AI summary by TwinMind
What Does It REALLY Mean To Do Things That Don't Scale? – Dalton Caldwell and Michael Seibel
Y Combinator
18m 51s
Introduction & Core Concept
- Michael Cybel with Dalton Caldwell discussing "do things that don't scale"
- Term popularized by PG essay from 2013, advice given to YC founders for years
- Founders often confused about concept
- Definition: doing something provocatively manual where founder personally does tasks, not relying on code or scalable processes
Reddit Example - Fake Users Strategy
- Day one: Reddit existed as blank site with no links, no users
- Scalable approach would be: launch subreddits → run ads → hire influencers → big launch with hundreds of thousands signups
- What Steve & Alexis actually did:
- Built tool allowing them to submit links with extra username field
- Started inventing different usernames and submitting links themselves
- Day one when emailed to 6 friends, looked like page full of users (actually only 2 users)
- Steve's milestone stories:
- First day someone else submitted (not day one)
- Got drunk, forgot to load morning's fake links, feared blank site
- Opened site next morning - enough real users submitting that site wasn't blank
- Estimated 20-30 real people at that point
DoorDash Example - Founder Deliveries
- Food delivery requires hundreds of orders daily at scale
- Founders got started by doing all deliveries themselves
- No need to hire drivers, build driver app, or recruit drivers
- When miracle order came in, founders would drop everything and deliver food personally
- Benefits of founder delivery approach:
- Delight users through personal service
- Race to restaurant, ensure correct order, add extras
- Personal thank you from founder
- No driver network management needed
- Direct user feedback: "here's your burrito, I'm the founder, do you like it, would you order again?"
- Tony (DoorDash founder) reportedly still does deliveries occasionally
Cruise Example - Basic Autonomous Vehicle
- Cruise recently allowed general public to ride in driverless cars
- When Cruise was in YC, V1 built by Kyle's team in 3 months
- Retrofitted Audi with basic features:
- Big red button on driver's seat floor (purpose unexplained but clearly important)
- Only worked on highway
- Basically adaptive cruise control that stayed in lane
- Not more sophisticated than that
- Memorable test ride moment: Kyle said "oh a shadow, let's see how car handles that"
- Google had cars with fancy equipment while Kyle had Audi that couldn't handle shadows
- That's how it starts
CEO Role in Early Stages
- Job of CEO/co-founders: do shittiest, worst, low-status work possible
- Opposite of typical CEO mental model (visionary, powerful)
- Must learn to love and embrace the work
- Pre-product market fit job: make everything happen personally
- Be personally accountable for entire system running end-to-end
- Many founders mistakenly emulate late-stage CEOs (Elon Musk, Mark Zuckerberg)
- Smart people often not cut out for founder role because they don't want to do manual work
- Special type: smart but willing to get hands dirty
Customer Support & Front Lines
- Founder's actual job: answering customer support emails
- Dealing with angry customers who hate them
- No one else to yell at in tiny startup
- Every DoorDash order screw-up meant founders personally apologizing
- Founders want to put others on front lines but founders ARE the front lines
- More scalable to have others handle, but "doing things that don't scale" means embracing the suck
Learning Advantages
- Learn what others unwilling to learn → build better product
- Serve customers better than incumbents
- Senior management at big companies not on front lines
- They use researchers/outsourced customer surveys
- Front-line founders learn 10x more
Social Pressure & Friends
- Friends at Google with free food, laundry, bus mock founder work
- "You're hand delivering food? Hand adding links to your website?"
- Friends think founder is idiot for starting company
- This stage screens out most people because it sucks
- If easy to start company, more people would do it
- Few Googles exist because this stage is barrier
- Not lack of good ideas, innovation, or funding - lack of people willing to do this stage
Big Tech Company Training Problem
- Many young founders think they need big company experience first
- Google/Facebook interviews focus on scalable algorithms
- Inside big tech: rolling out to millions, scalability beaten into head
- Only scalable solutions considered good solutions
- Unscalable feature requiring 100x more servers would be rejected
- Every product/feature viewed through scalability lens
- This training actually harmful for early startup stages
- Helpful for late stage but not early stage
- College grads without big tech experience never told to worry about scalability
- Big tech founders can't launch because nothing's perfect
- Incentive structure at big companies: don't launch unscalable features
Founder Mindset & Expectations
- Biggest challenge: screwed up expectations
- Founders think "I launch it and then it's gonna work"
- Should go in with mindset: "I don't care if it takes two years to work"
- Maintaining motivation is 3/4 of the game
- What founders want to discuss vs. what they need help with are different
- Should ask about: emotional well-being, maintaining motivation, high stress, criticism
- Many unprepared for internet criticism
- Many startups die because founders run out of gas, get sad, want to return to Google jobs
Motivation & Reality Check
- Motivation not about how shitty something is, but expectations vs. reality
- Dalton tells college kids: "this life sucks, like getting punched in face every day"
- Real face punch, not fake - memorable for rest of life
- Still have bad dreams about startups
- Hardest thing they've ever done and probably will ever do
Key Principles Summary
- Expect it to be counterintuitive
- People around you won't understand, will actively think you're stupid
- Friends will stage interventions
- Expect to look way different than larger competitors
- You are primary actor, not acting through others - you are the doer
- Don't copy well-funded competitors' strategies
- Focus on getting one customer before trying for thousands
- Delivering one burrito separates you from the pack
- Actually doing something puts you in top half of startup founders
- Get first customer to come back - measure retention
