark (also sponsors EMF Camp and us versus them)
- Using native UI to avoid emoji copyright issues
- GitHub uses Apple's emoji images, unclear legal status
- Font vs. image copyright questions unresolved
- Kickstarter for open emoji set failed at 10% completion
Privacy and Data Ethics
- Will not do natural language processing on conversation data
- Not Facebook or OKCupid - no user experiments
- Privacy policy allows internal record keeping but won't use
- Explicit commitment to user privacy
Q&A Highlights
- Android support: backend ready, frontend needs development and testing
- Investment offers: would consider selling, not interested in running startup
- Trademark: no, costs money, not investing more
- User-generated emoji drawings: rejected due to moderation burden
- Taking investor money: prefer current careers, don't want startup life
