LLenny's Podcast
← All episodes
Sam Schillace (Microsoft deputy CTO, creator of Google Docs)11 January 2024

How to be more innovative

5Frameworks
15Insights

Frameworks in this episode

Self-Mastery4 steps

Guilty-Pleasure Career Bet

Find the thing you feel guilty getting paid for, do the hell out of it with intensity, and take non-linear risks.

Read →
Innovation3 steps

North Star Exploration

Turn aimless tech-tinkering into learning by chasing an arbitrary-but-real goal and being stubborn about reaching it.

Read →
Innovation3 steps

The Love-Hate Disruption Test

Gauge whether an idea is truly disruptive by how polarized the reactions are, not how positive.

Read →
Strategy4 steps

User-Value Thermodynamics

Users are lazy; adoption only happens when total value delivered exceeds total effort spent, ideally by 10x.

Read →
Innovation4 steps

What-If Before Why-Not

Evaluate a disruptive idea by imagining its upside first, then treat the objections as your build list.

Read →

Insights & moments

The myth-busts, hot takes, explainers, and tools worth keeping.

Hot Take· 5

Hot Take09:00

Better to Be Optimistic and Wrong Than Pessimistic and Right

Sam argues engineers are fundamentally pessimistic — they see everything as a problem to be solved — but he's concluded he has lost far more by being pessimistic than by being optimistic too early. His mantra is that there's very little prize for being pessimistic and right, especially in a moment of rapid change. Optimism, for him, is a deliberate choice rather than a personality trait.

  • Engineers 'come into the world broken,' seeing everything as a problem
  • Sam has missed out more from pessimism than from early optimism
  • There's almost no reward for being 'pessimistic and right'
  • Optimism is a conscious habit, not his natural disposition

“it's much better to be optimistic and wrong than pessimistic and right”

Sam Schillace · 09:30
#optimism#mindset#innovation
Hot Take22:00

Why Crypto Never Passed Sam's 'What If It Works' Test

Sam spent a lot of time genuinely stress-testing his own skepticism about crypto, checking whether he was just reacting to something that threatened his identity. He never found a 'what if this works' scenario that produced real value — even in the best case, the user outcomes felt dystopian rather than desirable. He concluded the pessimism was justified.

  • Sam deliberately checked whether he was reacting from threatened identity
  • His test: 'what if it works — how valuable is it?' Crypto failed even best-case
  • He couldn't name a single thing valuable to him as a user
  • Conclusion: sometimes the pessimism really is justified

“what if it works like then I have to run upsc on my personal finances that sounds dystopian I don't want that”

Sam Schillace · 22:30
#crypto#skepticism#user-value
Hot Take55:30

Do the Thing You'd Feel Guilty Getting Paid For

Sam pushes back on the belief that work has to be unpleasant to be valuable — we systematically undervalue the things we're good at. His advice: find the thing you'd feel guilty getting paid for and do the hell out of it, as hard as you can. The most impact you'll ever have comes from that flow state where you can't quite believe you're getting away with it.

  • We undervalue what's easy and fun, assuming valuable work must be unpleasant
  • Find the thing you feel guilty to get paid for and pursue it at full intensity
  • That work doesn't have to be hard — the best case is that it isn't
  • Peak impact comes from the flow state of 'getting away with' doing what you love

“you should go do the thing that you feel guilty to get paid for if there's a thing like that and do the hell out…”

Sam Schillace · 55:30

“the most impact you'll ever have is where you're in that mode where you're just like in the flow”

Sam Schillace · 56:00
#career#advice#flow
Hot Take1:02:30

AI Isn't a Feature of Your Product — Your Product Is a Feature of AI

Sam argues you can add some value by bolting AI onto an existing product, but the transformative value comes from apps and solutions that wouldn't work at all without it — products that treat AI as a true platform. He inverts the usual framing: rather than AI being one feature among many, the product itself becomes downstream of AI's capabilities.

  • Bolting AI onto an existing product adds only marginal value
  • Massive value comes from products that can't exist without AI
  • Treat AI as a true platform, not a feature
  • The framing inverts: your product becomes a feature of AI

“AI isn't a feature of your product your product is a feature of AI”

Sam Schillace · 1:02:30

“really transformative massive value will come from building apps and solutions that won't work at all without it”

Sam Schillace · 1:03:00
#ai#product#platform
Hot Take1:04:30

Just as the Internet Made Distribution Free, AI Will Make Pixels Free

Sam sees AI driving a giant disruption in software analogous to how the internet made distributing information free. Producing pixels — UI, images, software output — is currently expensive, requiring programmers and infrastructure, and many businesses are predicated on that cost. AI will collapse it, the way image generation already replaced the Photoshop-skill pipeline with a plain-language request.

  • Internet made distribution of information free; AI will make pixels free
  • Producing pixels today is expensive — programmers plus infrastructure
  • Many businesses are predicated on pixels being costly to produce
  • Image generation already shows the shift — 'a cat riding a bike' vs learning Photoshop

“I think the way that the internet made distribution of information free I think AI is going to make pixels free”

Sam Schillace · 1:04:30
#ai#disruption#software

Explainer· 3

Explainer11:30

A Polarized Love/Hate Reaction Is the Signal of True Disruption

Sam explains that a product's impact shows up in the shape of the reaction. A bell curve of indifference and mild dislike means an incremental product. But when a small group loves something intensely and others want it to 'die in a fire,' that polarization signals genuine disruption — as with early Google Docs and later ChatGPT. He notes the haters can even outnumber the fans for a long time and it doesn't matter.

  • Modern indifference and mild dislike = incremental, non-disruptive product
  • Intense love plus intense hatred = a truly impactful, disruptive product
  • ChatGPT ('amazing' vs 'terrible') is a clear example of the pattern
  • Early Google Docs had haters outnumbering fans for a long time — that's fine

“the entire world is like this is amazing or this is terrible and there's like not a whole lot in between”

Sam Schillace · 12:00

“for a long time the haters outnumbered the people who were fans”

Sam Schillace · 12:30
#disruption#product#signals
Explainer25:00

Adoption Follows 'Thermodynamic' Rules — and Friction Is Hidden Cost

Sam frames adoption almost like physics: add a little value and adoption creeps; add a lot and it explodes (ChatGPT). But builders forget the whole system around their feature — the user has to hear about it, remember it, learn it, build a habit, and get past sign-up friction. To win, the total ease gained has to exceed the energy spent, ideally by 10x. On Writely they cut friction so hard they didn't even ask for an email at first.

  • Small value = slow adoption; large value = explosive adoption
  • Builders overweight their feature and ignore the surrounding effort cost
  • Hearing, remembering, learning, habit-building, and sign-up are all friction
  • Net value must exceed effort — aim for ~10x better than what it costs to use
  • Writely deferred asking for an email until ~2 minutes of use to kill friction

“when we did uh rightly at the beginning we didn't even ask for an email address because it was such a novel thing we didn't…”

Sam Schillace · 26:30

“it has to make your life a lot better hopefully a really a lot better like 10x better than what you spend to use it”

Sam Schillace · 27:30
#adoption#friction#product#onboarding
Explainer49:30

Why Sam Thinks the Younger Generation Is More Pessimistic

Sam has puzzled over why students today seem less energetic about solving problems than earlier generations. He traces it to a reduced willingness to take risks and fail: filtered content presents rare five- and six-sigma successes as normal, living 'out loud' makes failure feel more painful, elite kids grow up highly curated with everything needing a reason and output, and children stopped being allowed to play unsupervised. The through-line is that you can't succeed if you're afraid to embarrass yourself.

  • Filtered content presents rare extreme successes as the norm
  • Public/'living out loud' life makes failure feel more painful
  • Elite kids are highly curated — everything must have a reason and output
  • Loss of unsupervised childhood play reduced tolerance for experimenting to failure
  • Core skill: being willing to make a fool of yourself and fail

“you see a lot of filtered content and that filtered content presents low probability events like five and Six Sigma events as though they were…”

Sam Schillace · 50:30

“you cannot succeed if you're afraid to fail”

Sam Schillace · 52:30
#risk#failure#generations#mindset

Story· 3

Story04:00

The Very First Google Doc From 2005 Still Opens Today

Sam still has the original document from before Google Docs existed — a 2005 test file written in C on rented Windows servers in Texas, later ported to Java on Bigtable, then Spanner, with the front end rewritten twice. Despite all those back-end and front-end migrations, no user data was ever lost. The document is just him messaging a colleague about whether collaborative typing was colliding, plus a pasted image of Edna from The Incredibles.

  • Started in 2005 in C on three rented Windows file servers in a Texas data center
  • Migrated to Java on Bigtable, then Spanner, front end rewritten twice — nothing lost
  • The content is just a note to Steve testing whether collaboration was colliding
  • The Computer History Museum wanted to curate it but the live doc can't really be preserved

“we moved all the data over in big table and we we didn't lose anything like we never lost anybody's stuff”

Sam Schillace · 04:30

“it is the old Google doc from like October of 2005 or something like that”

Sam Schillace · 05:00
#google-docs#history#migration
Story06:00

They Moved All of Google Docs to Google's Infrastructure and Nobody Noticed

When Writely migrated onto Google's infrastructure, the team put the site into read-only 'maintenance mode' for eight hours on a Sunday, moved all the data, and brought the new system up. Days later Sergey asked when they'd migrate — and Sam had to tell him it was already done. The only person who spotted it was a blogger in Germany who noticed the IP address had changed.

  • Migration done quietly in an 8-hour Sunday read-only window
  • Sergey Brin asked when they'd move over — days after it had already happened
  • The only detection was a German blogger noticing an IP address change
  • Sam values doing high-stakes engineering changes invisibly to users

“oh yeah we did it this weekend no one noticed some blogger in Germany like noticed the IP address changed and that was it like…”

Sam Schillace · 06:30
#google#migration#engineering
Story28:00

Google Docs' Innovator's Dilemma: 'We Have All the Users, They Have All the Money'

Sam describes deliberately stripping features to make Google Docs zero-install, easy, and collaborative against an Office incumbent that was asymptotically adding low-value features. They even cut too much — students demanded word count early. It was a textbook innovator's dilemma: a small product chipping away from the bottom of a market the incumbent didn't understand. Microsoft survived it well; Google took the users, Microsoft kept the money.

  • Deliberate tradeoff: zero-install and collaboration in exchange for far fewer features
  • They cut too far — word count was an early demand from student users
  • Classic innovator's dilemma: chipping from the bottom of a neglected market
  • Outcome: Google took market share, Microsoft kept the revenue

“we took a bunch of market share but they kept all the money basically”

Sam Schillace · 30:30
#innovators-dilemma#google-docs#microsoft#strategy

Takeaway· 4

Takeaway10:30

When People Call It a 'Toy,' You've Probably Built Something Real

Sam says every new idea looks dumb at first — but so do genuinely dumb ideas, so it's an imperfect signal. The tell he listens for is dismissive language, especially the word 'toy.' If people can't muster a better criticism than 'it's just a toy,' that's often a sign the thing is actually real and threatening.

  • Every new idea looks dumb at first — but so do actually dumb ideas
  • The more disruptive an idea, the dumber it feels initially
  • Listen for dismissals like 'it's a toy,' 'it's stupid,' 'I don't get it'
  • 'Toy' is a strong signifier of something real people can't otherwise argue against

“if you hear people saying something's a toy that's often a really good signifier that it's actually something real and threatening”

Sam Schillace · 10:30
#innovation#disruption#signals
Takeaway22:30

Users Are Lazy — Only Real User Value Wins

Sam's root principle is that everything comes down to significant user value. People are lazy and cynical at heart; nobody adopts something because it's friendly, nice, or has a pretty logo. If you can't point at real user value, no amount of marketing dollars or articles will save the product.

  • It's all about user value — everything else is noise
  • People are lazy and only care whether something makes their life better
  • Marketing spend and articles can't rescue a product with no real value
  • Convenience wins because users won't expend effort for weak payoff

“users are lazy right we're all lazy we don't really care that much at the end of the day”

Sam Schillace · 22:45
#user-value#product#adoption
Takeaway53:00

Sam's Personal Motto: 'From Error Comes Virtue'

Sam describes himself as a maker with poor motor skills who makes mistakes constantly — and then figures out how to turn each mistake into something good. He adopted 'virtue from error' as his personal motto this year, the first he's ever had, and says the more he repeats it the more he likes it. It captures a whole disposition toward creating from your own mistakes.

  • Sam constantly makes mistakes as a maker and turns them into virtue
  • 'Virtue from error' is his first-ever personal motto, adopted this year
  • He frames mistakes as raw material to be creative with, not failures to hide
  • The more he repeats it, the more it resonates

“I have this personal motto which is like from error comes virtue”

Sam Schillace · 53:00
#motto#failure#creativity
Takeaway1:09:30

'Just Play With It' Is Bad Advice — Pick a Concrete Goal Instead

Sam disagrees with the common advice to just play with AI. The best way to learn a new technology is to pick a concrete thing to build with it — even a goofy or arbitrary goal — and be a little stubborn about reaching it. 'Mess around for an hour' isn't a goal; 'build a GPT that can do this part of my job' is. Aim goals somewhere at least slightly interesting.

  • 'Just play with it' isn't really useful advice for learning AI
  • Pick a concrete goal — even a goofy one — and be stubborn about reaching it
  • An arbitrary but real goal beats aimless messing around
  • Choose goals that lead somewhere at least a little interesting

“just play with it is not really great advice”

Sam Schillace · 1:09:30

“picking in some arbitrary goal and being a little bit stubborn about trying to get yourself to it is a good way to learn stuff”

Sam Schillace · 1:10:00
#ai#learning#advice