AI summary by TwinMind

Twitter vs. X: Product Lessons For Startup Founders

Y Combinator

22m 42s

Introduction to YC Product Analysis Series

  • New YC series analyzing popular consumer products
  • Tom and host have extensive consumer product building experience
  • Focus on Twitter/X as first product deep dive
  • Perfect timing given recent changes under new ownership

Twitter's Major Feed Algorithm Changes

  • Biggest change: shift from chronological feed of followed topics to TikTok-style algorithmic engagement feed
  • Host's experience: used to see left-leaning content, now sees "MAGA rally after MAGA rally"
  • Content quietly morphed without obvious UI changes
  • More random videos: fights in Mexico City, people getting smacked, cat videos
  • Increased dopamine hits but decreased learning/educational value
  • Elon explicitly stated users should "feel equally bad if you're on far left or far right"

Product Metrics vs User Value Problem

  • Teams optimize for single metrics (dwell time, engagement, minutes used) without considering quality
  • Example: watching fist fight videos increases engagement but doesn't provide actual user value
  • Business model based on advertisers selling seconds of video time incentivizes more hours on site
  • Contrast with Google: aims to get users off site quickly with good first search result
  • Counter-argument: poor long-term retention if users highly engaged with content they don't actually like
  • Quality users eventually churn to other platforms

User Control and Product Design Issues

  • Users can click "not interested" on three-dot menu to improve feed
  • Many users don't know about this feature
  • Reddit threads exist teaching users "how to stop Twitter from showing random content"
  • When users need instructional videos to use your product, something's wrong
  • Product gives many tools but unclear which to use (mute, block, not interested)

Social Network Evolution Pattern

Facebook

  • Started with personal walls, no newsfeed
  • Biggest change: introduced newsfeed showing content from people you chose to follow
  • Initial user protests but all metrics went up
  • Later diluted as "mom joined, sister joined" - broader social group
  • Feed became less relevant to core users

Instagram

  • New network with different content type and network
  • Initially early adopters and people you wanted to follow
  • Gradually everyone joined, brands joined
  • Drive for engagement led to more unwanted content in feed
  • Now "approximately zero photos of friends, just random clickbait"

TikTok

  • "Skipped straight to the end game"
  • Shows all possible content, lets algorithms determine what human brain wants
  • "Deeply unsatisfying experience, empty calories"

Product Builder Lessons

  • Define what problem you're solving: boredom vs real human connection
  • Human communities based on Dunbar's number (~150 people)
  • Social network constrained to smaller groups might be interesting
  • Different monetization needed since fewer hours = less ad revenue
  • Pressure ultimately pushes toward algorithmic feed

Twitter/X Specific Improvements Needed

  • Algorithmic feed sorting user-selected content makes sense
  • Problem: introduction of high "entropy content" (fist fights, car crashes)
  • Need more judicious content selection and clearer user feedback mechanisms
  • Curated lists feature exists but hard to find
  • Lists are advertising-free, high-quality (Ukraine war, UK politics examples)
  • Inheriting credibility of others' curation work

Twitter's Strengths

  • Still best for breaking news events (assassination attempts, Ukraine war, UK politics)
  • Easy for first-party information owners to post content quickly
  • Community Notes effective at distinguishing fake vs real news
  • Firsthand videos during events that traditional media can't produce quickly

Blue Checkmark Changes

  • Originally signal of credibility and legitimacy
  • Now anyone can pay $7-9/month for blue check
  • Overloaded the credibility signal
  • Can't distinguish between "12-year-old in basement" vs legitimate person
  • Intended to filter out bots through payment requirement
  • Edit tweets feature implemented successfully (was controversial but works fine)

Product Feature Bloat

  • Sidebar filled with unused icons/features
  • Grok (ChatGPT inside Twitter) doesn't make sense to most users
  • Example of product expanding beyond user perception
  • Lack of clear articulation of what product is for and how to use it
  • Deviation from core product purpose

Elon Musk's Product Taste

  • Jeff Bezos quote: "taste is not transferable across domains"
  • Host drives Tesla, considers it "best car ever sat in"
  • Tesla taste/decisions don't match Twitter/X decisions
  • SpaceX achievements (booster catching) are incredible
  • Mass market consumer product decisions "not super well thought out"

Name Change from Twitter to X

  • Neither host calls it "X" unironically
  • Twitter "got the verb" - "to tweet"
  • "When your product gets the verb, you've won"
  • Examples: Google, Bump (14 years later people still say "bumping" phones)
  • Monzo originally "Mondo" - sued by German company, users suggested new name
  • X change based on Elon's obsession with letter X
  • Possible grand plan for "everything app" including payments platform
  • Should show future vision before changing name

Naming Advice for Founders

  • Best time to change name is early rather than later
  • Pick something people can easily say and spell
  • Should be googleable without asking for spelling
  • Ideally relates to product function
  • Best companies bring meaning to name rather than vice versa
  • Examples: Monzo, Google, Amazon didn't have inherent meaning

Revenue Impact

  • Ad revenue down significantly due to advertiser concerns about content adjacency
  • Subscription revenue "probably okay but not taking over"
  • Engagement metrics "off the chart"
  • Classic problem: team optimizing single metric while everything else suffers
  • "AI tasked with making paperclips enslaves world" analogy

Product Leadership Lessons

  • Top-down ownership important for product vision
  • Example: Google Photos designed for how designer wanted to use it
  • Consistent worldview of product purpose needed
  • Elon may have vision but "we're not ready for it"

Final Product Founder Advice

  • Articulate what product is for and who should use it
  • Define good user states vs states that look good but are actually bad
  • Not always a single metric to optimize for
  • Example: Twitter engagement up but user feels "that was waste of time, wish I'd been playing with kids"
  • Don't mindlessly optimize single metric excluding all other factors
  • Leads to "engagement farms"
  • Product leader role: provide consistent vision and clear purpose
  • Founder CEOs have "moral authority" - "I started this company, this is what I believe"
  • Steve Jobs iPhone example: clear decisions on what to include/exclude, no internal debate

TwinMind

Get summaries like this for your own meetings

TwinMind takes notes in the background during meetings, lectures, and calls, then writes the summary for you.

Free plan to get started

No bot joins your calls

Phone, desktop & Chrome apps

Your next meeting, summarized like this

TwinMind listens in the background and turns meetings, lectures, and calls into searchable notes you can chat with. Start free.

Available on phone, desktop & Chrome

Your next meeting, summarized like this

TwinMind listens in the background and turns meetings, lectures, and calls into searchable notes you can chat with. Start free.

Available on phone, desktop & Chrome

Your next meeting, summarized like this

TwinMind listens in the background and turns meetings, lectures, and calls into searchable notes you can chat with. Start free.

Available on phone, desktop & Chrome