如何更具创新力 | Sam Schillace(微软副CTO,Google Docs 创造者)
How to be more innovative | Sam Schillace (Microsoft deputy CTO, creator of Google Docs)
Sam Schillace: We tend to undervalue the things we’re good at. We tend to think work has to be unpleasant. And so if something is easy and fun, we don’t tend to think it’s valuable. So I think lots of people gravitate in this direction of like, let’s go do unpleasant things and grind our way through the career because that’s the way to make it. But the reality is 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 of it, right? Do it as hard as you can. If you get pleasure from doing something that people want to pay you for, do it the best you can do it, as hard as you can do it. And if that’s messing around and playing around with cool ideas, do the hell out of that. Work doesn’t necessarily have to be hard.
Introducing the Guest
Lenny: Today my guest is Sam Schillace. Sam has an incredible resume that is very hard to summarize succinctly. I’ll give it a shot. Currently, he is corporate vice president and deputy chief technology officer at Microsoft, where he leads efforts in the consumer product space, infrastructure, and AI. Sam is most known for basically inventing Google Docs with his company Writely, which was acquired by Google, and became the foundation for what is now Google Workspace, which currently has over one billion active users a month.
After joining Google, Sam ended up responsible for many of Google’s consumer applications, including parts of Gmail, Maps, Automotive, Groups, Reader, and more. He’s also founded six startups, was senior vice president of engineering at Box through their IPO. He’s also worked at Intuit, Macromedia. He was even a VC at Google Ventures for a time. As you’d suspect, we had a fairly wide-ranging conversation, but the core focus was around innovation, how to think big, how to come up with original ideas, why optimism is so important and powerful, and also a ton of career advice. Sam is hilarious and not what I imagined a corporate vice president at Microsoft would be like, which gives me even more respect for Microsoft. A big thank you to Brett Berson for making this introduction. With that, I bring you Sam Schillace, after a short word from our sponsors.
Over 5,000 fast-growing companies use Vanta to automate up to 90% of the work involved with SOC 2 and these other frameworks. For a limited time, Lenny’s Podcast listeners get $1,000 off Vanta. Go to vanta.com/lenny. That’s V-A-N-T-A dot com/lenny to learn more and to claim your discounts. Get started today. Sam, thank you so much for being here, and welcome to the podcast.
Sam Schillace: Thank you. Happy to be here.
The First Google Doc
Lenny: A really fun fact about you is that apparently you have the very first Google Doc file. I don’t know what you call it. The very first Google Doc document saved somewhere from before even Google Docs was a thing. And does it still work in today’s Google Docs? And what is in this document?
Migrating to Google Infrastructure
Sam Schillace: Yeah, it does actually still work. It’s pretty funny. Actually, if I move my camera for a second you might, if you’re on YouTube, you can see the Writely thing in the background. Writely was the company that did Google Docs. Yeah, it still works. And it’s funny though because it’s like the document of DCS, right? So we started off 2005, wrote this thing in C#, which is a little known, in our own… It was pre-cloud, so we had three file servers that we rented that were Windows machines in a data center in Texas with a sysadmin in the Philippines running them. So it started there.
And then when we moved to Google, we ported everything to Java, we moved all the data over into Bigtable. And we didn’t lose anything. We never lost anybody’s stuff. So it’s still there and moved across. And then so that’s one backend migration, and there’s another one with Spanner. And then the front end has been rewritten twice as well. So it’s like, is it really the same document? I don’t know. The front end, back end have been rewritten. It’s not much. It’s just me saying something to Steve about, is collaboration working? Are we colliding on each other? Because we were trying to figure out typing on one line, if that algorithm was working.
And then there’s a picture of Edna from The Incredibles pasted into it. I don’t know why. I think that came after, so I might’ve gone back and pasted that in. I’m not sure when. We must’ve been testing pictures or something. Unfortunately, we don’t have the version history anymore, so I don’t know what was original, original. But it is the oldest Google Doc from October 2005 or something like that.
Understanding Disruptive Innovation
Lenny: I love that this philosophical answer of, is it still the same Google Doc, considering all the code has been redone.
Sam Schillace: Well, the Computer History Museum wants to curate it. I talked to these guys and they were like, “Oh, that’s so cool. We’ll take that.” I’m like, “How?” I can make you a PDF. Now it’s not the document. I can share you into it, but please don’t edit it. How do you curate this? So that’s funny.
”What If We Could?”
Lenny: Yeah. It needs to be on the blockchain.
Sam Schillace: Yeah. Yeah. If you made the NFT of it, that would be more authentic, I think, than the document almost. It’s kind of funny.
Polarizing Reactions Signal True Disruption
Lenny: That’s amazing. And it’s amazing that it still works. That’s a testament to, I don’t know, you/Google.
Sam Schillace: Well, I’ll tell you a quick Google story. When we migrated in to Google, we were very sneaky about it, and we put the site into “maintenance mode” for eight hours on a Sunday, where everything was just read-only. And then we migrated all the data and moved everything and brought the new system up. And three days after that, Sergei was in a meeting with me and he’s like, “So when are you guys going to move over to Google infrastructure?” And I got to tell them, “Oh yeah, we did it this weekend.” No one noticed. Some blogger in Germany noticed that the IP address changed and that was it. Nobody noticed it at all, so we were really good about it.
Optimism Is a Choice
Lenny: Man, I love these sneaky stories. I’m hoping we hear more. There’s a bunch of stuff I want to cover, the first is this broad idea of disruptive innovation. I know that you spend a lot of time thinking about this. Google Docs is a great example of this. It feels like Microsoft increasingly is getting really good at this. Just the idea of doing something completely new, oftentimes things that people didn’t think were possible. So let me just ask a broad question. Why is this important to you? Why do you spend a lot of time thinking about this? And then just what are some tools you’ve found to help you and other people think more innovatively, more originally?
Lowering the Cost of Failure
Sam Schillace: It’s an interesting question. The why it’s important part, I don’t know, it just is. Everything you’re wearing, eating, using, listening to, sitting on, was a disruptive innovation at some point. That’s how everything happens, right? I think there’s this really interesting thing where everything new is threatening at some level at the beginning. I mean, probably literally the first guy who invented chairs got shit from his tribe mates for making a chair. And they’re all obvious in retrospect, right? Everything is obvious in retrospect.
But I think there’s this really deep thing that people have where if something is disruptive of your worldview, it feels threatening, and you have this very stark choice to make that’s either you’re wrong or it’s wrong. And humans are storytellers. It’s very easy for us to tell stories about why something is right or wrong if we’re motivated to. And so I call these why-not questions. People ask these why-not questions a lot. So a new thing pops up, and if you’re not ready to receive it for some reason, you’re not already half there or you don’t have a problem that it solves or whatever, it’s just threatening and irritating, and you come up with a why-not question.
We heard a bunch of these with Google Docs, with Writely, in the early days about the browser wasn’t ready, people wouldn’t… The whole model of the cloud was like, people aren’t going to trust you to store your files. That’s really weird. What if there’s no connectivity? I heard the no connectivity on an airplane story 100 times from journalists. Like, “What if I’m on an airplane and I write stuff?” I’m like, “I don’t know. There’ll be connectivity on airplanes soon.” Which there is. And those are all just why-not questions.
I think the more interesting ones are the what-if questions, like, what if this does work? Just use your imagination. Think about, how far can I extend the curve? What are the implications of that? I’m an engineer, and engineers are fundamentally pessimistic people. Somebody once told me engineers come into the world broken. They just look at everything as a problem to be solved. And I think there’s something to that. But I feel like I’ve missed out more by being pessimistic than I have by being too optimistic too early.
So I have this kind of mantra now that there’s just not that much of a prize for being pessimistic and right, particularly in a moment like this. It’s much better to be optimistic and wrong than pessimistic and right, I think. So I don’t know. And I’m an impatient person. I’m a creative person. I’m a messy person. I just like to create and explore and find stuff. So disruptive innovation just seems natural to me. But I think it’s not an exaggeration to say, literally, that wheat you had in your bread this morning, if you eat bread, some weirdo was messing around with plants 1,000 years ago and everybody thought he was a nut, or she was a nut, and then we had wheat because somebody… Just everything, right? Everything is like that.
Just Try It Out
Lenny: Along the same lines, I was actually just working on a post around first-principles thinking. And I found this quote from Steve Jobs just reminding us that everything around us was designed by some person that wasn’t necessarily that much smarter than you. And there’s no reason there isn’t a better way. It just happens to be the way it is today.
Sam Schillace: Yeah. One of the other ones that I like to keep in mind is every new idea looks dumb at first. Unfortunately, the dumb ideas also look dumb at first. It’s not a perfect [inaudible 00:10:35]. But the more disruptive they are, the more dumb you’re going to feel they are. You always listen for stuff like if they say it’s a toy or if it’s practical or it’s stupid or I don’t get it or whatever, those are often… Toy is a good keyword. If you hear people saying something’s a toy, that’s often a really good signifier that it’s actually something real and threatening, and they can’t think of a better criticism for it than it’s just a toy right now.
Real Opportunities vs. False Hype
Lenny: Yeah, I imagine people thought about Google Docs that way initially. It was like, “Oh, this little toy in the browser.”
Sam Schillace: Yeah, we got all this stuff. I mean, the real interesting thing, like I said at the beginning, you have this very binary reaction that’s possible, right? You either understand it, in which case, you’re super excited about it like, “Cool, the world’s going to change in this exciting way.” Or you don’t and you reject it. And to the degree that something is really disruptive, that reaction, that binary reaction gets really strong. And so with something like G Docs, we got this thing with G Docs that was really confusing in the early days to me, where there was a small group of people that really liked it. Some of them liked it more than we liked it. Nate Torkington over at Writely was this super huge early booster for it. And I did not understand what he saw in it at first. But then we had people that just wanted it to die in a fire.
And that bifurcation of love it, hate it, is really how you have an idea of whether you have impact in what you’re building. If you get more of the bell curve of modern indifference and maybe mild like and mild dislike, that’s an’s an incremental product. That’s not really disrupting anything. But if you look at something like ChatGPT where the entire world is like, “This is amazing.” Or, “This is terrible.” And there’s not a whole lot in between, that’s a very good signifier of it being truly impactful and disruptive. Whether it’s actually good or bad is a separate question. But there’s no denying that that’s a disruptive technology.
Convenience Always Wins
Lenny: That’s an awesome framework. Basically, if it feels like people sort of like it, mostly people don’t care, very few people love it or hate it, probably not disruptive. If some people absolutely love it and a lot of people really hate it, good sign.
Sam Schillace: Right. Yeah. Actually, weirdly enough. It’s not voting, right? In the early days, we had, I don’t know, a couple of million users, five million users, and there were still executives at Google telling me that it was a stupid idea and that it should stop and we shouldn’t be doing it. So, for a long time, the haters outnumbered the people who were fans. And who cares? Whatever. It’s fine. As long as you don’t run out of the people who love it, that’s fine.
Writely’s Zero-Friction Onboarding
Lenny: Is there another example of you using this what-if approach either on a product you worked on or something you’ve seen and it working out?
From Challenger to Incumbent
Sam Schillace: I’m doing a lot of it right now, honestly. I mean, that’s probably the most immediate example. But I could almost point at any product and there’s moments like that in there. But right now there’s a lot of why-not stories, right, around generative AI. So it’s expensive, it hallucinates, you can’t necessarily… It’s sarcastic, it’s random, it doesn’t do the same thing twice. Yeah, they’re real, they’re actual issues to solve, but I look at it and think, “Well, what if? What if we can build software around it? What if we can build more complicated programs than what we’ve been able to build? What if we actually have a reasoning engine that we can use to do meaningful things? What if this is really the second Industrial Revolution where, in the first one, we had a surplus of physical energy beyond just our bodies and things like water reels, and now we have a surplus of cognitive energy beyond just our brains, right?”
And that’s a really transformational idea. So I’m completely in that mode right now, honestly. I think that’s just the right mindset for something that’s obviously this disruptive, right or wrong. Tesla is a great example. SpaceX is a great example where people are like, “That doesn’t make any sense.” And Elon’s like, “Well, what if you could land rockets and reuse them and they get really cheap? That’s pretty amazing. What if I can fix the battery problems and a car is basically a software product, right?” Those are pretty amazing what-if questions, right, of those products.
The Birth of Google Docs
Lenny: Yeah. So in this work on understanding what first principles actually looks like when you’re thinking from first principles, the steps are essentially, figure out what you want to do, figure out the levers that keep you from achieving that thing, and then basically question every assumption that stands in the way of making this possible. So I think Elon’s a great… The classic example. You can’t talk about first-principles thinking without quoting Elon, telling stories of Elon. But essentially, it’s just, okay, how much would it cost to make this if we were to start over and not…
The Sequence of Luck
Sam Schillace: I mean, the why-nots, there are actually problems you need to pay attention to eventually to build stuff. But once you have the what-if, right, just to pick on SpaceX for a second, right? If you have what-if of like, “If I could make payload to space cost a lot less, what if?” Okay, that’s amazing. That’s an amazing world. Let’s see if we can work on that problem. And then now you have all the why-nots? Why not? Why isn’t it as cheap as it could be? And you can start to break the problem down and think about it that way. It’s a good model.
Lenny: This connects to something else that I know you’re big on, which is optimism, being optimistic. There’s this feeling that pessimism, you’re often right. There’s growing pessimism in the world in a lot of ways, especially in technology. I know you’re a big proponent of staying optimistic. Can you just talk about why you think that’s important and how you approach that?
Facing Massive Resistance
Sam Schillace: It’s funny, it’s a choice. I’m not an optimistic person by nature. I’m just not. All the people in my life, if any of them listen to this, they’ll just laugh at the idea that I’m a proponent for optimism per se. It’s just a conscious choice. I don’t think you get very much for being pessimistic necessarily. You definitely don’t get a lot for being careless. You can be optimistic to the point of being careless and causing harm, for sure.
Maybe a better way to say it is growth mindset, right? You want to look at the possibilities rather than the limitations, and suspend some disbelief and just work on these problems. I just personally feel like I’ve missed out on more than I’ve protected myself from. If I sum up both sides of that equation over my career, I wish I had been more open-minded and more optimistic, and more willing to try things, and more focused on possibilities rather than problems.
And so I’m just personally choosing to do that, try to do that as a habit. Nothing deeper than that. I just think it’s a better place to be, particularly… It’s kind of funny. When I came out here, I was pre-med. I dropped out of school. I came out to be a computer scientist with my friend. I didn’t think of it that way. I came out to have a job at Ashton-Tate with a friend of mine, and spent 10 years not understanding that I was actually in a career and thinking that it was a temporary thing where I had to go back to med school and get my degree and be a doctor or something boring like that. And so for 35 years of doing this, it hasn’t occurred to me that, oh, actually, I’m in this computer industry that’s this technical industry that’s constantly growing and constantly inventing things and constantly coming up with these new ideas. And actually, the best posture in that world is to be creative and curious and open and optimistic and try things and stuff like that.
The other thing I’ll say about optimism too is related to doing these disruptive things like G Docs. Going back to this idea of all the good ideas look bad at first. Okay, so that’s a first principle. That’s a sort of fundamental thing of you’re going to constantly be challenged by the really good ideas. So how do you overcome it? Well, one way you can overcome it is you want to be able to try things more easily. So part of that is being more optimistic, so being more willing to try stuff, and part of it is also just making it cheaper to try things.
Very early story of G Docs, when I had the idea for Writely, my two co-founders who were both deep domain experts in both app building and word processors were like, “The browser is never going to support this. It’s a bad idea. Let’s not do this.” And they were right and wrong at the same time. They were right that it didn’t support even what we have today and wouldn’t have supported a full experience, but wrong in that the world was going to change and evolve. And we would never have done the first experiment if it had been a long and costly thing to do, right?
So the fact that our tools were sharp and I could say, “Let’s do this thing.” And it only takes a couple of days to get it on its feet and see how it feels, it’s kind of a form of optimism, right? If you’re super pessimistic, you can be like, “Even that’s not worth it. Two days is a waste of time.” So there’s always a little bit of a leap of faith. And then you want to make those as consumable as possible. You want to be able to try things out quickly and learn things and do these experiments.
Now, lots of people have said that before, but I think all those pieces connect for me in this idea of being optimistic and open to trying stuff. Because stuff always is different. You’re always wrong about products. That’s one of my other rules is you’re just always wrong. And so you have to try it. You have to put it in front of people. You have to try it yourself before you’ll understand it. No one can really design products in their head completely as far as I can tell.
Lenny: Awesome. There’s a few threads I want to follow there. But this is also a tool that I found came up again and again in first-principles thinking, people that are really good at this, is just trying it. There’s a lot of just, “Nah, it’s not going to work.” And exactly as you just described, you often find out you’re completely wrong when you actually try it out. And you have this quote, I think in one of your newsletter posts, talking about building Google Docs. You describe it as just fuck around.
Product-Market Fit and Discovery
Sam Schillace: Yeah. Kind of.
Lenny: Get to the edge of something and fuck around. That’s the strategy.
The Future of Documents
Sam Schillace: Yeah. Yeah. Get to the edge is only… Get your tools as sharp as you can get them to be. Make it so that you can try lots of cheap experiments, right? And just mess around and see what happens, see what pops out. And just try to be observant. I think the other part of optimism too is there’s a receptiveness to it, right? If you’re very pessimistic, you might miss the surprising result that pops out of an experiment. You might force yourself to do a bunch of experiments grudgingly, but you’re like, “You know what? I hate this. I’m doing four experiments today because I have to do it because I want to be an entrepreneur, but it sucks and everything’s miserable and black.” And then you won’t notice that, oh, this thing didn’t work, but it didn’t work in an interesting way. And you’re more receptive to that kind of surprising thing, I think, when you’re in an optimistic frame of mind. Like, “Oh, let’s see how far I can get with this. Oh, it’s not working, but why isn’t it working? Well, that’s kind of interesting. It broke here.”
And we’ve done stuff like that in some of the things, the projects I’ve got going in Microsoft right now, we’ve got a chatbot thing we’ve been working on for a while and with memory, long-running memory so that you can have long conversations with it. And they work okay, but they don’t work great in some ways. And we were trying to get multiple versions of them working together, like multi-agents working together. And we gave them whiteboard working memory as a shared working memory thing to fix this problem. And that turns out to make them much smarter. Don’t know why. It just makes them smarter. So that was one of these nice little bits of discovery where if you’re in a pessimistic frame of mind, you might’ve said, “Well, these don’t work that well, let’s give up on it.” More optimistic frame of mind was like, “Well, let’s try to give them a whiteboard just like a person and see if they cooperate better.” And it turns out they really do. So another example of that mindset.
Tech Exploration and the North Star
Lenny: Along this thread, I was going to ask about this earlier, but there’s a lot of technologies people get optimistic about. Crypto comes to mind, not that there’s nothing there, but a lot of people got really optimistic and then it turned out there wasn’t really a lot of business to be built and then things entered wintertime. Is there anything you’ve learned that gives you a signal that, “Lets keep working, I’m going to stay optimistic about this thing?”
Purpose-Driven Exploration
Sam Schillace: I spent a lot of time really thinking hard about crypto and whether I was just reacting to it because it threatens some part of my identity or whatever. And I never came down to anything that seemed valuable. I mean, that was always the thing for me is just there has to be a what if that I can say, “Well what if this works? How valuable is it?” And for crypto I was always like, “Well what if it works? Then I have to run up sec on my personal finances. That sounds dystopian. I don’t want that.” I can’t think of anything as a user that I think actually is valuable here, even in the best case. So I feel like the pessimism is justified.
So that’s one of my other root principles is just like it’s all about user value. Users are lazy, right? We’re all lazy. We don’t really care that much at the end of the day. No one’s going to do something really in their life for any other reason other than it makes their life better. Nobody cares that you’re friendly or nice or the logo is pretty or whatever. They care about making their life easier. We’re all cynical at heart at some level, so if you can’t point at user value, significant user value, it’s not going to work. It doesn’t matter. Shove all the marketing dollars into it you want, you can write all the articles you want, but it’s got to actually solve a problem, a real problem, at the end of the day. I just never saw that with crypto.
Cultivating the “What If” Mindset
Lenny: Yeah. So I think the lesson there is truly understand if the value is real, versus, this sounds really cool, you think a lot of… Would you want it? I think is a nice exercise there.
Courage to Fail Without Shame
Sam Schillace: We pick on poor Elon, but I feel like with a lot of his products, at least he… He’s got lots of other issues, but he articulates clear user value even in the beginning when he’s hyping things up. Like Teslas, right? Okay, so electric cars weren’t ready, he did the Roadsters, whatever. But he at least articulated this idea that we’re going to put a lot of batteries in these things. They’re going to really be real cars. They’re going to have a real range. We’re going to figure out the charging problems. As an end user I’m like, “Okay, that’s great. Now I have a car that works like a car that solves some problems that’s way cheaper to operate, because the fuel is cheaper. He’s solving all the end user problems for me.” That at least makes sense. Even if you don’t believe that he’s going to do it or believe in the way he did it, at least the end user value proposition makes sense, right?
Lenny: You have this other great quote in your newsletter, “People are lazy. Look beyond cool too, on how much easier new tool or tech makes someone’s life. Convenience always wins.” Can you talk about that, just this realization people are just lazy and that’s the key?
Linear Effort, Extraordinary Returns
Sam Schillace: Yeah, well I mean that is the thing. I think as product builders, it’s hard to not love what you’re doing. You build a product because you love it, you build it because you understand some problem, you build it because you want a paycheck maybe sometimes. But we build it for all these reasons that just do not matter to the end user at all. And if there’s one thing I’ve learned about product, particularly in the consumer space, just kind of in general, is people are just lazy about stuff and don’t care about anything other than it making their life better. The thing that’s complicated with that is there’s two things about it that I think are interesting that follow from that principle. One is, I think products almost follow these thermodynamic rules where if you add a little bit of value, your adoption goes slowly and if you add a lot of value, your adoption goes really quickly, right?
I think ChatGPT is a great recent example of something that was just added a ton of new value to the world and got this explosive growth, and then you see lots of other AI stuff that people are doing that’s just not bad, but not great, and it’s sort of kind of adding a little bit of value and slowly lumbering along, or maybe it’s going to collapse under its weight. That’s one thing about the users are lazy part of this. And then the other one I think is, again, it’s almost like physics. I think of this as entropy, or people who are confused about entropy or will be like, “Entropy is not real. Look, it runs backwards all the time on earth. Life gets more complicated. What is the deal with entropy?” And like, “Well no, you have to consider the whole system, right? The entire system of the solar system, including the sun is increasing in entropy all the time. We’re just making use of some of it.”
And I think the same thing is kind of true about user laziness, where people are like, “This tiny thing that I’m focusing on, this feature that I added is better. Therefore, users should adopt it.” But you forget all the stuff around it. The user has to hear about it, the user has to remember it in the moment. The user has to learn how to use it, they have to build the habit. That’s all effort, right? Not to mention the fact that the actual use of your feature might have friction on the way in, right? It might be hard to sign up for it.
When we did Writely at the beginning, we didn’t even ask for an email address, because it was such a novel thing. We didn’t want any friction at all in the onboarding process. So you could just come in and make a document and start using it without telling us anything at all about yourself. And after about two minutes of typing, if you’re still there, we’d very gently just say, “Please give us your email address, no password, no anything. Just give us your email address so we can send you a URL of this document in case you care about it later. Because if you leave, we’ll never know where… We’ll never be able to find it again.”
But we were super focused on that, as little friction as possible. And I think it’s well known in the consumer space, you don’t have… The number of seconds you have is not many. 15 seconds, 30 seconds, right, to convince somebody that there’s some value there. They’re not going to hang out and grind their way through a bunch of high-friction stuff to sign up for your thing. That’s the other part of it is just, users will only adopt what you’re doing if that sum total of energy that they have to expend is less than the resulting ease in their life that they get, usually by a factor of at least a couple, right? So 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.
Lenny: What I think of as you’re describing this is, Microsoft Excel had a billion toolbars and buttons and options, which allowed Google Docs essentially to come in with a much simpler experience. Now you’re on the other side of that, which is I didn’t think about.
A Tangible Lesson From Mistakes
Sam Schillace: Yeah, it’s a really funny place to be. And it’s funny to be at Microsoft, because I’m kind of the enemy, right? Because I’m the guy who messed them up a little bit. So there’s some friction around that. Yeah, I mean I think there are similar trade-offs to be made right now, by the way, with AI. I think there’s similar opportunities. But we made this choice with… So it’s a little hard to remember, right? Because it’s like 18 years, 17 years ago now, 18 years ago almost. In that era, Office was impregnable, right? So software had to be distributed physically, right, it had to be shipped around, it had to be bought and installed. It was harder to use.
And so there’s a very high transactional cost. Because there’s a very high transactional cost, the buyers would always make this decision like, “Do I want the thing with 1,000 features or the thing with 995 features? I don’t know what those last five are, but I might as well have all of them.” And so that was just the lock-in for Microsoft, right? So we made this trade-off, we’re like, “Look, we’re easy to use, we’re zero install, you don’t have to ever deal with it, it’s super convenient. Plus you get this one new feature that’s really, really useful, which is collaborating with each other and not having to send attachments around the old file servers, but we’re going to take away most of the features, because we don’t care about them that much.”
And we took away a little bit more than we should have. In the early days, we’d get all these complaints about people who wanted word count, which I thought was a really weird… I thought that was going to be way at the end of the list. We didn’t have rulers, we didn’t have any kind of formatting at that point, any real pagination. We just had these basic documents. But page word count came in, of course it was students, was one of our early adopters. So they really wanted to know if they were at the word count for the essay that they had just had assigned to them.
So there was this dance with Microsoft that we deliberately made this trade-off. And I think it’s almost like a classic innovator’s dilemma model, right? We had this incumbent that was asymptotically approaching usefulness, they’re adding stuff. Whenever they added stuff, it wasn’t really that much more valuable. And then we were this small thing that came in into a market that they didn’t care that much about, that they didn’t understand that well, which is the internet stuff, this disruptive new thing. We just chipped away from the bottom, like the innovator’s dilemma. And I think it was hard for Microsoft to respond to it. I think it took them a while to even have a clear idea of how they were going to respond to it. In retrospect, they did fine. We took a bunch of market share, but they kept all the money basically. So we being Google. So they survived it. They did a good job surviving the challenge. We have all the users now, but they have all the money. We have all the money, I guess now.
Lenny: I want to spend more time on Google Docs and the story there. A couple questions. How long did it take from starting on it to feeling like it’s working? Whatever you consider product market fit.
Discovering Your Own Surprises
Sam Schillace: Almost immediately, honestly. It was weird, the process of it was, I had this idea, we set this thing up, we started working together. We’re like, “Ah, that’s actually pretty…” So basically history is like I noticed contents editable, so the browser would do some editing for you. And then I noticed JavaScript, I never realized that JavaScript is out there. And we had done word processors in the past, this team, and for a long time. In fact, my co-founder, Steve, the other person on that document, wrote this thing called Full Write way back in 1987 or something like that, that was a direct competitor. ‘85 I think even, that was a direct competitor to Word one.
So we knew word processors, and so we decided to just try it. What’s it like to build a word processor? And the fact that you could collaborate on them was kind of an accident. They’re just these things on the server. We hadn’t built the thing that would lock somebody out yet. So there was just like, “Here’s a document, you can edit these two things.” Which we, A, immediately realized was really cool. We could both work in the same document at the same time. And then, B, realized, “Oh, crap, we’re colliding with each other, because there’s no presence or anything like that. And there’s no collision detection or anything like that.”
So pretty quickly we’re like, “Oh, that’s kind of cool. That feels good as a development team to have these shared documents, not to send stuff around and not… So that’s cool. So let’s build that out.” But like, “Oh, bummer, collaboration’s a problem. We’ll have to go fix that.” And naively figured out that that’s a problem. And it took forever to get that working. It was really, really hard in the time, because we didn’t do operational transform. I don’t think that technique had been quite invented yet. And so we did three-way merge, which doesn’t work that well, because the browsers… The logical document, a document can be rendered differently in HTML. There’s not a canonical representation.
And so you’re doing merges where alphabetization can change, the order of attributes can change, the tree structure can change. Firefox would do a blank paragraph with a singleton BR tag, and IE would do it with an open closed paragraph tag. And so even the tree doesn’t match. So it’s a really hard merge problem. So that turned out to be a gnarly hard problem to solve. But once we had seen the value of working together, we were motivated to do that.
The interesting thing too with that is, I think if we’d gotten it in the other order, we might not have done it. It’s another good example of why not and what if, right? Where we got really lucky that we saw the what if part, that we saw how cool a document in a browser that you could collaborate on would be, because if we understood how hard the collaboration piece would’ve been first without understanding that value, we might’ve been like, “Eh, it’s not worth it. It’s going to be so hard to solve that problem. It’s probably not a useful app.” So I think it’s a good little counter example of that optimistic, pessimistic perspective we were talking about. We could easily have missed that idea, easily have missed that idea. And we just got lucky I think in the order it got presented to us.
The Future of AI and Software
Lenny: That is really interesting actually, that you need to be pulled to the what if getting you so excited that you’re going to spend however many years it took you to solve that problem, because you are so excited about this what if. I think that’s a really good-
Sam Schillace: Yeah, yeah, I mean I’ve spent a lot of time. It’s kind of funny, I keep expecting people to just be like, “All right, grandpa, stop talking about G Docs. It’s been a long time.” Right? So it has been a long time. It’s been 17 years, but it’s still very relevant. It’s got a couple billion users now I think, it’s a big thing. But I’ve spent a lot of the last 10 or 15 years just thinking about, why did that work? What worked about that? What lessons can I draw from it? There was a lot of energy around it, positive and negative. The first week I was at Google, an executive there refused to give me hardware, because he thought that Google was an app company not a… It was a search company, not an app company. And literally the guy in charge of hardware at the time refused to give me hardware for this service. And I had to threaten to either sue him or haul him in front of Eric Schmidt, because I had a contract and I had contracted earnouts.
So that was another one of these interesting lessons of sometimes the opposition is enormous. And if I had just been a random Google employee with this idea and no legal protection, and the CEO wasn’t a fan of the project, it would’ve died. There’s no way it would’ve made it through that negativity and that pessimism and that person being either challenged or afraid of the idea, or just not able to imagine it, or what. I’m not sure what. But we’ll keep coming back to this idea of optimism, but I have this very strong feeling about that most of the reason people don’t do really innovative good products is this kind of mindset. You’re just not seeing the opportunities.
There’s a lot of hard work for sure. There’s a lot of stuff that you can read about and best practices of doing iteration and user testing and user interviews and really listening and all the engineering best practices. That stuff is pretty mechanical. Once you know where you’re going, you can do that. It’s not that hard to learn and to master it. I think the hard stuff is this mindset of being open in the right ways and understanding that some kinds of pushback are good pushback. Some are bad.
I always think that product builders and entrepreneurs, you have this really hard problem of you have to be very rigid about your mission. I know where I’m going. I know what my mission is, and I’m going to go there because the world doesn’t care, it’s going to push back. But you also have to be really flexible about feedback. You’re probably aren’t going to be right about a bunch of it. And so you have to blend these two things together somehow. It’s like a samurai sword that’s hard on the back, but softer on the edge so it doesn’t break. Or the other way around I think. But there’s this hard thing you have to do as an entrepreneur, and I think it’s the real core of building really great products is finding that balance and really listening to those signals, being open to it.
Media Reaction to Gemini’s Launch
Lenny: And also knowing how long to commit to it versus time to move on to something else. So along those lines, what was the moment where you finally felt product market fit for what became Google Docs, and how long was that from the beginning of starting to work on it?
Sam Schillace: It depends on what market we’re talking about. I’ve been continually surprised at the adoption of G Docs. I think we knew there was something there pretty quickly, probably in the first couple months. There was a lot of energy around it. It was a weird ride, because we built this thing on a whim, and as an experiment. We liked it. We decided to go just advertise on Google. At the time, 37signals was the cool company. And we’re like, “That looks cool. We’ll just be some engineers and we’ll have a little subscription SaaS business thing and chill out. So let’s see what it costs to acquire customers. So let’s go advertise on Google and see how much it costs to get people to sort of show up, and then we’ll figure out if we have a subscription business or not.”
And that just got us noticed. That got us noticed by Google. It got us noticed. We were I think one of the first 10 articles at TechCrunch, like Michael Arrington. Another funny story is that I had a breakfast with Michael Arrington at Bucks, in that era where he was trying to decide, he had this spreadsheet idea he wanted to work on and he was trying to decide he should go do that and maybe join forces with us, because we were cool, or if he should continue to work on this blog thing he had gone called TechCrunch. So I might be partially responsible for TechCrunch, because we turned him down and said, “You should go do TechCrunch instead.” Every time I would bump into him, I would laugh about that one.
But when we got noticed, we really got noticed. There was just this period where we were the hot thing for a couple months, where every VC wanted to talk to us and everyone’s trying to figure it out. Because I think we’re like… When you see one point on the line, like Gmail, which came before us, you’re like, “Oh, that’s kind of cool. That’s an interesting quasi-app.” But it’s like a weird kind of app. It’s serialized. You can’t really interact with it that much. And then you see this as another point on the curve and you’re like, “Oh, that’s a real app. Oh, crap. I wonder, is there anything stopping us from doing the rest of Office? Oh, probably not. How far is this going to go?”
And so I think we were that second point that showed that there was actually this totally different paradigm. And so we just got this enormous amount of tension pretty quickly. And then the rest of it was feeling our way through what does it actually mean? How much of the functionality do we need to build? What’s the really important part about it? How much of its collaboration? We spent a bunch of energy on offline, which was miserable, which never turned out to matter that much.
Now that team has spent a long time replicating all these features that we abandoned by the wayside, which I think I’m not that interested in. I think the future of documents looks very different than what we have now. I think it’s kind of funny now that we’re spending billions of dollars on GPUs to emulate wind pulp and ink pressed by metal type. We’re building linear documents that are fixed, that are static. So one of the things we’ve been doing with these chatbot things is they also serve as documents. I say bots or docs all the time. And so you’ll do these things where you… I do this all the time, where we’ll interview… We’ll create a new one. It has a separate identity, as a separate document, and then you tell it like, “I’m going to write a technical document. Here’s roughly what it’s about. Why don’t you interview me?”
So it interviews you for an hour, and now you’ve got this nice linear artifact which you can read. It’s very readable because it’s conversational. But at the same time, you’ve been building all these semantically encoded virtual synthetic memories in the Spectre database. So you can come in and say, “Show me a diagram of this. Draw this diagram for me, change it in the following way. What is this? If I change this, what if I change that? Summarize this part of it.” So you can start to interact with it. That’s still creating stuff at the bottom of this linear artifact. But the next step that we’re working on now is just making that dynamic where you just come to something and you talk to it and interact with it.
I think one of the things that’s going to happen is, just like it seems… Well, in the early days of G Docs, people would say, “Well, what if I’m not connected?” And one of the things I would say is, “In three or five years, if you get handed a device that’s not connected to the internet, your word for it is going to be broken.” Which is true, right? It’s anachronistic and weird if something’s not connected. I think we’re going to feel the same way about intention and interactivity in our products very soon. If I can’t tell something what my intent is and have it configure itself in an intelligent way, have it converse with me, whether that’s a device or a piece of GUI UX somewhere, I think it’s going to feel anachronistic. It’s going to feel really weird.
There’s that scene in one of the early Star Trek movies where Scotty tries to talk to the mouse, right? He’s like, “Computer make the…” He’s pissed off because he can’t talk to the computer. We’re all going to be like that in five years I think, about it. And it’s going to seem really weird that we have these applications that I can’t collaborate with the application. Why can’t I collaborate with the applications? It’s like the application’s locked on a file server just like the pre G Docs days. Why can’t I just interact with it and have it configure itself the way I want it to configure itself, and show me the data the way I want to see this, and let me build the workflow the way I want it, and remember it for me and bring it back later, and all that stuff? So that was a long digression, but I would just like… You’re asking about features and functionality, and I feel like where we are now with these feature wars, it’s just silly. It’s not the point at all. I think documents are going to change radically in the next few years.
How to Keep Up With AI
Lenny: I want to follow that thread. Before I do, I found the first TechCrunch post about you guys. Starts with, “Imagine Word, but as an Ajax browser application.”
Characteristics of the Current Stage
Sam Schillace: Oh, yeah, there you go. Yeah. It wasn’t even JavaScript. It was Ajax.
Why Microsoft Keeps Innovating
Lenny: Ajax, so hot back then.
Sam Schillace: It’s also funny too, because I’ll talk to young front end developers these days. I’m like, “I don’t want to scare you too much, but jQuery didn’t even exist when I wrote this thing. This is like bare metal in the DOM, and there were bugs.” When I went to Google, I had to write this little network stack at the bottom of the JavaScript that in theory [inaudible 00:43:21] you could interrupt, you could have multiple requests inflate and you can interrupt them and discard them and stuff.
But the stack at the time was really buggy, and I think it was IEE. And so I wrote this little network queue that would keep track of whether there were requests in flight, and kill them in a way that didn’t break everything. And it was hard to do, because it’s this weird asynchronous programming. And that piece of code, when I went to Google, they made me reformat it for the JavaScript readability standards, and I could not get it to work with their formatting. There was some bug in the JavaScript compiler of the time that whitespace mattered. And so I wound up checking it in broken, got the readability badge and immediately fixed it. It’s like that was another one of our little hacks to get this working.
Quick Fire Q&A
Lenny:
There’s often this criticism as an engineer, you just want to work on interesting things and work on the technology before you find a problem that it’s solving. It feels like with this example it was, you just think this is a cool technology, let’s see what happens. Do you have any, I don’t know, learnings or advice for when it’s actually fine. Let’s just play with this tech, be at the edges, you said, and maybe it’ll lead somewhere, versus you should probably try to avoid that and first focus on a problem.
Great Questions for Interviews
Sam Schillace: I’m guilty of that. I mean, I like to play with stuff. I tend to think with my fingers as much as anything else. So I actually think there’s a good place for just play with the tech a lot and figure out what it’s good for. What I’ve evolved to doing these days with my teams is I pick what I call north stars that I think are interesting, useful things to get to rather than just messing around. What’s a cool thing that I think might be buildable with this?
So right now we’re doing these multi-agent systems. We’re trying to figure out how much independent work they can do without a person holding their hand. And so a nice domain to test that out in is programming, because you don’t have a whole lot of… You just give something a Python environment and a file system and that’s it and that’s all it needs. And so you’re not distracted by connectivity issues or whatever.
So one of the problems right now is go write the eye in Python. That’s a problem I could give to an intern and it would take them a summer to do some halfway decent job of it. It’s a thing you could expect a reasonably competent programmer to do, mostly independently. And so it should be possible for the system, if it’s independent at all, to go do that. So is that useful by itself? No, because we already have the eye, it doesn’t matter. But if we build a system of programming agents that can self-monitor and self-correct and bug themselves, that can build things that are roughly that scale of complexity, that’s valuable. That would be a valuable thing to have.
It’s kind of interesting too, because that system already, it’s produced a bunch of good insights. One of them is its kind of complicated and then hard to debug it. It’s this asynchronous system of stochastic agents. That’s a lot of stuff to deal with. So we wrote a debugger agent. And debugger agent watches stuff, and when there’s a problem somewhere, it goes and figures out what the problem is and then gives you a nice explanation of what you broke and what needs to be fixed. And we haven’t turned it loose on actually fixing things yet because we don’t trust it, but it’s very helpful as an assistant [inaudible 00:47:15]. We had one that documented itself too. That’s the other one we did recently. Just turned it loose on documenting the code base and did a pretty good job of it.
So it’s starting to produce interesting stuff, right? Because we have these north stars that we aim things at. And I think that’s maybe a good antidote to this. Just playing with tech without being focused doesn’t tend to produce anything that’s super valuable. But picking these, even if they’re arbitrary goals, as long as they’re real goals that you’re trying to get to, that’s useful, right? Where you’re like, “I wonder if I can get this to work. I wonder if I can build this thing.” Then grind away at that for a week and see how close I can get. See what I learned about why it’s hard. That’s probably better than just like, “Let me poke at JavaScript for a while.”
Lenny: It’s also different, I think, at a bigger company where you need to achieve something, versus I think as just an engineer out of college just playing around, like, go for it, right? It’s just like, what’s the worst that could happen?
Electric Muscle Cars
Sam Schillace: Even the early days of Writely, the very… I mean, we had a goal from the beginning. The beginning was like, “Can I write a word processor in a browser?” That was literally the problem statement, right? It was like, “I have content edible, I have Ajax or JavaScript. Can I put these together and something that feels like a word processor? Let’s go do that.” It’s kind of half messing around with tech, but it’s also half an actual goal. So I don’t know.
I like playing around with… I think a lot of the good product ideas, most of the good product ideas actually come up from engineering. So I think there’s a lot to be said for, get familiar with tools, particularly weird esoteric combinations of tools can often be useful. If you understand or three things… At Google, I was one of only two people in the company who had the code readability, which is the right to check in code in this language, and both a backend language, which is the monitoring language, Org Mode, a middle tier language, which is Java, and a front language, which is JavaScript. No one else would do that full stack. I think it’s useful to have that broad perspective sometimes.
Lenny: Sam, the Renaissance man of all languages.
A Guiding Life Motto
Sam Schillace: Yeah, ADD more like it, but yeah, I pay attention to things.
The 200-Pound Blood Sausage Story
Lenny: I wanted to follow this thread a little further around being good at these what if questions. It feels like you’ve built this, or maybe you were born with this skill of thinking in the future, thinking about what’s possible, thinking about where things are going. Is there anything that you could recommend to people listening to get better at the skill? Because for a lot of product people, this is really important to figure out, where could we be going and let’s work back.
Sam Schillace: This is a really interesting question. And I may actually write, I’ve been thinking about writing a book from some of my Sunday letters, and this is maybe the frame of it. So I’m curious to see how flamed I get from saying this, it’ll be interesting to see. I think there’s this weird thing that I’ve noticed. I go talk to university kids and stuff like that, and there’s this weird thing I noticed where when I was in university and I would talk to old guys like me, they would all say the PC is this stupid toy, whatever, it’s not real computing, go on a mainframe or whatever. And my attitude was like, “Out of the way, old man, just like, you’re irrelevant. I’m going to go do this thing. It’s awesome.” And go, go, go.
And now when I talk to kids, I actually had a slide up at Michigan when I was talking recently that was titled, okay humor, and the professor actually put it up there. Because this generation is very pessimistic and doesn’t seem to be quite as engaged and energetic about solving problems. And I’ve been puzzling through it. And I think maybe there’s a bunch of different things that intersect.
I think one of them is, I think they all have to do with the willingness to take risk and to fail, honestly. I think that’s really where it comes from. So I think 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 normal. So you see everybody makes 100 million, you’re an idiot. There’s that stuff. There’s also, you’re living out loud, so when you fail in that context it feels very painful.
But I think there’s also for elite students, like people at these elite schools, they’re hard to get into. I went to Michigan, “You’re kind of smart. Michigan’s a good school. You live nearby, go apply to that one.” Nothing serious about it, but kids in the elite schools, their lives are highly curated going up to getting into a school like that now, right? Those students like, “I didn’t do sports, I didn’t do extracurricular. I was just a weird nerd having to be good at math.” And so I think there’s that as well. If you’re highly curated where you’ve spent a lot of your life thinking, “Everything I do has to have a reason and an output.” It’s very hard to just mess around and do something that might lead down a surprising path, right? So that’s the curation is part of it.
And then I think just in about the mid ’80s when I graduated from high school, we stopped letting kids just play on their own, unsupervised outside with other kids. I grew up in this neighborhood full of, it was like the faculty ghetto for this small university my dad taught at, and we just ran wild. It was on the estate of widow of Dodge Motor, founder of Dodge Motor, so we had a couple hundred acres of swamp and fields to go run around in. We did hair-raisingly, dangerous things that my parents never knew about, and really explored and had fun.
I think if you put all those pieces together, I think there’s much less of an ability and willingness and skillset around experimenting to the point of failure, making a fool of yourself, having bad ideas. I send stupid emails to Satya at Microsoft all the time where I’m just like, “I don’t know what the hell he thinks of me at this point.” Because I send him all these goofy ideas. I think he actually gets it and he’s like, he likes it, because I don’t think people usually do that for him. But I’m just like, “Man, this is…” Then I’ll send him an email a week later. I’m like, “Yeah, that was a dumb idea. Sorry about that. I’ve decided that wasn’t a very good one.”
But I think you cannot dance if you can’t… If you’re afraid to embarrass yourself. You cannot succeed if you’re afraid to fail. That’s just how it is. You have to have that sense of play. You have to have that sense of, it’s okay if this doesn’t work, I’ll iterate on it. I have this personal motto, which is, from error comes virtue. Because I’m a maker, I make stuff. And I fuck it up all the time. I have poor motor skills, so I make mistakes constantly, and then I just figure out how to make the mistake into a virtue somehow. So I think it’s a really good skill to have. The saying I took on this year and I really, really like it as my… I never had a personal motto before. I think it might be my personal motto, it’s like, virtue from error.
Outro and Contact Info
Lenny: Amazing. On this topic of failing. I think a lot of people hear this advice and they’re like, “Yeah, okay, I need to fail more.” It’s hard to do. And oftentimes your performance at a company is negatively impacted. And it feels like for you it was just, you’ve done it enough times where fine, okay, it’s going to be fine. I launch this thing, no one cares. I email Satya this thing, he ignores it or he doesn’t. It’s going to be okay. Is that maybe the key to this or is there anything else that you’ve done to allow you to be okay with failure?
Sam Schillace: I feel like you can have a linear return on your effort if you manage things in a linear way, which is I think that tightly managing, okay, nothing’s going to be surprising, I’m going to be within this boundary, I’m going to slowly accrete value, I’m going to play this game, whatever. I think you can have a nice linear, boring return to your career and you’ll climb the ladder, it takes 30 years or whatever. I don’t have the patience for that. And I think the way you get extraordinary returns is you do extraordinary things, right? You have to take bigger risks, and have more interesting shots to have this kind of extraordinary result in your career.
I feel like I always tell people, I think… I mean, I pitch this because I’ve observed myself and thought about what has been successful in my career. It’s not a thing everybody can do. I’m just kind of like this. I never really fully grew up. I’m kind of this weirdo. I still feel… I’m 57 now, and I feel like I’m about 17. I’m still very immature and like to mess around with stuff and play with things. So not everybody can do it. But I think there’s, at the end of the day, the reason you get ahead in your career is you had a lot of impact. And the reason you had a lot of impact was because you picked something that you’re good at that you did with a lot of intensity that wound up having impact, right? And so I think the good at part of it is hard too. We tend to undervalue the things we’re good at. We tend to think work has to be unpleasant. And so if something is easy and fun, we don’t tend to think it’s very valuable.
So I think lots of people gravitate in this direction of, let’s go do unpleasant things and grind our way through the career, because that’s the way to make it. But the reality is, 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 of it, right? Do it as hard as you can. If you get pleasure from doing something that people want to pay you for, do it the best you can do it, as hard as you can do it. And if that’s messing around and playing around with cool ideas, do the hell out of that.
Work doesn’t necessarily have to be hard. It often is, but it doesn’t have to be. And the best case is that it isn’t. The most impact you’ll ever have is where you’re in that mode where you’re just in the flow and doing your thing and you’re happy to do it and you can’t quite believe they pay you and you don’t understand how you’re getting away with this, but it’s super cool anyways, right? I think that’s the career thing that makes sense to me. At least that’s what I’ve done. Who knows? It’s all luck sometimes, so it’s hard to replicate these. Everybody has a different path.
Lenny: Amazing. I love that advice. It’s exactly where I was going to take our conversation, so I love that you took us there. It makes me think of, I’m reading Charlie Munger’s Almanac, which just came out through Stripe Press. And Warren Buffett and Charlie Munger’s whole philosophy is, when you find an advantage, just go huge. Just go big, right? Make one bet a year. But when you find that, go for it. Don’t buy a little bit at a time. And I love that. That’s exactly what you’re saying.
Sam Schillace: Sometimes things don’t make sense either. So I’ll lean over and show you, for people with the video. That’s an instrument I made. It’s an instrument. So the top of that’s a piece of redwood, this one right behind me with the cat eyes, the reason it’s got these weird cat eyes, by the way, this is virtue from error right there, I dug this piece of wood out of the forest. It’d been sitting on the forest floor for 80 years trying to not rot and doing a pretty good job of it, because it’s Redwood.
And there were knots in it. So those two cat eyes are where the knots were. There’s another knot right here that I couldn’t get out. But there’s two knots in there that I had to carve out of there. That’s a very weird design that I did by hand. It doesn’t quite work. It’s kind of a failure. The arch of it’s a little bit too high, so it’s a little hard to play, because the pick hits the top because the strings get a little bit close to the top. So like that…
But that’s an experiment. I was playing around. I wanted to do this thing. It was fun to do. It was a passion project. Now it’s just hanging on the wall. Not everything works. Clearly I don’t have a career as a luthier either. So it’s more just a fun thing to do. But that’s just a good example of the… I don’t know. Sometimes I don’t even understand why I do stuff. You just do it because you do it. Because it makes some sense to you.
Lenny: Many people are in the opposite boat where they don’t like what they’re doing, they’re miserable, but they have to have a job. They need income, they need to pay their rent, feed their family.
Sam Schillace: I know, I realize what I’m saying is very privileged, and I am sorry about that. But from some perspective.
Lenny: But I imagine you were also in those situations occasionally. Is there anything you recommend to folks that aren’t in that, I would do this for free, I’m so excited about this work? Do you recommend try to get out that as soon as you can? Is it enjoy it as much as you can, get the most out of it?
Sam Schillace: I mean, I’ve done plenty of things for money. I’ve done plenty of jobs to make money for my family, things I did not enjoy doing. All I can say is really, I stopped doing those things as soon as I could stop doing them. Not only as fast as I physically could, because I definitely had that Calvinist, oldest boy thing of, must provide, must suffer kind of thing. So it took me a long time to realize, no, actually I’m really creative. I don’t have to be like everybody else. I can have my own path, and I can be this weird engineer. I always joke that I’m an engineer. Two is a prime number. It’s just like I’m kind of a programmer, but not real… I’m this weird non-linear person that only barely fits into the programming world. But it’s okay. It took me a long time to figure out that I could do that, that I could be comfortable with that part of myself.
And I’m fortunate enough now that I’ve done enough things and have enough of a connection and network that people understand who I am and the value I can bring. And so I get away with doing that stuff that I like doing. So it is kind of privileged advice, and it’s not something everybody can do at every stage in their career. Certainly earlier stages you often have to make compromises, but I still think it’s worth paying attention to, right? When you’re working, what makes you happy? What is the stuff that you feel guilty for getting away with? When I started managing people, I couldn’t understand why people were paying me and I wasn’t writing code, because all of my energy was attached to, “I can produce a lot of lines of code every day.” And I asked my boss at the time, “Why are you so happy with this? I’m not writing anything.” He’s like, “I don’t know what you’re doing. Everywhere you go, it gets better. So just keep doing whatever it is you’re doing.”
And that was one of these moments where I was just like, “Oh, I could do something else and it’s kind of fun. I like talking to people all day. That’s great. They’re going to pay me for talking to people all day. If they seem happy, I’m happy. Let me just lean into this for a while and see where it goes.” So I think you just look for those moments when somebody is willing to let you do something that you feel happy to do, surprisingly happy to do, or doesn’t feel like it’s the thing you “should be doing”. If you get those surprises like this, I think this goes back to this openness, the optimism that we were talking about. You have to be receptive and attentive to those moments when they show up. So I think they are in every career. If you listen for them, you’ll see stuff show up where you don’t think of it as who you are, but somebody else sees it in you and if you can be open to it, you can do these pivots.
Lenny: When you talk about that, that makes me think of Seth Godin has this really important advice that’s always stuck with me, that no matter what job you’re in, just try to enjoy it and do the best version of that job you can, because you’ll enjoy it more, that you’ll end up being more successful, and it’s just a good habit to just like, “I’m just going to do the best I can at being a waitress at this place. I’m just going to do the best at greeting people entering the Apple Store.”
Sam Schillace: Absolutely. Yeah, I think that’s absolutely right. And even more than that, just find a way to bring yourself to it, right? What is the thing that you can do in this role that is unique to you that you’re the most comfortable with, right, where you really have high impact. I don’t know. I tend to be very, it’s kind of a joke because I’m a programmer, I tend to be a very binary person, right? I’m either all in or all out on something. So whenever I do something, I do it with just a ridiculous amount of intensity, for better or for worse. So I find the ways to do stuff. I tend to be very unhappy if I can’t be intense in something successfully, and then if I can be intense in it successfully, I’m happier and the people around me are probably less happy, but stuff happens. Anyways, I get things done.
Lenny: Speaking of getting things done and being intense and working things that are really interesting, you’re responsible for some of the cutting-edge work happening at Microsoft in AI. You’re spending a lot of time in AI. I’m curious to get your take on just what you find interesting, where you think things are going, what people should know about AI. I’ll share a couple of quotes that you put out somewhere that I have here that I think are cool. One is, “AI isn’t a feature of your product. Your product is a feature of AI.”
Sam Schillace: I love that one. Yeah.
Lenny: Another is, “It’ll be possible to add some value by building AI into your product, but really transformative massive value will come from building apps and solutions that won’t work at all without it, that treat it as a true platform.”
Sam Schillace: Yeah, I think both of those are really true. So what I’m working on is, most of the industry right now is focused on, when you talk about somebody who’s working in AI, it’s somebody who’s creating models, right? It’s somebody who’s figuring out how to do some new open source model or somebody who’s doing some new training or make some model bigger. And I think that’s very, it’s valid, useful work. It’s just not the kind of work I like to do very much. And a lot of people are doing it.
And so I’m an app builder, I’m a tool builder, and so I don’t create models, I consume them, right? I want to build things around them. And so when I started with Microsoft, started working on GPT-4 with Microsoft in September of last year, my immediate reaction to it after picking my jaw up off the floor, which we were all doing in the early days, was, “Okay, this is cool, but in some computer science sense, it’s just this function, this stochastic pure function that just takes a character array and rearranges it and hands it back to you. That’s not much of a building block for building programs. We need state and we need control flow, and orchestration, and call-outs.”
So that just started me down this rabbit hole of thinking about building the Semantic Kernel, which we built, and then building Infinite Chatbot, which was next, and these other projects we’ve been working on. And the more I think about this stuff, the more I do think those two quotes are good quotes. I think what’s going to happen over time… I actually think we’re at the beginning of this just gigantic disruption in the software industry. I think the way that the internet made distribution of information free, I think AI is going to make pixels free.
So pixels are expensive to produce now, they take programmers and they take lots of infrastructure, and putting a pixel in front of the user is a hard thing to do and lots of software is predicated on that. Lots of business is. The way lots of businesses were predicated on it being hard to distribute information 25 years ago. But you can see this already with things like just images, right? Two years ago, if you wanted a piece of digital art, you had to go invent Photoshop, learn to use Photoshop, use Photoshop to do the drawing, build the skills up. That’s a lot of work to produce those pixels. Now it’s like, I want a picture of a cat riding a bike eating a banana, done, right? So those pixels got really free.
And the similar things are happening in the business world as well, and I think it’s just going to start to happen everywhere. So you can draw… This is what if, let’s go ask some what ifs. So what if the models get really good at planning, so they get more independent, they can do longer and complicated things? What if the multimodal stuff gets really good so that they can both consume and produce dynamic UI like I was talking about? What if we figure out a good way to store state, this is my bots or docs thing. So what if do we figure out a good way for you to really highly personalize something so it knows you really well and you trust it with confidential information?
If you have all of those things, you’re just going to spend a lot of time talking to that agent. It’s like, what would you do… If you imagine you’re the richest person in the world, you’ve got 100 of the best people working for you, and a chief of staff, and they’re tireless, and they never fight with each other, they do everything you want. With that staff supporting you, what are you doing with software? What are you doing when you’re sitting in front of a screen? Well, you’re probably communicating intention and you’re probably consuming some either entertainment or some of the products of that intent. And that’s about it. You’re not messing around with pokey static apps and stuff. That doesn’t work, right? You’re just telling your staff to deal with stuff for you.
So I think that’s where we’re headed, I think, in the world of software at least. Things are going to get more dynamic, more intentional, more semantic, more fluid, then more personalized. I think there’s a ton of problems to be solved to make that vision real. But I think this feels to me a little bit like seeing the Palm maybe, or the early iPhone, where you’re just like, “Okay, I get it. Phones are going to get interesting. That’s a new device. Now we got to go do a whole bunch of engineering before they actually are as useful as they’re today.” Right?
So I get it. I think software is going to change radically now. I had the same feeling when we started doing… This is going back to G Docs again. It’s another lesson. It’s another one of these category shifts. The second we got Writely up on its feet and I was like, “Ah, the browser is actually a platform that you can actually build real apps in. I get it, the world’s going to change.” And we had a ton of stuff to do, right? Nobody really understood distributed systems. Nobody understood how to build stuff in multiple places at once or how you deal with replication, how you deal with security, all kinds of hard… All the development patterns had to shift from Waterfall, to Agile, to CI-CD, all this stuff had to change to fully realize that world.
But I remember back in 2005, this very quick… And the people who were the strong proponents, I think all saw this, instantly saw that the world had changed and there was this new category. And I have exactly the same feeling about generative AI. Like, yep, software’s going to totally change. These businesses are going to totally change. It might take 10 years to really work through all of it, but yep, door open, new room, new game, start coloring in the blanks, right? Let’s go.
So that’s where I think we’re going. And that sounded really certain, it probably sounded more certain than I should sound. I think there’s a lot of, probably a quarter at least, if not a half of what I just said is wrong in some way. So we’re going to learn a bunch of stuff along the way. And there’s a lot of work to do, a whole lot of work to do and a whole lot of unanticipated side effects are going to pop out, and there’s just a whole lot of stuff to get that to be real the way there was with all of the last transformations. But I think this is just a giant category shift. I think it’s just incontrovertible that it is.
It’s kind of funny when Gemini came out, all the press take was like, “Oh, it’s not that different from GPT-4. I guess we’re done with AI now, we can go back to bed.” I’m like, “That is the dumbest possible interpretation of that story that you could come up with I think.” Of all the takes you could have had on that, I think that was the dumbest one honestly. It could say many things about either company. It could say many things about the science. But, “Guess there’s nothing here to see.” Is not one of them.
Lenny: For somebody listening that wants to not fall behind on this and or find opportunity for their product, other than just playing with it, which is what everyone is always saying, just like, “Play with it. Use ChatGPT, use Bard, and all these things.” Is there any advice you’d give listeners for how to approach thinking about AI, how it integrates into the stuff they’re doing?
Sam Schillace: Yeah, I agree with you. Just play with it, is not really great advice. I think the best technique I’ve really seen for learning things is to pick a thing to do with the thing you’re trying to learn, right? Even if it’s an unreasonable… Even if it’s a goofy, weird thing, right? Like I’m going to figure out how to draw funny pictures with this programming language or whatever. Even if it’s a dumb thing like that, picking some arbitrary goal and being a little bit stubborn about trying to get yourself to it is a good way to learn stuff.
And then the question is, “What goals are you picking?” So try to pick goals that lead somewhere maybe at least a little bit interesting. So if your goal is just like, “I’m going to mess around with ChatGP for an hour.” That’s not really much of a goal. If it’s like, “I’m going to go try to GPT that can do this part of my job, let’s see how close I can get.” That’s more interesting, right?
And I do think unfortunately, one of the other ways in which this is very reminiscent of the early dot com era, and I think plenty of people have said this, is this sense of exhaustion in keeping up. There’s so much stuff going on right now. And that’s I think another good strong indicator that something really big is going on, where it’s just very, very difficult to keep track of all the stuff that’s happening. It’s kind of interesting, because I remember, just to kick crypto’s corpse one last time, there was a tweet at the beginning of the year that I saw that somebody was like, “Yeah, it took the AI bros a week to come up with as many use cases as crypto came up with in a decade.” Which definitely feels true, right? It’s just so much stuff going on. And I think you just have to try to keep track of it.
One of the other things I think is going on in the moment, which feels a lot like the cloud moment to me, is it’s hard to get the first idea. The zero to one is hard. Understanding that there’s something there at all is the really hard part, because you have to be lucky and you have to be talented and you have to look in the right place and do some very hard work. But once you understand that there’s something there, like the cloud model works, or their generative AI matters and scale works and stuff like that. Once you’re there, the one to many, all the optimization stuff, that happens in parallel, it happens really quickly. Many, many, many people can do it. There’s a lot of energy. It’ll just go really fast.
So I think we’re in that phase where just like we’re in the elaboration phase where we understand this step and people are just filling in all the white space as fast as they can. So that’ll slow down eventually, hopefully. We’ll see what the next year brings. But yeah, it’s a hard time. It’s hard for professionals even. I think you just have to read a lot. You have to think a lot. You have to play with stuff. You have to choose your battles. You have to pick good targets. Pick a goal in your domain that would matter to you if you can get to it, and then go try to solve that problem with some specific technology and get to know that technology.
Maybe pick the technology based on how popular it seems, if you want to learn something lots of people [inaudible 01:12:44]. Those are all good. I think you just have these kind of mundane… I don’t think it’s really a secret, I don’t think. They’re kind of mundane strategies, but you just have to pick some stuff and do some homework. And there’s no magic single bullet to learning this stuff. You just have to run. It’s like, “How do I run this sprint without getting out of breath?” You don’t. You’re going to be out of breath, run hard. It’s just what it is.
Lenny: I love that advice. And I love it connects to everything else that you’ve been talking about is, find some problem someone has, find some value you could provide, and then think about, how can AI potentially provide that? I think that’s really practical. Great advice. Maybe a last question just around Microsoft. It feels like Microsoft is firing on all cylinders. It feels like it’s become one of the most innovative companies out there. It feels like Satya is known now as the most innovative, best executing CEO out there potentially. That’s just what it feels like. Being on the inside, I’m just curious, what is it that you think Microsoft is doing so right or how they think that enables them to be so innovative and continue to be such a behemoth as so much has changed in tech?
Sam Schillace: There’s a couple of things. First of all, I think very, very highly of Satya, or I wouldn’t be there. I really, really like Satya. He is very much what he appears to be from the outside. I’ve had plenty of candid private conversations with him and watched him in meetings and stuff like that. He’s a very decent, genuine, honest, high energy, caring individual with a ton of empathy. He’s really motivational in a way that is not destructive. He believes very strongly that a leader’s job is to raise the energy of the organization and he really lives that. And I watch him in meetings where I’m a domain expert and I cannot believe how engaged he is in stuff and how much he understands about something that I know he’s not as deep an expert as some of the people in the room are and he’s fully in there. So he is a really incredible leader in many ways. That’s one thing.
I think another is that culture is very humble, honestly. I mean, it’s interesting going from Google to Microsoft. And I don’t want to draw comparisons or anything like that, but I think there are definitely similarities and differences between the companies. And I think one thing that stands out with Microsoft is it’s a humble culture. It does unglamorous work all the time to make businesses be successful. So it’s got that sort of mindset of just hard work and humility, which I value a lot. And then I think there’s just, I mean honestly, there’s just a fantastic number of really talented people working there.
It’s kind of funny, this year I’ve been writing a lot of patents, because there’s just a lot of stuff going on in the world. And I’ve written more patents this year by a lot than the rest of my career combined. And I commented on the number. I’m like, “I’ve written 15 patents this year.” I commented to the patent attorney that I work with. He’s like, “Ah, that’s nothing. The chief science officer wrote 700 of them one year during the mobile boom.” It’s like, whatever you feel about patents, and I don’t necessarily love patents, although it’s part of my job to write them, but that’s the kind of people that are there, these just fantastically talented people with just really deep experience at a lot of different levels.
I think that’s part of it too. There’s just a lot of really good folks there. Kevin Scott, who I work for, is one of the smartest… He’s probably the smartest person I’ve ever been fortunate to work directly with. He’s definitely one of the smartest I’ve ever met. He’s pretty fantastic. And the group around him is pretty fantastic, and the leadership in Windows and Office is pretty fantastic. So it’s just a lot of good people and a lot of good attitude, and a good leader I think is the answer. And luck. It’s always luck too, right? Kevin and Satya made a bet on OpenAI a few years back, and they’re doing some extraordinary things with capital raise and the support of that technology and stuff like that. That doesn’t happen by accident either.
Lenny: No drama there, by the way.
Sam Schillace: I wouldn’t know anything [inaudible 01:16:42] there. I hear it’s calm. I stay far away from it as possible, honestly.
Lenny: Most under-the-radar startup out there. Amazing. Sam, is there anything else you want to share before we get to our very exciting lightning round? Is there anything you want to leave listeners with?
Sam Schillace: Mostly I think probably take all of this as more my personal opinions and not Microsoft’s official stands. This is just me being an engineer. I’m not here as a Microsoft representative necessarily. But yeah, I don’t know. Build stuff, solve problems, build stuff. That’s what it is, right? That Jobs quote-
Lenny: Fuck around.
Sam Schillace: Fuck around. Well, that Jobs quote is right. The world was built by people just like you. That’s that’s the thing. It’s really true. You don’t have to have permission, you just have to have energy.
Lenny: With that, we’ve reached our very exciting lightning round. Are you ready?
Sam Schillace: Probably not.
Lenny: Great answer. Sam. What are two or three books that you’ve recommended most to other people?
Sam Schillace: I like weird book. There’s this book called Invisible Cities by Italo Calvino, which is this very beautiful meditation on the nature of cities and the nature of Venice. And I read it the first time I was in Italy right after college, and it just blew all the circuits in my brain when I went to Venice, and so that was pretty cool. The other one I recommend to people with some caution is this very intense and disturbing book called The Wasp Factory by Ian Banks, which is probably the creepiest and hardest book I’ve ever read, but it’s a very interesting psychological deep dive. So those are both fiction. If you’re looking for business advice, I think the business advice one is probably, Where Good Ideas Come From, Steven Johnson. I really like that book. I think some of the stuff about the adjacent possible. I think it’s an old book now, but I think it’s pretty timely still. I think he had some good stuff in there.
Lenny: What is a favorite recent movie or TV show that you really enjoyed?
Sam Schillace: My favorite one right now, my guilty pleasure is watching Gary Oldman be a completely disgusting over-the-hill British spy in Slow Horses, which is pretty fun. And I’m actually having a little fun with the retro monster stuff like Monarch and stuff like that. It’s kind of fun to watch. I don’t know. I have absolute junk food taste when it comes to media.
Lenny: Love creepy monsters and bugs.
Sam Schillace: Yeah.
Lenny: I love it.
Sam Schillace: Shooty science things that blow up a lot. I’m not very deep.
Lenny: By the way, have you seen Scavengers Reign, Scavenger Reigns, Scavenger Reign? Okay. You’d love it. It’s on HBO. It’s incredible. It’s an animated sci-fi thing, and it’s very creepy, slimy, kind of alieny things, and it’s so beautiful. Do you have a favorite interview question that you like to ask candidates when you’re interviewing them?
Sam Schillace: I have one that got banned at Google, and I liked it. I still think it’s a fun question. It’s, how many zeros are at the end of 100 factorial? And the reason I like it… Yeah, it’s like, you made the face, right?
Lenny: Going to ask ChatGPT for this answer.
Sam Schillace: Well, here’s the thing. Well, don’t though, because the reason I ask it is, it seems like an unreasonable and impossible answer. And if you sit down and think about it a little bit, I’m not going to tell you how to reason through it, but you can figure out the answer to it in a few minutes. And so the reason I ask is not to get to the answer. It’s because I just want to see how people react when I give them something that seems impossible and unreasonable.
And some people just back off and refuse to engage with it. And some people are just like, “I don’t know, let me roll my sleeves up and see how far I can get in this thing.” And if you do that, you can actually get through it. And I just think it’s interesting, because that’s building stuff, right? That’s a good signal of when somebody tells you you can’t write a word processor in the browser and you can’t do collaboration. Do you roll your sleeves up and deal with it or do you fall over?
Lenny: And what was the question again? Just to [inaudible 01:20:34].
Sam Schillace: How many zeros are at the end of 100 factorial in decimal?
Lenny: And then why did it get banned?
Sam Schillace: I think it got known, and somewhere… Actually, the funny thing is, one of their more senior directors, actually he’s a SVP now I think, Dave [Bezeras 01:20:51], I interviewed him when he came in, and I asked him that question. And his response was, “I don’t do math. Next question.” And so I failed him. I was the veto. And they have this policy where… We’re friends now. They have this policy where one veto is actually a good signal. If you’re controversial as a candidate, they would take a hard look at you. And so my veto may have gotten him higher, because I was like, “I don’t know, he wouldn’t answer this question. That’s a red flag for me, so don’t hire him.” And he turned out to be a great person. So it’s maybe not a good question anyways.
Lenny: That comes back to one of your other lessons of the best ideas have some people that are just very anti that idea.
Sam Schillace: Yeah. Yeah. Yeah.
Lenny: It all circles back. Next question, what is a favorite product you’ve recently discovered that you really love?
Sam Schillace: My father-in-law and my brother-in-law worked for the American car companies. And I’ve never had American cars. I just drive Japanese cars, because I grew up in Detroit and just hated that culture. And so recently we bought a Ford Mustang Mach-E, the electric Ford Mustang, and I just love the shit out of that car. I don’t know why. It’s just a really fun car to drive. It’s very surprising to have this American muscle car that I’m just really… An electric American muscle car that I really like. So that’s probably the current product I really am enjoying the most right now.
Lenny: Next question, do you have a favorite life motto that you often repeat to yourself, find useful share with friends or family, either in work or in life?
Sam Schillace: Virtue from error. Yeah, I think that’s… It’s become one for me, and the more I say it, the more I like it. But I just like this idea of, you’re going to fuck up, make something from it and be creative with your mistakes. I like that a lot. So I think that’s at least my current one.
Lenny: And say it again, just so people get it.
Sam Schillace: There’s lots of different ways to say it, but I just like, virtue from error, is probably the cleanest way to say it, or from error, virtue.
Lenny: Final question. Apparently you’re the only person who has sold both a company to Google… I like that you already know where I’m going. Both a company and also 200 pounds of blood sauce.
Sam Schillace: Yes.
Lenny: Both to Google. Tell us the story.
Sam Schillace: The story? So I have a friend who dropped out of the tech industry to start a company up in San Francisco called Boccalone, which was an artisan salumi thing in the Ferry Building. And so he’d make blood sausage and stuff. So there was this wonderful insane chef back in the day when Google had really high-end cuisine in the campuses, probably like 2005 or something, or no, sorry, probably 2008, something like that. And so my friend Mark shows up. This Chef JC had the word foie gras tattooed onto his knuckles. So that’s the kind of guy he was, he was just super awesome.
And so I had Mark show up to talk to JC about buying some of those products, because I was an investor in that company. And he showed up with a bag of blood sausage that was, he’s like, “Here, you should take this one and put it in the refrigerator.” And like, “Why?” He’s like, “It’s dripping.” I’m like, “It’s all right. It’s fine.” He’s like, “It’s dripping blood.” Because they hadn’t got the packaging right. So we showed the blood sausage to JC. He cooked some of it up. It was really good, and he was like, “Yeah, that’s awesome. I’ll buy a couple hundred pounds of it.” And so technically, because I was an investor in that company, I sold both the company Writely and 200 pounds of blood sausage to Google, which I think is a unique accomplishment. And I would just absolutely love to meet anybody who has also done that. We’ll have a party.
Lenny: Did you get stock though, for that blood sausage?
Sam Schillace: No, I did not get stock or anything. That guy was crazy, at one point… Just a quick JC story. At one point Google rented some goats to graze the hillside across the way. And JC was a very non-politically correct, non-woke kind of guy, and he did not like all the sort of attitude at Google. So when these goats were across the hill, he bought a goat carcass from somewhere else and roasted it over a spit and carried it whole through the line for lunch one day to serve up, just to completely tweak people. So that was a different time at Google.
Lenny: The good old days.
Sam Schillace: Yeah.
Lenny: Amazing. Sam, that’s it. We did it. Two final questions. Where can folks find you if they want to potentially follow up on any of this? And then how can listeners be useful to you?
Sam Schillace: Well, I have a Substack, which is Sunday Letters From Sam, that I write a letter roughly every Sunday, that I’ve been doing for about 10 years. Not that particular… Well, I’ve been writing letters to my engineering team on Sunday since I was the head of engineering at Box, and I think that’s 12 years now or something. I just started doing it to keep myself accountable, and people liked it, so I just kept doing it. So now I do it in public. I repost them on LinkedIn. You can find me there. You can message me on LinkedIn if you want to. I’m hesitant to give out my personal email address, because this is probably going out to a lot of people and I don’t want to get spammed.
Lenny: Smart, smart man.
Sam Schillace: Last funny story. At Google, my email address at Google got leaked somehow and a spammer… I was a little but notorious during the early days of Writely, and so a spammer used it for what’s known as a Joe job, where you send something out, fake emails out, with somebody else’s email address as the reply to. So several hundred million emails went out, and all bounced. And so for a while I had my own Gmail front end server that would filter them out for a couple of weeks until that died down.
Lenny: Let’s make sure no one does that to you right now.
Sam Schillace: Oh, yeah.
Lenny: You didn’t answer the final question. How can listeners be useful to you other than not doing that, spamming things to you?
Sam Schillace: Oh, I guess, the thing I guess I’m interested in is people making interesting progress in the direction of that product vision that I talked about. Independent action, the UI part of this stuff, generating UI, consuming UI, all that stuff, I think I’m curious about that. I mean, and interesting ideas. Anything surprising that seems that you’d like to have somebody pay attention to. I entertain weird ideas all the time. I do my best to entertain weird ideas and live what I preach. So if you think you have something that’s really resonating that you think you want to have somebody pay attention to, you can connect me. I’ll take a look at it, I’ll do my best. I won’t look at stuff that’s incremental and boring. It has to be actually interesting and disruptive. So I don’t care. I’m not going to review the 27th memo writing AI chatbot thing that plugs into Outlook or whatever. I don’t care.
Lenny: I think that’s a final good takeaway, is that’s a litmus test for, are you working in something innovative? Which I think has been a great theme of this conversation.
Sam Schillace: Yes. Tell me something that’ll piss me off. That’d be more…
Lenny: Sam, thank you so much for being here.
Sam Schillace: My pleasure.
Lenny: Bye, everyone. Thank you so much for listening. If you found this valuable, you can subscribe to the show on Apple Podcasts, Spotify, or your favorite podcast app. Also, please consider giving us a rating or leaving a review, as that really helps other listeners find the podcast. You can find all past episodes or learn more about the show at lennyspodcast.com. See you in the next episode.
Glossary
| English | 中文 |
|---|---|
| 37signals | 保留原文写法 |
| adjacent possible | 相邻可能(创新理论中的概念) |
| adoption | 采用(指产品被用户接受和使用) |
| Agile | 敏捷式(软件开发方法论) |
| Ashton-Tate | 保留原文写法(公司名) |
| Bard | 保留原文写法(Google 的 AI 对话产品) |
| Boccalone | 保留原文写法(旧金山手工肉制品品牌) |
| Box | 保留原文写法(云内容管理公司) |
| Bucks | 保留原文写法(餐厅名) |
| Calvinist | 清教徒式的 |
| Charlie Munger | 查理·芒格 |
| CI-CD | 保留原文写法(持续集成/持续交付) |
| Dave Bezeras | 保留原文写法(Google 高管) |
| debugger agent | 调试器 agent |
| earnout | earnout(合同约定的业绩对赌条款) |
| Eric Schmidt | 埃里克·施密特 |
| first-principles thinking | 第一性原理思维 |
| G Docs | Google Docs(Google 的在线文档服务) |
| Gemini | 保留原文写法(Google 的 AI 模型) |
| Gmail | 保留原文写法 |
| GPT-4 | 保留原文写法 |
| growth mindset | 成长型思维 |
| IE | 保留原文写法(Internet Explorer) |
| Infinite Chatbot | 保留原文写法(项目名称) |
| Invisible Cities | 《看不见的城市》(Italo Calvino 著作) |
| iPhone | 保留原文写法 |
| Joe job | 保留原文写法(邮件伪造攻击手法) |
| jQuery | 保留原文写法 |
| Kevin Scott | 保留原文写法(微软首席技术官) |
| luthier | 乐器制作师 |
| Michael Arrington | 保留原文写法 |
| Monarch | 《帝王》(Apple TV+ 怪兽剧集) |
| multi-agent | 保留原文写法 |
| Nate Torkington | 保留原文写法 |
| north star | 北极星(指引方向的目标) |
| Office | 保留原文写法(指 Microsoft Office) |
| Palm | 保留原文写法(早期智能手机设备) |
| product-market fit | 产品市场匹配 |
| Roadster | 保留原文写法(Tesla 车型名) |
| SaaS | 保留原文写法 |
| Sam Schillace | 保留原文写法 |
| Satya | 保留原文写法(指 Satya Nadella,微软 CEO) |
| Scavengers Reign | 《拾荒者统治》(HBO 动画科幻剧) |
| Scotty | 保留原文写法(《星际迷航》角色) |
| SEC | 保留原文写法(美国证券交易委员会) |
| Semantic Kernel | 保留原文写法(Microsoft 的 AI 编排框架) |
| Seth Godin | 保留原文写法(美国知名营销专家、作家) |
| Slow Horses | 《慢马》(Apple TV+ 电视剧) |
| Spectre | 保留原文写法(数据库名) |
| Star Trek | 保留原文写法 |
| stochastic agent | 随机性 agent |
| Stripe Press | 保留原文写法 |
| Substack | 保留原文写法(内容发布平台) |
| TechCrunch | 保留原文写法 |
| The Wasp Factory | 《黄蜂工厂》(Ian Banks 著作) |
| VC | 保留原文写法 |
| Warren Buffett | 沃伦·巴菲特 |
| Waterfall | 瀑布式(软件开发方法论) |
| what-if questions | ”如果可以呢”问题 |
| Where Good Ideas Come From | 《好点子从哪里来》(Steven Johnson 著作) |
| why-not questions | ”为什么不”问题 |
| Writely | 保留原文写法(产品名) |
Reformatted by reformat_english.py
如何更具创新力 | Sam Schillace(微软副CTO,Google Docs 创造者)
如何更具创新力 | Sam Schillace(微软副CTO,Google Docs 创造者)
访谈记录
Sam Schillace: 我们往往低估了自己擅长的事情的价值。我们倾向于认为工作必须是令人不快的。所以如果一件事做起来轻松又有趣,我们就不太会认为它有价值。所以我认为很多人会走向这样一个方向——去做那些令人不快的事情,在整个职业生涯中苦苦煎熬,因为那才是成功的路径。但现实是,你应该去做那件让你拿了工资都觉得心虚的事——如果真有那么一件事的话——然后拼尽全力去做,对吧?尽你所能去做。如果你从某件事中获得了乐趣,而人们又愿意为此付钱给你,那就尽你最大的能力去做,倾尽全力去做。如果那件事就是摆弄酷点子、玩转有趣的想法,那就把这件事做到极致。工作不一定非得是苦差事。
嘉宾介绍
Lenny: 今天的嘉宾是 Sam Schillace。Sam 的履历非常精彩,很难简短概括。我试试看。他目前是微软的公司副总裁兼副首席技术官,负责消费产品领域、基础设施和 AI 方面的工作。Sam 最广为人知的,基本上就是他和他的公司 Writely 发明了 Google Docs——这家公司被 Google 收购后,成为了如今 Google Workspace 的基础,而 Google Workspace 目前每月有超过十亿活跃用户。
加入 Google 之后,Sam 最终负责了 Google 的许多消费级应用,包括 Gmail、Maps、Automotive、Groups、Reader 等部分。他还创立过六家创业公司,曾在 Box 担任高级工程副总裁并经历了 IPO。他也在 Intuit、Macromedia 工作过,甚至还曾在 Google Ventures 做过一段时间投资人。正如你所预料的,我们的对话范围相当广泛,但核心焦点围绕创新——如何做大、如何提出原创想法、为什么乐观主义如此重要且强大,以及大量的职业建议。Sam 非常风趣,完全不是我想象中微软公司副总裁的样子,这反而让我对微软更加敬佩。非常感谢 Brett Berson 促成这次介绍。接下来,请出 Sam Schillace。
(赞助商广告已跳过)
Lenny: Sam,非常感谢你能来,欢迎收听播客。
Sam Schillace: 谢谢,很高兴来到这里。
第一份 Google Doc
Lenny: 关于你有一个非常有趣的事实——据说你拥有史上第一份 Google Doc 文件。我不知道该怎么称呼它。就是那份在 Google Docs 还不存在之前就保存下来的最早的 Google Doc 文档。它在现在的 Google Docs 里还能打开吗?里面写了什么?
Sam Schillace: 对,确实还能打开。挺有意思的。其实如果我挪一下摄像头,如果你在 YouTube 上看的话,可能看到后面有个 Writely 的东西。Writely 就是做 Google Docs 的那家公司。是的,它还能打开。不过挺好笑的,因为那份文档的内容基本上就是一堆 DCS 的东西。我们 2005 年开始,用 C# 写的这个东西——这一点鲜为人知——用的是我们自己的……当时还没有云,所以我们租了三台文件服务器,是放在得克萨斯一个数据中心的 Windows 机器,由一个在菲律宾的系统管理员在运维。最初就是这样起步的。
然后当我们转到 Google 时,我们把所有代码移植到了 Java,把所有数据迁移到了 Bigtable。我们没有丢失任何东西,从来没有丢过任何用户的数据。所以数据都还在,而且迁移过去了。那次是一次后端迁移,后来还有一次用 Spanner 做的迁移。前端也重写了两次。所以这就好比——它还是同一份文档吗?我不知道。前端、后端都重写了。内容不多,就是我给 Steve 留了几句话,问他协作功能是否正常工作,我们有没有互相冲突。因为我们当时在测试在同一行打字的算法是否正常工作。
里面还贴了一张《超人总动员》里 Edna 的图片。我不知道为什么。我觉得那张图是后来贴的,可能我后来回去贴上的。不确定什么时候贴的。当时我们肯定是在测试图片功能之类的。遗憾的是,我们已经没有版本历史了,所以我不知道哪些是最最原始的内容。但它确实是现存最古老的 Google Doc,大概是 2005 年 10 月的。
Lenny: 我喜欢这个哲理性的回答——考虑到所有代码都重写了,它还是同一份 Google Doc 吗?
Sam Schillace: 嗯,计算机历史博物馆想收藏它。我跟那些人聊过,他们说:“哦,太酷了,我们收。“我说:“怎么收?“我可以给你做个 PDF,但那就不是那个文档了。我可以把你加为协作者,但请不要编辑它。你怎么收藏这个东西?所以挺有意思的。
Lenny: 是的。得放到区块链上才行。
Sam Schillace: 对,对。如果做成 NFT 的话,可能比文档本身还更”正宗”。有点好笑。
Lenny: 太不可思议了。而且它居然还能打开。这是对……我不知道该说是你还是 Google 的最好证明。
迁移到 Google 基础设施
Sam Schillace: 嗯,我给你讲个 Google 的小故事。当我们迁移到 Google 内部时,我们做得很偷偷摸摸。我们在某个周日把网站设为”维护模式”八个小时,期间一切都是只读的。然后我们迁移了所有数据,搬了所有东西,把新系统跑了起来。三天后,Sergei 在一个会议上跟我说:“你们什么时候搬到 Google 的基础设施上?“我得以告诉他:“哦,已经搬了,这个周末刚搬的。“没人注意到。只有一个德国的博主发现 IP 地址变了,就这些。完全没有人察觉,所以我们做得确实很漂亮。
颠覆性创新
Lenny: 天哪,我最爱这种偷偷摸摸的故事了。希望还能听到更多。我有很多内容想聊,第一个就是”颠覆性创新”这个大概念。我知道你花了很多时间思考这个话题,Google Docs 就是一个很好的例子。感觉微软在这方面的能力越来越强了。就是说去做一些全新的事情,往往是那些人们认为不可能的事。所以我想先问一个宽泛的问题:为什么这件事对你来说这么重要?你为什么花这么多时间思考这个?还有,你发现了哪些工具可以帮助你自己和其他人更有创新性地、更原创地思考?
Sam Schillace: 这是个很有意思的问题。至于为什么重要——我不知道,它就是重要。你身上穿的、吃的、用的、听的、坐着的,所有这些东西在某个时刻都曾经是颠覆性创新。万事万物都是这么来的。我觉得有一个非常有趣的现象:所有新事物在最初都是某种程度上的威胁。我猜,发明椅子的第一个人,多半被部落里的人嘲讽了个够。而这一切事后看来都很明显,对吧?所有事情事后看来都理所当然。
但我认为人性中有一个很深层的东西:如果某件事颠覆了你的世界观,你就会感到威胁,然后你必须做出一个非常 stark 的二选一——要么你错了,要么它错了。而人类是天生的故事讲述者,如果我们有动机,很容易就能编出一套故事来论证某件事是对还是错。所以我管这些叫”为什么不”问题。人们经常提这种”为什么不”问题。一个新东西冒出来,如果你因为某种原因还没准备好接受它——你还没走到那一步,或者它解决的问题你还没有——那它就是威胁,就是令人不快,于是你就会想出一个”为什么不”的问题。
Google Docs、Writely 早期就听到了一大堆这类说法:浏览器还不成熟、人们不会……整个云计算的模型就是,人们不会信任你来存储他们的文件。这太奇怪了。如果没有网络怎么办?飞机上没有网络的故事我从记者那里听了不下一百遍。“要是我坐在飞机上想写东西怎么办?“我说,“我不知道,飞机上很快就会有网络的。“现在确实有了。这些全都是”为什么不”问题。
“如果可以呢”
我觉得更有意思的是那些”如果可以呢”问题——如果这个真的行得通呢?发挥一下想象力。想想看,这条曲线能延伸多远?会带来什么影响?我是工程师,工程师本质上是悲观主义者。有人曾跟我说,工程师天生就觉得世界是坏的,他们看什么都觉得是个需要解决的问题。我觉得这话有几分道理。但我因为悲观而错失的东西,比因为太早乐观而错失的要多得多。
所以我现在有一句口头禅:悲观猜对了,奖赏并没有多少,尤其是在当下这样的时代。乐观猜错了,远比悲观猜对了要好。我不知道怎么说。而且我是个急躁的人,是个有创造力的人,是个混乱的人。我就是喜欢创造、探索、发现新东西。所以颠覆性创新对我来说是很自然的事。但我认为毫不夸张地说——你今早吃的面包里的小麦,如果你吃面包的话,就是一千年前某个怪人在那里摆弄植物,所有人都觉得他疯了,或者她疯了,然后我们就有了小麦,因为某个人……所有事情都是这样的,对吧?所有事情都是如此。
Lenny: 顺着这个思路,我最近正好在写一篇关于第一性原理(first-principles thinking)的文章。我找到了史蒂夫·乔布斯的一句话,提醒我们周围的一切都是某个人设计的,那个人并不一定比你聪明多少。没有理由不存在更好的方式,只不过恰好今天是这个样子而已。
Sam Schillace: 对。另一句我喜欢记在心里的就是:每一个新想法一开始看起来都很蠢。不幸的是,那些真正愚蠢的想法一开始看起来也很蠢。这不是一个完美的筛选器。但颠覆性越强,你就越觉得它蠢。你要留意一些信号——如果他们说这是个”玩具”,或者说它不实用、很蠢、搞不懂之类,这些往往……”玩具”是个很好的关键词。如果你听到有人说某个东西是”玩具”,这往往是一个非常好的信号,说明它其实是真正有影响力的、有威胁性的东西,只是他们想不出比”它现在就是个玩具”更好的批评了。
Lenny: 是的,我猜人们最初对 Google Docs 也是这样看的。就像”哦,浏览器里的一个小玩具”。
Sam Schillace: 是的,这些我们都听到过。就像我一开始说的,你会有一种非常二元化的反应。你要么理解它,然后就会非常兴奋,“太好了,世界会以一种令人兴奋的方式改变”;要么你不理解它,就排斥它。而一个东西越是具有颠覆性,这种二元反应就越强烈。比如 Google Docs,我们在早期遇到了一件让我很困惑的事:有一小群人非常喜欢它。有些人比我们自己还喜欢它。Nate Torkington 在 Writely 那边就是它的超级早期支持者。一开始我并不理解他看中了什么。但同时,也有人恨不得它死掉。
爱恨分立才是颠覆的信号
这种”爱的爱死、恨的恨死”的两极分化,正是你判断自己正在构建的东西是否具有影响力的一种方式。如果你得到的是一条正态分布曲线——大多数人无所谓,有些人略微喜欢,有些人略微不喜欢——那就是一个渐进式产品,并没有真正颠覆任何东西。但如果你看看 ChatGPT 这样的东西,全世界的人要么说”这太棒了”,要么说”这太糟糕了”,中间地带很少——这就是一个非常好的信号,表明它是真正有影响力和颠覆性的。至于它到底是好是坏,那是另一个问题。但它是颠覆性技术,这一点毋庸置疑。
Lenny: 这个框架太棒了。基本上就是:如果感觉大部分人觉得还行、多数人无所谓、极少数人热爱或痛恨,那大概率不是颠覆性的。如果有些人绝对热爱,又有很多人真的恨它——好兆头。
Sam Schillace: 没错。是的,说来也奇怪。这不是投票,对吧?早期我们有——我不知道——几百万用户、五百万用户,Google 内部仍然有高管跟我说这是个愚蠢的想法,应该停掉,我们不该做这个。所以在很长一段时间里,讨厌它的人比喜欢它的人还多。但谁在乎呢?无所谓。只要喜欢你的人没有耗尽,那就没关系。
Lenny: 你还有没有其他例子,是在你做过的产品上或者你见过的东西上,运用这种”如果可以呢”的方法并且成功了的?
Sam Schillace: 说实话,我现在就经常在做这件事。这大概是最近的例子。但我几乎可以指着任何一个产品说,里面都有类似的时刻。不过现在围绕生成式AI,确实有很多”为什么不”的故事。它很贵,会幻觉,你不一定能……它讽刺、随机,同样的事不会做两次。是的,这些都是真实的、需要解决的问题。但我看着它会想,“如果呢?如果我们能围绕它构建软件呢?如果我们能构建比过去更复杂的程序呢?如果我们真的拥有一个可以用来做有意义事情的推理引擎呢?如果这真的是第二次工业革命——第一次我们获得了超越自身体力和水车之类的物理能量盈余,而现在我们获得了超越自身大脑的认知能量盈余呢?”
这是一个真正具有变革性的想法。所以我现在完全处于这种模式中。我认为面对这种显然具有颠覆性的事物,不管对错,这种心态都是对的。Tesla 是一个很好的例子。SpaceX 也是一个很好的例子,人们会觉得”这说不通”,而 Elon 的想法是,“如果火箭可以着陆、可以重复使用,成本变得很低呢?那太了不起了。如果我能解决电池的问题,汽车本质上变成一个软件产品呢?“这些都是相当了不起的”如果可以呢”问题。
Lenny: 是的。所以在理解第一性原理思维实际是什么样的过程中,步骤基本上是:弄清楚你想做什么,找出阻碍你实现目标的杠杆,然后基本上质疑每一个挡在实现目标路上的假设。我觉得 Elon 是一个很好的……经典例子。谈论第一性原理思维不可能不引用 Elon,不可能不讲 Elon 的故事。但本质上就是,好吧,如果我们从零开始做这个,成本会是多少?
Sam Schillace: 我的意思是,“为什么不”——那些其实是最终需要关注和解决的问题。但一旦你有了”如果可以呢”,拿 SpaceX 来说吧,如果你的”如果可以呢”是”如果我能让太空载荷的成本大幅降低呢?如果可以呢?“好吧,那太了不起了。那是一个了不起的世界。让我们看看能不能攻克这个问题。然后你就面对所有的”为什么不”。为什么做不到?为什么没有尽可能地便宜?然后你就可以把问题拆解开来,用这种方式去思考。这是一个很好的模型。
乐观是一种选择
Lenny: 这让我想到另一个我知道你很看重的东西,就是乐观主义,保持乐观。有一种说法是悲观主义者往往是对的。现在世界上很多方面的悲观情绪都在增长,尤其是在科技领域。我知道你非常提倡保持乐观。你能谈谈为什么你觉得这很重要,以及你是怎么做的吗?
Sam Schillace: 说来有趣,这是一种选择。我天性不是一个乐观的人。真不是。我生活中的所有人,如果他们听到这话,都会笑出声,觉得我居然成了乐观主义的倡导者。这就是一个有意识的选择。我觉得悲观不会给你带来太多东西。当然,莽撞也不行。你可以乐观到莽撞的地步,乐观到造成伤害,这毫无疑问。
也许更好的说法是成长型思维(growth mindset),对吧?你想要关注可能性而非局限性,悬置一些怀疑,然后去攻克这些问题。就我个人而言,我觉得自己错过的东西比我保护自己免受的东西要多。如果我把我职业生涯中天平两边加起来,我真希望自己当初思想更开放、更乐观,更愿意去尝试,更专注于可能性而不是问题。
所以这只是我个人的选择,试图把它养成一种习惯。没有更深的原因。我只是觉得这是一个更好的状态,特别是……说来好笑。我当初来这儿的时候是医学预科生。我辍了学,和朋友一起来当计算机科学家。但我当时没这么想。我来是为了和朋友一起在 Ashton-Tate 找份工作,然后花了十年都没意识到自己其实已经在一条职业道路上,一直以为这只是一个临时状态,最终还得回到医学院拿到学位当个医生之类的无聊职业。而做这一行三十五年了,我才意识到,哦,原来我身处的是计算机行业,是一个不断成长、不断发明、不断涌现新想法的技术行业。而在那个世界里,最好的姿态就是保持创造力、好奇心、开放和乐观,去尝试各种事情。
降低试错的成本
关于乐观我还想说的一点是,它和做 Google Docs 这种颠覆性的事情也是相关的。回到”所有好主意一开始看起来都像坏主意”这个观点。好吧,这是一个第一性原理,一个基本事实——真正好的想法会不断受到质疑。那你怎么克服它?一种克服的方式是让自己更容易去尝试。其中一部分是更加乐观,更愿意去尝试;另一部分是让尝试的成本更低。
G Docs 早期有一个故事。当我有了 Writely 的想法时,我的两位联合创始人——他们都是应用开发和文字处理领域的深度专家——说:“浏览器永远不会支持这个。这不是个好主意。别做了。“他们同时是对的也是错的。他们对的地方在于,当时的浏览器确实不支持,甚至到今天也不能说完全支持一个完整的体验。但他们错的地方在于,世界会改变、会进化。而如果第一次实验需要花费很长时间和很多成本,我们就永远不会去做。
正因为我们工具够锋利,我可以说”我们来试试这个”,然后只需要几天就能把它搭起来看看感觉如何——这本身就是一种乐观的形式,对吧?如果你极度悲观,你会觉得”连这都不值得。两天是浪费时间。“所以总要有一点信仰的飞跃。而你要做的就是把这种飞跃的代价降到最低。你要能够快速尝试、快速学习、做这些实验。
很多人以前说过类似的话,但我认为这些要素在我的认知中通过”保持乐观、开放地去尝试”这个理念串联在了一起。因为现实总是和预期不同。你对产品的判断总是错的。这是我的另一条法则——你总是错的。所以你必须去尝试。你必须把它放到用户面前。你必须自己先用。只有这样你才能真正理解它。据我所知,没有人能完全在脑子里设计出产品。
动手试试
Lenny: 太棒了。这里有几条线索我想继续展开。但”动手试试”也是我发现的第一性原理思维中反复出现的一个工具,那些真正擅长第一性原理思维的人都会用这个方法。很多人会说”不行,这不会成功的。“而正像你刚才说的,你实际动手试了之后,往往发现自己完全错了。你有句话,我想是你在一篇 newsletter 里写的,谈到做 Google Docs 的时候。你把它描述为就是——瞎折腾(fuck around)。
Sam Schillace: 是的。差不多吧。
Lenny: 走到某个前沿,然后瞎折腾。这就是策略。
Sam Schillace: 是的。走到前沿,就是——把你的工具磨得尽可能锋利。让自己能做大量低成本实验,对吧?然后就是瞎折腾,看看会发生什么,看看能冒出什么东西来。尽量保持敏锐的观察力。我觉得乐观的另一层含义是一种开放接纳的心态。如果你非常悲观,你可能会错过实验中冒出来的意外结果。你可能会不情不愿地逼自己做一堆实验,但心里想的是,“你知道吗?我讨厌这个。我今天做四个实验,是因为我不得不做,因为我想当创业者,但这事儿糟透了,一切都灰暗无光。“那你就不会注意到:哦,这个东西没成功,但它失败的方式很有意思。我认为当你处于乐观的心境时,更容易接收到这类意外发现。比如,“哦,看看我拿这个能走多远。哦,不行,但为什么不行呢?嗯,这倒是挺有意思的。它在这里断了。”
我们在微软目前的一些项目里就碰到过类似的情况。我们有一个聊天机器人项目,做了一阵子,带有长期记忆功能,可以和它进行长时间对话。效果还行,但某些方面不够好。我们尝试让多个版本协同工作,就像 multi-agent 协作一样。我们给了它们一个白板作为共享工作记忆,来解决这个问题。结果发现这让它们聪明了很多。不知道为什么,就是变聪明了。这就是那种令人愉悦的小发现——如果你处于悲观的心态,可能会说,“嗯,这些效果不太好,放弃吧。“而更乐观的心态则是,“嗯,让我们试试给它们一块白板,就像人一样,看看它们能不能更好地协作。“结果它们确实做到了。这也是那种思维方式的又一个例子。
区分真正的机会与虚幻的热潮
Lenny: 沿着这条线索,这个问题我本来想更早问的——有很多技术会让人变得乐观。加密货币就是一个例子,不是说它一无是处,但很多人变得非常乐观,然后发现其实没有太多可建立的商业模式,于是进入了寒冬。你有没有学到什么方法,能给你一个信号,让你判断”继续做下去,我要对这件事保持乐观”?
Sam Schillace: 我花了很多时间认真思考加密货币,反复审视自己是不是因为它威胁到了我身份中的某一部分才产生那种反应。但我始终没有找到任何有价值的东西。对我来说,关键始终是——必须有一个”如果可以呢”能让我说,“好吧,如果这个成了,它有多大价值?“而对于加密货币,我总是想,“好吧,如果它成了,那我就得在我的个人财务上跑 SEC 合规了。听起来反乌托邦。我不要这个。“作为用户,我想不出任何我认为真正有价值的东西,即使在最好的情况下也是如此。所以我认为这种悲观是有道理的。
我的另一个根本原则就是——一切都要回归用户价值。用户是懒的,对吧?我们都懒。说到底我们并没有那么在乎。没有人在生活中会无缘无故去做一件事,除非它让自己的生活变得更好。没人关心你是否友善、是否nice、logo 是否漂亮之类的。他们关心的是让自己的生活更轻松。我们在某种程度上骨子里都是务实的,所以如果你指不出用户价值——显著的用户价值——那就行不通。什么都不管用。往里面砸多少营销费用都没用,写多少文章都没用,它必须真正解决一个问题,一个真实的问题。我在加密货币上始终没有看到这一点。
Lenny: 是的。所以我觉得这里的教训是——真正搞清楚这个价值是否真实存在,而不是”这听起来很酷”。你会想很多人……”你自己想要吗?“我觉得这是一个很好的检验方法。
Sam Schillace: 我们老拿 Elon 开涮,但我觉得他的很多产品,至少他——他有各种其他问题——但他能清晰地阐述用户价值,即使在早期大肆宣传的时候也是这样。比如 Tesla,对吧?电动车还没成熟的时候,他做了 Roadster 等等。但他至少把这个理念讲清楚了——我们要往车里塞很多电池,它们会是真正的汽车,会有真正的续航里程,我们会解决充电问题。作为终端用户,我的反应是,“好的,很好。现在我有一辆车,像真正的车一样运作,解决了一些问题,而且运营成本更低,因为燃料更便宜。他在帮我解决所有的终端用户问题。“这至少说得通。即使你不相信他能做到,或者不认同他的做法,至少终端用户的价值主张是成立的,对吧?
便利永远获胜
Lenny: 你在 newsletter 里还有一句很精彩的话:“人是懒的。超越’酷’这个层面,去看新工具或新技术能让某人的生活轻松多少。便利永远获胜。“能不能聊聊这个——就是意识到人就是懒的,而这就是关键所在?
Sam Schillace: 是的,确实如此。我觉得作为产品构建者,很难不爱自己在做的事。你做产品是因为热爱,是因为理解某个问题,有时候可能只是因为想要一份薪水。我们出于所有这些原因构建产品,但这些原因对终端用户来说完全不重要。如果说我在产品方面学到了一件事,特别是在消费领域,但基本上也通用,那就是——人对东西就是懒的,除了能让生活变好之外什么都不在乎。这里面比较复杂的是,从这个原则可以推导出两个我觉得很有意思的推论。一个是,我认为产品几乎遵循某种热力学规律——如果你增加一点价值, adoption 就慢;如果你增加大量价值,adoption 就非常快。
我觉得 ChatGPT 就是一个很好的近期例子——它给世界增加了巨大的新价值,然后获得了爆发式增长。然后你会看到很多其他人在做的 AI 产品,不算差,但也不够好,只是在某种程度上增加了一点点价值,缓慢地拖着步子,或者可能会在自身重压下崩塌。这是”用户很懒”这个命题的一个方面。
另一个我觉得也很有意思的是,这同样几乎是物理学层面的。我把它想成熵。有些人对熵感到困惑,会说,“熵不是真的。你看,地球上熵一直在反向运行。生命变得越来越复杂。熵到底是怎么回事?“答案是,“不,你得考虑整个系统。整个太阳系的系统,包括太阳在内,熵一直在增加。我们只是在利用其中的一部分。”
我觉得用户的懒惰也是同样的道理。人们会说,“我关注的这个小功能,我加的这个特性,确实更好了。所以用户应该会采用。“但你忘了周围所有其他的东西。用户得先听说它,用户得在需要的时刻想起来,用户得学会怎么用它,还得建立使用习惯。这些都是成本。更不用说你的功能在实际使用中可能还有摩擦,比如注册流程可能很麻烦。
Writely 的零摩擦引导
当初我们做 Writely 的时候,我们甚至不要求用户输入邮箱地址,因为这太新颖了。我们不想在引导过程中有任何摩擦。所以你直接进来就能创建文档,开始使用,完全不需要告诉我们任何关于你自己的信息。大概打了两分钟字之后,如果你还在那里,我们会非常轻柔地提示一句:“请给我们你的邮箱地址,不需要密码,什么别的都不需要。只是给我们邮箱地址,这样我们可以把这个文档的链接发给你,万一你以后还关心这个文档的话。因为一旦你离开,我们永远不会知道在哪里……我们永远也没法再找到它了。”
但我们对此非常执着,追求尽可能少的摩擦。我觉得在消费级产品领域这是众所周知的——你没有多少时间……你能争取到的秒数并不多。15秒,30秒,对吧,来说服某人这里面有价值。他们不会耐着性子,磕磕绊绊地走过一堆高摩擦的注册流程来使用你的产品。另一方面的道理是,用户只有在他们需要付出的总能量,小于他们获得的生活便利的时候,才会采用你做的东西,而且通常至少要好上好几倍才行。所以它必须让你的生活好很多,最好真的好非常多,比如比你的使用成本好上十倍。
Lenny: 听你这么描述,我想到的是,Microsoft Excel 有十亿个工具栏和按钮和选项,这基本上让 Google Docs 可以以一种简单得多的体验切入。而现在你站在了另一边,这是我没想到的。
从挑战者到在位者
Sam Schillace: 是的,这是一个很有趣的位置。在 Microsoft 工作也很有意思,因为我某种程度上是”敌人”,对吧?因为我是那个给他们制造了一点麻烦的人。所以这方面还是有一些摩擦的。是的,我觉得现在做 AI 也有类似的权衡要做。我觉得有类似的机会。但我们当时做了这个选择……这有点难回忆了,对吧?因为已经18年了,快18年前的事了。在那个时代,Office 是坚不可摧的,对吧?软件必须通过物理方式分发,必须被运来运去,必须被购买和安装。使用起来也更难。
所以交易成本非常高。因为交易成本非常高,买家总是做这样的决定:“我是想要有一千个功能的东西,还是九百九十五个功能的东西?我不知道最后那五个是什么,但我不如全都要。“这就是 Microsoft 的锁定效应,对吧?所以我们做了这个权衡,我们说:“看,我们好用,我们零安装,你完全不用操心,超级方便。而且你还得到一个真的非常有用的新功能,就是互相协作,不用再通过旧文件服务器来回传附件了。但我们要去掉大部分功能,因为我们并不那么在意它们。”
我们实际上去掉得比应该去掉的还多了一些。早期我们会收到各种抱怨,有人想要字数统计,我当时觉得这很奇怪……我以为这个需求会排在列表很靠后的位置。我们那时候没有标尺,没有任何格式设置,没有任何真正的分页功能。我们只有这些基本文档。但页面的字数统计功能呼声很高,当然,用户是学生,是我们最早的采用者之一。所以他们真的想知道自己的论文写到规定字数了没有。
所以就有了和 Microsoft 之间的这种博弈,我们有意识地做了这个权衡。我觉得这几乎就是经典的创新者窘境模型,对吧?我们面对的这个在位者正在渐近地逼近实用性极限,它在不断添加东西。但每当它添加东西,并不真的增加多少价值。然后我们这个小东西进来了,进入了一个他们不太在意、也不太理解的市场,也就是互联网这个东西,这种颠覆性的新事物。我们就像创新者窘境描述的那样,从底部一点一点地蚕食。我觉得 Microsoft 很难回应这件事。我觉得他们花了很长时间才清楚地知道自己要怎么回应。事后来看,他们做得还不错。我们抢了一大块市场份额,但钱基本上都在他们那里。这里的”我们”是指 Google。所以他们撑过来了。他们在应对挑战方面做得不错。现在用户都在我们这里,但钱都在他们那里。不过现在钱也都在我们这里了,我想。
Google Docs 的诞生故事
Lenny: 我想多花点时间聊聊 Google Docs 以及背后的故事。几个问题:从开始做到感觉它行了的这个过程花了多长时间?不管你怎么定义产品市场契合(product-market fit)。
Sam Schillace: 老实说,几乎是立竿见影的。说来也怪,过程是这样的:我有了这个想法,我们搭了这个东西,我们开始一起工作。我们觉得:“啊,这个确实挺……”基本上历史是这样的:我注意到了 contenteditable,所以浏览器可以做一些编辑功能。然后我注意到了 JavaScript,我之前一直没意识到 JavaScript 已经在那里了。我们这个团队以前做过文字处理器,而且做了很长时间。实际上,我的联合创始人 Steve,就是那个文档上的另一个人,早在1987年还是什么时候,写了一个叫 Full Write 的东西,我想可能甚至是1985年,那是 Word 1.0 的直接竞争对手。
所以我们了解文字处理器,于是我们决定试一试。做一个文字处理器是什么体验?而可以在上面协作这件事其实是个意外。它们就是在服务器上的东西而已。我们还没来得及做那个把别人锁出去的功能。所以就只是:“这是一个文档,你可以编辑这两个东西。“对此我们,A,立刻意识到这真的很酷。我们可以同时在同一个文档里工作。然后,B,意识到:“糟糕,我们互相冲突了,因为没有在线状态之类的东西。也没有冲突检测之类的机制。”
所以很快我们就觉得:“哦,这挺酷的。作为开发团队,有这些共享文档感觉很棒,不用把东西传来传去……这很酷。那我们把它做出来吧。“但是:“哦,糟糕,协作是个问题。我们得去解决这个。“然后天真地去解决,发现这是个难题。而要把它做好花了非常久。在那个时候真的非常、非常难,因为我们没有用 operational transform。我觉得那个技术当时还没有完全被发明出来。所以我们做了三方合并(three-way merge),效果不太好,因为浏览器……一个逻辑文档在 HTML 里可以被渲染成不同的样子。没有一种规范的表示方式。
所以你在做合并的时候,字母顺序可能变了,属性的顺序可能变了,树结构可能变了。Firefox 会用一个带单独 BR 标签的空段落,而 IE 会用一个开闭段落标签。所以连树都对不上。所以这是一个非常棘手的合并问题。结果这确实是一个极其困难的难题。但一旦我们看到了协同工作的价值,我们就有了动力去解决它。
幸运的顺序
Sam Schillace: 这件事还有一个有意思的地方,我觉得如果我们先看到了另一面,可能就不会去做了。这又是”为什么不”和”如果可以呢”的一个很好的例子,对吧?我们真的很幸运,先看到了”如果可以呢”的那一面——看到了在浏览器里可以协作的文档有多酷。因为如果我们先理解了协作那部分有多难,而还没有理解那个价值,我们可能就会说:“算了吧,不值得。解决这个问题太难了,这个应用可能也没什么用。“所以我觉得这是一个很好的反面例子,跟我们之前讨论的那个乐观与悲观的视角有关。我们很容易就错过那个想法,非常容易错过。我觉得我们只是运气好,它呈现给我们的顺序恰好是对的。
Lenny: 这真的很有意思——你需要先被”如果可以呢”吸引,兴奋到愿意花上好几年的时间去解决那个问题,因为你对这个”如果可以呢”如此兴奋。我觉得这是一个非常好的……
Sam Schillace: 对,对,我的意思是我在上面花了大量时间。说来也有点好笑,我一直等着有人跟我说:“好了老爷子,别再讲 Google Docs 了,都过了这么久了。“对吧?确实过了很久。已经17年了,但它仍然非常相关。现在大概有几十亿用户了,是个很大的产品。但在过去10到15年里,我花了很多时间思考:为什么那个东西成功了?它成功的因素是什么?我能从中汲取什么教训?围绕它有很多能量,正面的和负面的都有。我在 Google 的第一周,一位高管拒绝给我硬件设备,因为他认为 Google 是一家搜索公司,不是应用公司。当时负责硬件的那个人,真的拒绝为我这个服务提供硬件。我不得不威胁他说要么起诉他,要么把他拖到 Eric Schmidt 面前去,因为我有合同,有合同约定的 earnout 条款。
巨大的阻力
所以这又是一个有趣的经验教训:有时候反对的力量是巨大的。如果我只是一个普通的 Google 员工,带着这个想法,没有法律保护,CEO 也不支持这个项目,它就死定了。它绝对不可能熬过那种负面情绪、那种悲观主义、那个人对此或是不满或是恐惧,又或者是无法想象它的可能性,或者其他什么原因。我不确定具体是哪种。但我们会不断回到”乐观”这个主题。我对此有非常强烈的感受:大多数人之所以做不出真正创新的好产品,根本原因就是这种心态。你就是看不到机会。
努力工作当然必不可少。你可以读到很多最佳实践——做迭代、用户测试、用户访谈、真正倾听用户,以及各种工程最佳实践。这些东西其实相当机械化。一旦你知道自己要去哪里,你就能做到。学习和掌握这些并不难。我觉得真正难的是这种心态——以正确的方式保持开放,理解有些反对意见是好的反对,有些是坏的反对。
我总觉得做产品的人和创业者面临一个非常困难的挑战:你必须对自己的使命非常坚定。我知道我要去哪里,我知道我的使命是什么,我一定要到达那里,因为世界不在乎,它会不断推回你。但你对反馈又必须非常灵活。你很多东西可能想错了。所以你必须把这两者融合在一起。就像武士刀一样——刀背坚硬,刀刃柔软,这样才不会折断。或者反过来,我记不太清了。但作为创业者,你必须完成这件难事,而我认为这才是打造真正伟大产品的核心——找到那种平衡,真正倾听那些信号,对它们保持开放。
Lenny: 还有就是知道要坚持多久,什么时候该转向别的事情。沿着这个话题,你最终什么时候感觉到 Google Docs 实现了产品市场匹配(product-market fit)?从开始做这件事到那一刻,过了多久?
产品市场匹配与被发现的历程
Sam Schillace: 这取决于你说的是哪个市场。G Docs 的采用程度一直让我持续感到惊讶。我觉得我们很早就知道这个方向有东西,大概在头几个月就有感觉了。周围有很多能量。那是一段很奇妙的旅程,因为我们是一时兴起做出来的,作为一个实验。我们喜欢它。然后我们决定去 Google 上打广告。那时候 37signals 是最酷的公司。我们就想:“那看起来很酷。我们就当几个工程师,搞一个小订阅制 SaaS 业务,悠哉悠哉的。所以先看看获取客户要花多少钱。去 Google 上打广告,看看拉人过来要多少成本,然后再决定能不能做成订阅生意。”
然后这就让我们被注意到了。被 Google 注意到了,也被其他人注意到了。我们应该是 TechCrunch 最早的那10篇文章之一,就是 Michael Arrington 写的。还有一个好玩的故事:我当时在 Bucks 跟 Michael Arrington 吃早餐,那是他正在纠结要不要去做一个电子表格的想法,他在犹豫是该去做那个、也许跟我们合作,因为我们当时挺酷的,还是继续搞他那个叫 TechCrunch 的博客。所以 TechCrunch 的诞生可能有一部分是我的功劳,因为我们拒绝了他,说:“你应该去做 TechCrunch。“后来每次碰到他,我都会拿这事跟他笑。
但当被关注到之后,是真的被大规模关注了。有那么几个月我们就是当时最火的东西,每个 VC 都想跟我们聊,所有人都在试图搞清楚这是怎么回事。因为我觉得我们就像是……当你看到线上出现一个点,比如 Gmail,它比我们早,你会说:“嗯,这挺酷的。这是一个有趣的准应用。“但它是那种奇怪的应用,是串行的,你没法真正跟它交互多少。然后你看到我们作为曲线上的第二个点,你会说:“哦,这是一个真正的应用。哦,天哪。那有什么能阻止我们做 Office 的其余部分吗?可能没有。这到底能走多远?”
所以我觉得我们就是那第二个点,证明了确实存在一种完全不同的范式。于是我们很快就获得了巨大的关注度。接下来的过程就是一路摸索:这到底意味着什么?我们需要构建多少功能?真正重要的部分是什么?协作占多大比重?我们花了很多精力在离线功能上,那非常痛苦,而且后来证明根本没那么重要。
文档的未来
现在那个团队花了很长时间去补全我们当初放弃的那些功能,我对这些其实不太感兴趣。我觉得文档的未来会跟我们现在的非常不同。我觉得现在有点好笑——我们花费数十亿美元在 GPU 上,去模拟木浆和金属活字压出来的墨迹。我们在构建线性的、固定的、静态的文档。所以我们用这些聊天机器人做的其中一件事就是,它们同时也是文档。我总说 bots or docs,这两个词我老混着用。所以你会做这些事情……我经常这么做——我们会创建一个新的,它有独立的身份,作为一个独立的文档,然后你告诉它:“我要写一份技术文档,大概是关于这个的。你来采访我怎么样?”
Sam Schillace: 于是它花一个小时采访你,你就得到了一份漂亮的线性文档,可以直接阅读。它的可读性很好,因为是对话式的。但与此同时,你一直在 Spectre 数据库中构建所有这些语义编码的虚拟合成记忆。所以你可以过来说,“给我画个这个的图表。帮我画这个图,按这种方式修改。这是什么?如果我改了这个,如果改了那个会怎样?帮我总结这一部分。” 于是你开始与它交互,而这一切仍然在这个线性文档的底部生成内容。但我们现在正在做的下一步,就是让它变成动态的——你只需要来到一个东西面前,跟它说话、与它交互。
我觉得即将发生的一件事就是——就好比在 Google Docs 早期,人们会说,“如果我没联网怎么办?“我当时会说的一句话是,“三五年后,如果有人递给你一个没连网的设备,你会觉得它就是坏的。“事实确实如此,对吧?如果一个东西没联网,那就是过时的、奇怪的。我觉得我们很快就会对产品中的意图表达和交互性产生同样的感觉。如果我没办法告诉一个东西我的意图是什么,让它以智能的方式自我配置,让它跟我对话——不管那是一个设备还是某个 GUI UX 界面——我觉得都会显得过时。会感觉非常奇怪。
早期《星际迷航》电影里有那么一个场景,Scotty 试着对着鼠标说话,对吧?他就像,“计算机,把那个……”他很生气,因为他没办法跟计算机对话。我觉得五年后我们都会变成那样。我们会觉得非常奇怪:为什么我有了这些应用程序,却不能跟应用程序协作?为什么我不能跟它们协作?这就好像应用程序被锁在文件服务器上,就像 Google Docs 出现之前的年代一样。为什么我不能直接跟它交互,让它按我想要的方式自我配置,按我想看的方式展示数据,让我按我想要的方式构建工作流,替我记住这一切,之后还能带回来,所有这些功能?所以这是很长一段跑题,但我只是想说——你问的是功能和特性的问题,而我觉得我们现在这些功能竞争真的很愚蠢。根本不是重点。我认为文档在未来几年会发生根本性的改变。
Lenny: 我想沿着这个话题继续。不过在此之前,我找到了 TechCrunch 最早一篇关于你们的文章。开头是这样写的:“想象一下 Word,但是是一个 Ajax 浏览器应用。”
Sam Schillace: 哦,对,就是这样。没错。当时甚至都不是 JavaScript,是 Ajax。
Lenny: Ajax,那时候可太火了。
Sam Schillace: 这也挺好笑的,因为我现在跟年轻的前端开发者聊天,我会说,“我不想吓到你,但我写这东西的时候连 jQuery 都还不存在。就是直接操作 DOM,赤手空拳,而且还有各种 bug。“我去 Google 的时候,不得不在 JavaScript 底层写了这么一个小网络栈——理论上你可以在中途打断,可以同时有多个请求在飞,然后你可以打断它们、丢弃它们之类的。
当时的网络栈 bug 很多,我觉得是 IE 的问题。所以我写了这么一个小型网络队列,会追踪是否有请求正在进行中,然后以一种不会把一切搞崩的方式把它们杀掉。这个很难做,因为是一种奇怪的异步编程。而那段代码,我去 Google 的时候,他们让我按照 JavaScript 可读性标准重新排版,而我怎么都没法让它在他们那种格式下正常运行。当时的 JavaScript 编译器有个 bug,空格是有影响的。所以我最后以坏掉的状态提交了上去,拿到了可读性认证,然后立刻修好了它。这也是我们为了让这东西跑起来而搞的小 hack 之一。
Lenny: 太有意思了。
技术探索与北极星
Lenny: 工程师经常受到一种批评——你只是想做有趣的事情,想在找到要解决的问题之前先玩技术。从你这个例子来看,感觉就是你觉得这个技术很酷,先看看会发生什么。你有没有什么心得或建议,关于什么时候可以放心地先玩技术、待在边缘、看看能不能走到某个地方,而什么时候应该尽量避免这样做、先聚焦在一个问题上?
Sam Schillace: 我对此是有罪的。我是说,我喜欢玩这些东西。我倾向于用手指思考,至少跟用大脑一样多。所以我确实认为,单纯地玩技术、搞清楚它擅长什么,是有其价值的。不过我这些年跟团队协作时演变出来的做法是,我会选择我所谓的”北极星”——那些我认为有趣且有用的目标,而不是漫无目的地瞎折腾。什么是用这个技术有可能做出来的酷东西?
所以现在我们在做这些 multi-agent 系统。我们在尝试弄清楚它们能在没有人手把手指导的情况下独立完成多少工作。编程是一个很好的测试领域,因为你不需要太多其他东西——给它一个 Python 环境和一个文件系统就够了,不需要分心去处理连接性之类的问题。
所以我们目前的一个任务是:用 Python 把 AI 写出来。这是一个我可以交给实习生的问题,他们会花一个暑假做出一个还过得去的成果。这是一个你可以期望一个还算合格的程序员基本独立完成的任务。所以如果这个系统有任何独立性的话,它应该能够去完成这个任务。那这个任务本身有用吗?没有,因为我们已经有 AI 了,这不重要。但如果我们能构建一个编程 agent 系统,能够自我监控、自我纠错、自己发现 bug,能够构建大致这种复杂度的东西——那就有价值了。那会是一个很有价值的成果。
这也挺有趣的,因为那个系统已经产生了不少好的洞察。其中一个就是——它相当复杂,而且很难调试。它是一个由随机性 agent 组成的异步系统。要处理的东西太多了。所以我们写了一个调试器 agent。调试器 agent 会监控各种情况,当某个地方出现问题时,它会去搞清楚问题是什么,然后给你一个清晰的解释——你哪里搞坏了,需要修什么。我们还没让它直接去修复东西,因为我们不信任它,但它作为助手已经非常有帮助了。我们还做了另一个,让它给自己写文档。这也是最近做的——直接让它去给代码库写文档,做得还挺不错的。
以目标驱动探索
Sam Schillace: 所以它开始产出一些有趣的东西了,对吧?因为我们有这些北极星来指引方向。我觉得这也许是一个很好的解药。漫无目的地玩技术,不做聚焦,通常不会产出什么特别有价值的东西。但选择一些目标,哪怕这些目标是任意设定的,只要它们是你真正想达成的真实目标,那就是有意义的,对吧?比如你会想,“我能不能让这个东西跑起来?我能不能造出这个东西。“然后花一周时间去死磕,看看能走多近。看看你能从中学到什么关于它为什么难的道理。这大概比”让我随便摸摸 JavaScript”要好得多。
Lenny: 我觉得在大公司里这也不一样,大公司需要你达成某些成果;而如果你只是个刚毕业的工程师随便玩玩,那就放手去干,对吧?最坏能怎样呢?
Sam Schillace: 即便是 Writely 早期,最开始的阶段……我们一开始就有目标。最初的目标就是,“我能不能在浏览器里写一个文字处理器?“这就是我们的问题陈述,就是字面意义上的——“我有 contenteditable,我有 Ajax 和 JavaScript,能不能把它们组合成一个感觉像文字处理器的东西?来干吧。“这有点像一半在瞎玩技术,但也有一半是真正的目标。所以我也说不好。
我喜欢玩技术……而且我觉得很多好的产品想法,其实大多数好的产品想法都是从工程中冒出来的。所以我觉得广泛熟悉工具是有很多好处的,尤其是一些奇怪的、冷门的工具组合往往很有用。如果你懂三样东西……在 Google 时,我是全公司仅有的两个拥有某个语言组合的代码可读性权限的人——代码可读性就是允许你用某种语言提交代码的资格。我同时拥有后端语言(一种监控语言)、中间层语言 Java、前端语言 JavaScript 的权限。没有其他人会去做这种全栈。我觉得有时候拥有这种宽广的视角是很有用的。
Lenny: Sam,语言界的文艺复兴人。
Sam Schillace: 嗯,更像是注意力缺陷多动,不过确实,我对各种东西都会关注。
如何培养提出”如果可以呢”问题的能力
Lenny: 我想沿着”擅长提出’如果可以呢’问题”这条线再往下聊聊。感觉你已经建立了这种能力——或者说也许你天生就有——能够面向未来思考,思考什么是可能的,思考事物的发展方向。对于正在听这个节目的朋友们,你有什么建议能帮助他们在这一方面变得更好吗?因为对很多做产品的人来说,这真的非常重要——搞清楚我们可能走向哪里,然后反向推导。
Sam Schillace: 这是一个很有意思的问题。我其实可能在写一本书,我一直在考虑从我的一些周日信件中整理出一本书来,这也许就是那本书的框架。所以我很好奇我说出这番话之后会被骂成什么样,会很有意思。
我注意到一个很奇怪的现象。我去跟大学里的学生们交流之类的,我发现了一个很奇怪的事情——我上大学的时候,跟像我这样的老家伙们聊天,他们都会说 PC 是个愚蠢的玩具,不算是真正的计算,去用大型机之类的。而我的态度是,“让开,老头,你已经过时了。我要去做这件事,它太棒了。“然后闷头往前冲。
现在我再去跟学生们交流——我最近在密歇根大学做了一次演讲,有一张幻灯片的标题带有调侃意味,教授还真的把它放上去了——因为这一代人非常悲观,似乎不像以前那样充满干劲地去解决问题。我一直在琢磨这件事。我觉得可能有好几个因素交织在一起。
不怕失败、不怕丢脸的勇气
我认为其中一个核心问题是——坦白说,是冒险和接受失败的意愿。我觉得根源就在这里。大家看到大量的经过筛选的内容。这些经过筛选的内容把低概率事件——五西格玛、六西格玛级别的事件——当作常态来呈现。你看到所有人都在创业头三个月赚了一亿美元。所以如果你的创业没有赚一亿美元,你就是个白痴。这是一种。还有一种,你的一切都在众目睽睽之下,所以在那种环境下失败会感觉非常痛苦。
但对于精英学校的学生来说,我觉得也有另一个因素——这些学校很难考进去。我上的密歇根大学,人家就是”你还挺聪明的。密歇根是个好学校。你就住附近,去申请吧。“没什么大不了的。但那些精英学校的孩子,他们的人生是被精心策划过的,一路走到考入那样的学校,对吧?那些学生会说,“我没做体育,没做课外活动,我只是个只会数学的怪宅。“所以我觉得这也是一部分原因。如果你的人生一直被精心规划,花了大半辈子在想”我做的每一件事都必须有理由、有产出”——那你很难放得开手脚去随意折腾、去做一些可能通向意外路径的事情,对吧?这种精心策划是原因之一。
然后我觉得,大概在八十年代中期我高中毕业前后,我们不再让孩子们自己在外面无人看管地跟其他小孩一起玩了。我长大的那个社区全是——那就像是一所小型大学的教职工住宅区,我爸在那里教书,我们就是一帮野孩子到处疯跑。那个地方在道奇汽车创始人遗孀的庄园上,所以我们有几百英亩的沼泽和田野可以到处跑。我们做了一些惊险得令人发指的危险事情,我父母完全不知道。我们真正地去探索、去玩。
我觉得把这些拼在一起来看,人们在实验到失败为止、让自己出丑、提出糟糕想法这方面的能力、意愿和技能都大大降低了。我经常给微软的 Satya 发一些愚蠢的邮件,就是那种”我都不知道他现在怎么看我”的程度。因为我会发各种稀奇古怪的想法给他。我觉得他其实能理解,而且挺喜欢的,因为我觉得通常没人会对他这样做。但我就是那种——然后过一周我又给他发一封邮件,“嗯,那个想法挺蠢的。抱歉。我后来觉得那不是个好主意。”
但我觉得,如果你怕丢脸,你就没法跳舞。如果你怕失败,你就没法成功。事情就是这样。你必须有那种玩乐的精神。你必须有那种”就算搞砸了也没关系,我会继续迭代”的心态。我有一个个人座右铭——从错误中生出美德。因为我是一个动手做东西的人,我一直在做东西。我经常搞砸。我的动手能力不太好,所以我老是犯错,然后我就想办法把错误变成某种美德。我觉得这是一项非常好的技能。我今年给自己找了这句话,我真的很喜欢它,把它当作——我以前从来没有过个人座右铭,我觉得这可能就是我的个人座右铭——从错误中生出美德。
Lenny: 太棒了。关于失败这个话题,我觉得很多人听到这种建议会想,“好吧,我需要多失败。“但真正做到很难。而且很多时候你在公司的绩效表现会受到负面影响。感觉对你来说,似乎是因为你已经做了足够多次——没关系,天塌不下来。我发布这个东西,没人在乎。我给 Satya 发了那个东西,他忽略了,或者他没有。总会没事的。这也许就是关键所在?还是说你做了其他什么事情来让自己能够坦然面对失败?
线性与非凡回报
Sam Schillace: 我觉得,如果你用线性的方式管理事情,你的努力可以得到线性的回报。也就是说,严格管理,不出现任何意外,待在边界之内,缓慢地积累价值,按部就班地玩这个游戏。我觉得你可以有一条不错的、线性的、无聊的职业回报曲线,你会一步步往上爬,花个三十年什么的。我没有这个耐心。而且我觉得,获得非凡回报的方式就是做非凡的事情,对吧?你必须承担更大的风险,进行更有趣的尝试,才能在职业生涯中得到这种非凡的结果。
我总是跟人说这件事,因为……我推广这个理念是因为我观察了自己,思考过我职业生涯中什么才是成功的。这不是每个人都能做到的。我就是这样的人。我从没有真正完全长大过。我就是个怪人。我现在五十七岁了,但感觉自己大概十七岁。我仍然很不成熟,喜欢鼓捣东西、玩各种东西。所以不是每个人都能做到。但说到底,你能在职业上出人头地,是因为你产生了很大的影响力。而你能产生很大的影响力,是因为你选择了一件你擅长的事情,并且以极高的强度去做,最终产生了影响力,对吧?所以我觉得”擅长”这部分也很难。我们倾向于低估自己擅长的事情。我们倾向于认为工作一定是不愉快的。所以如果一件事既轻松又有趣,我们往往不觉得它有多大的价值。
很多人会朝着这个方向走——去做不愉快的事情,在职业生涯中苦熬,因为觉得那才是成功之路。但现实是,你应该去做那件让你拿了工资都觉得心虚的事情,如果真有这么一件事的话,然后拼了命去做。全力以赴地去做。如果你从某件事中获得了乐趣,而人们又愿意为此付钱,那就尽你所能做到最好,拼尽全力去做。如果那意味着鼓捣和玩各种酷想法,那就拼命去鼓捣。
工作不一定要很辛苦。通常确实辛苦,但不一定非得如此。最好的情况是不辛苦。你最有影响力的时刻,就是当你处于那种心流状态、做着自己的事情、乐在其中、不敢相信居然有人为此付钱、不明白自己怎么侥幸过关,但总之就是觉得特别酷的时候,对吧?我觉得这才是对我有意义的职业观。至少我就是这样做的。谁知道呢?有时候全凭运气,所以很难复制。每个人的路径都不同。
Lenny: 太棒了。我非常喜欢这个建议。这恰好是我想把对话引向的方向,所以我很高兴你带我们走到了这里。这让我想到,我正在读 Charlie Munger 的 Almanac,刚刚通过 Stripe Press 出版的。Warren Buffett 和 Charlie Munger 的整个哲学就是,当你发现一个优势时,就下重注。放手一搏,对吧?一年只做一个赌注。但当你找到的时候,全力以赴。不要一点一点地买。我非常喜欢这一点。你说的正是这个意思。
从错误中生出美德的实物展示
Sam Schillace: 有时候做的事情也没什么道理。所以我倾过身子给你们看看,给看视频的人看看。那是我做的一件乐器。它是一件乐器。上面那块是红木,我身后这个带猫眼的,之所以有这些奇怪的猫眼,顺便说一下,这就是从错误中生出美德——我从森林里挖出了这块木头。它在森林的地面上躺了八十年,努力不腐烂,而且做得还不错,因为它是红木。
木头上有节疤。所以那两个猫眼就是节疤所在的位置。这里还有另一个节疤我没能去掉。但那两个节疤我必须从里面刻出来。那是我手工完成的一个非常奇怪的设计。它不太好用。算是有点失败。它的弧度有点太高了,所以有点难弹,因为拨片会碰到顶部,琴弦离顶部有点太近了。
但那是一个实验。我在玩。我想做这个东西。做起来很有趣。它是一个热情驱动的项目。现在就挂在墙上。不是所有事情都能成功。显然我也当不了乐器制作师。所以这更多只是一件好玩的事情。但这是一个很好的例子,说明了那种……我不知道。有时候我甚至不理解自己为什么做某些事。你就是做了。因为你自己觉得有某种意义。
Lenny: 很多人处于相反的境地——他们不喜欢自己正在做的事,很痛苦,但必须得有一份工作。他们需要收入,需要付房租,养家糊口。
Sam Schillace: 我知道,我意识到我说的这些非常有特权色彩,我对此表示抱歉。但从某种角度来看——
Lenny: 但我猜你偶尔也处在过那种境地。对于那些并不处于”我愿意免费做这件事、我对这份工作充满热情”状态的人,你有什么建议吗?你建议尽快脱离那种状态吗?还是说尽量享受它,从中获取最大的价值?
Sam Schillace: 我也为了钱做过很多事。我做过很多纯粹是为了给家里挣钱的工作,那些我并不享受的事情。我只能说,我真的尽可能快地停止了那些事情。不只是尽可能快,因为我内心确实有那种清教徒式的、长子心态——必须供养、必须受苦之类的东西。所以我花了很长时间才意识到,不,其实我非常有创造力。我不必跟其他人一样。我可以有自己的路径,我可以做这个奇怪的工程师。我经常开玩笑说我是个工程师——“二”是质数嘛。就好比说我算是程序员,但不算真正的……我是这种奇怪的非线性的人,勉强能塞进编程的世界里。但没关系。我花了很长时间才意识到我可以那样做,可以接受自己的那一面。
我现在足够幸运,做过足够多的事情,有足够多的人脉和网络,人们了解我是谁、我能带来什么价值。所以我能够去做那些我喜欢做的事情。所以这确实是带有特权色彩的建议,也不是每个人在职业生涯的每个阶段都能做到的。尤其在早期阶段,你常常不得不做出妥协。但我仍然觉得这件事值得留意,对吧?当你在工作的时候,什么让你快乐?什么是那种让你觉得”侥幸过关”而心虚的事情?当我刚开始管理人的时候,我无法理解为什么人们付我工资而我却没有在写代码,因为我所有的精力都附着在”我每天能产出大量代码”这件事上。我问当时的老板,“你为什么对这个这么满意?我什么都没在写。“他说,“我不知道你在做什么。但你去哪儿,哪儿就变好了。所以你继续做你正在做的事就好。“
发现自己的意外之处
Sam Schillace: 那就是这样一个时刻,我心想,“哦,原来我还可以做别的事情,而且还挺有意思的。我喜欢整天和人聊天。太好了。他们居然愿意为此付我工资。他们看起来开心,我也开心。那就让我先全身心投入这件事,看看会怎样。” 所以我觉得,你要寻找的就是这样的时刻——有人愿意让你做某件事,而你做这件事时感到快乐,出乎意料地快乐,或者这件事并不是你”应该做的”那种事。如果你遇到了这样的意外之喜,我觉得这又回到了我们之前谈到的开放心态和乐观精神。你必须对这些时刻保持敏感和留心。我觉得它们存在于每一个人的职业生涯中。如果你留心倾听,你就会发现一些东西浮现出来——你并不认为那是你自己的特质,但别人在你身上看到了它,如果你能以开放的心态接受它,你就能实现这些转型。
Lenny: 你这么说让我想到 Seth Godin 有一条非常重要的建议,一直让我印象深刻:无论你做什么工作,尽量去享受它,把它做到你能做到的最好版本,因为你会更享受其中,最终也会更成功。这就是一种很好的习惯——“我就尽我所能做好这家餐厅的服务生。我就尽我所能做好 Apple Store 的接待。”
Sam Schillace: 完全同意。没错,我觉得这完全正确。不仅如此,找到一种方式把你自己带入到工作中去,对吧?在这个角色中,什么是你能做的、独一无二的、你最得心应手的事情,什么是你能产生最大影响的事情。怎么说呢,我这个人非常……这有点像个笑话,因为我是程序员,但我倾向于是一个非常二元化的人,要么全力以赴,要么完全不做。所以每当我做一件事,我都会以一种荒谬的强度投入其中,无论结果好坏。所以我会找到各种方式把事情做成。如果我不能在某件事上全力以赴并取得成功,我往往会很不开心;而一旦我能全力以赴并成功,我就很开心——虽然周围的人可能就没那么开心了,但事情确实做成了。总之,我总能把事情搞定。
AI 的未来与软件行业的巨变
Lenny: 说到把事情做成、全力以赴以及做真正有趣的事情,你在 Microsoft 负责一些最前沿的 AI 工作。你在 AI 上花了大量时间。我很好奇你对 AI 的看法——你觉得什么有趣,方向在哪里,人们应该了解什么。我想分享你之前说过的几句话,我觉得很棒。第一句是:“AI 不是你产品的一个功能,你的产品才是 AI 的一个功能。”
Sam Schillace: 我很喜欢这句。没错。
Lenny: 另一句是:“把 AI 加入你的产品确实能增加一些价值,但真正具有变革性的、巨大的价值,将来自于构建那些没有 AI 就完全无法运行的应用和解决方案——把 AI 当作一个真正的平台来对待。”
Sam Schillace: 对,我觉得这两句话都非常正确。我现在做的工作是——目前业内谈到做 AI 的人,大多数都是在做模型,对吧?要么是搞新的开源模型,要么是做新的训练,或者把模型做得更大。我觉得这些都是有效的、有用的工作,只是不是我喜欢做的那种。而且已经有很多人在做这件事了。
我本质上是个应用构建者,工具构建者。我不创造模型,我消费模型。我想在模型之上构建东西。所以当我加入 Microsoft,去年九月开始和 Microsoft 一起做 GPT-4 的时候,我的第一反应——在把下巴从地上捡起来之后,我们早期都在做这个动作——是:“好吧,这确实很酷,但从某种计算机科学的角度来看,它就是一个函数,一个随机性的纯函数,接收一个字符数组,重新排列,然后交还给你。这不是一个很好的构建程序的基石。我们需要状态,需要控制流,需要编排,需要外部调用。”
这让我开始往这个方向越钻越深——思考构建 Semantic Kernel,我们已经把它做出来了,然后又做了 Infinite Chatbot,以及我们一直在做的其他项目。我越是思考这些东西,越觉得那两句话说得好。我认为随着时间的推移……实际上我觉得我们正处于软件行业一场巨大颠覆的开端。就像互联网让信息分发变得免费一样,我认为 AI 将让像素变得免费。
现在生产像素是昂贵的——需要程序员,需要大量基础设施,把一个像素放到用户面前是一件困难的事,大量软件正是建立在这之上的。大量商业也是如此。就像 25 年前大量商业建立在信息分发困难这一前提之上一样。但你现在已经可以看到这种趋势了,比如就拿图像来说。两年前,如果你想要一件数字艺术作品,你得去搞到 Photoshop,学会使用 Photoshop,用 Photoshop 来画,慢慢积累技能。生产那些像素需要做大量的工作。现在呢?“我想要一张猫骑着自行车吃香蕉的图。“搞定。那些像素一下子就变得免费了。
类似的事情也正在商业世界中发生,我认为它将开始无处不在地出现。所以你可以推演……这就是”如果可以呢”问题,让我们来问几个”如果可以呢”。如果模型在规划方面变得非常强大,变得更加独立,能够执行更长时间、更复杂的任务呢?如果多模态能力变得非常出色,它们既能消费也能生成动态 UI,就像我之前说的那样呢?如果我们找到了一种好的方式来存储状态——这就是我之前提到的 bots 和文档的问题——如果我们找到了一种方式让你能够高度个性化某个东西,让它真正了解你,而你又信任它能处理你的机密信息呢?
如果所有这些都实现了,你就会花大量时间和那个 agent 交流。想象一下——如果你是世界上最富有的人,你手下有 100 个最优秀的人为你工作,有一个幕僚长,他们不知疲倦,彼此之间从不争执,你想要什么他们就做什么。有这样一支团队支持着你,你还需要用软件做什么?你坐在屏幕前面在做什么?你可能只是在传达意图,然后消费一些娱乐内容,或者是那些意图的产出物。大概就这些。你不会去摆弄那些笨拙的静态应用之类的。那行不通。你只需要告诉你的团队去帮你处理事情。
所以我认为这就是我们的方向,至少在软件世界里。一切会变得更加动态、更加以意图驱动、更加语义化、更加流动,也更加个性化。要实现这个愿景,还有大量的问题需要解决。但对我来说,此刻的感觉有点像看到了当年的 Palm,或者早期的 iPhone——你就是觉得,“好吧,我明白了。手机要变得有趣了。这是一种新的设备。接下来我们需要做大量的工程工作,它们才能真正变得像今天这样有用。“
从云计算到生成式 AI 的范式转换
Sam Schillace: 所以我明白了。我认为软件即将发生根本性的变化。我们刚开始做……的时候也有同样的感觉。再说回 Google Docs,这又是一个教训。又是一次这种品类级别的转变。Writely 一跑起来,我就觉得,“啊,浏览器实际上是一个平台,你真的可以在上面构建真正的应用。我明白了,世界要变了。“而我们还有大量的事情要做,对吧?没有人真正理解分布式系统。没有人知道如何在多个地方同时构建东西,不知道如何处理数据复制,如何处理安全性,各种各样的难题……所有的开发模式都必须从瀑布式,转向敏捷式,再到持续集成/持续交付,所有这些都必须改变,才能充分实现那个世界。
但我记得那是 2005 年,这个过程非常快。那些坚定的拥护者,我想他们都看到了这一点,立刻意识到世界已经改变,出现了一个新的品类。我对生成式 AI 有完全相同的感觉。没错,软件将彻底改变。这些商业也将彻底改变。可能需要 10 年才能真正理清所有这一切,但是,没错,门已经打开了,新房间,新游戏,开始填补空白吧,对吧?我们开始吧。
所以我认为这就是我们的方向。刚才听起来非常笃定,可能比我应该表现出来的还要笃定。我觉得我说的话里,大概至少有四分之一,甚至一半,在某些方面是错的。我们会在过程中学到很多东西。还有大量的工作要做,非常大量的工作要做,还会有大量始料未及的副作用冒出来,就像之前的每一次变革一样,有太多东西需要我们去落实才能让它变为现实。但我认为这是一次巨大的品类转变。我认为这是不可辩驳的事实。
关于 Gemini 发布的媒体反应
Gemini 出来的时候,有点好笑,所有媒体的论调都是,“哦,它跟 GPT-4 差不多。看来 AI 就到此为止了,我们可以回去睡觉了。“我就觉得,“这大概是你能从这个故事里得出的最愚蠢的解读了。“在所有可能的解读中,我觉得那真的是最蠢的一个。它可以说出关于其中任何一家公司的很多东西。可以说出关于这门科学的很多东西。但是”这里没什么可看的”绝对不是其中之一。
如何跟上 AI 的步伐
Lenny: 对于那些不想落伍、或者想为自己的产品寻找机会的听众来说,除了大家总在说的那句”去玩玩它”——比如”去玩 ChatGPT,用用 Bard,试试这些东西”——你有没有什么建议,关于如何思考 AI、如何将它整合到自己正在做的事情中?
Sam Schillace: 嗯,我同意你的看法。“去玩玩它”并不是什么真正好的建议。我认为学习东西最好的方法是,为你正在学的那个东西挑一件具体的事去做,对吧?哪怕是不太合理的……哪怕是一件傻傻的、奇怪的事,对吧?比如”我要搞清楚怎么用这个编程语言画搞笑图片”之类的。哪怕是这种看起来很蠢的目标,选一个随意的目标,然后稍微执拗地逼自己去实现它,这是一种很好的学习方式。
接下来的问题是,“你选的是什么目标?“所以要尽量选那些至少能通向某个稍微有点意思的方向的目标。如果你的目标只是”我要花一个小时随便玩玩 ChatGPT”,那不算什么真正的目标。如果是”我要试着用 GPT 来完成我工作中的这个部分,看看能多接近”,这就更有意思了,对吧?
而且我认为,遗憾的是,这种现象跟早期互联网时代非常相似的地方还有一个——我相信很多人也说过——就是那种跟不上的疲惫感。现在发生的事情太多了。而我认为这也是一个很好的强烈信号,说明确实有非常重大的事情正在发生,就是很难、很难跟上所有正在发生的事情。挺有意思的,我记得——最后再踩一下加密货币的尸体——年初的时候我看到一条推文,有人写说,“AI 圈的人花了一个星期就想出了加密货币花了十年才想出来的那么多用例。“这确实感觉很真实,对吧?就是有太多事情在发生。我认为你只能尽量去跟踪。
当前阶段的特征
我认为当下正在发生的另一件事,让我觉得非常像云计算那个时刻的,是从零到一很难。第一个想法很难得到。理解那里确实有东西存在,这才是真正难的部分,因为你需要运气,需要天赋,需要找对方向,还需要做一些非常艰苦的工作。但是一旦你理解了那里确实有东西——比如云计算的模式是可行的,或者生成式 AI 确实重要、规模确实有效等等——一旦到了那一步,从一到多,所有的优化工作,都是并行推进的,进展非常快。很多人可以做,有大量的能量投入其中,速度就是会非常快。
所以我认为我们正处于这个阶段,就像我们处于一个展开阶段,我们理解了这一步,人们正在尽可能快地填补所有空白。这个过程最终会慢下来,希望如此。看看明年会带来什么。但确实,这是一个艰难的时期。即使对专业人士来说也很难。我认为你只能大量阅读。大量思考。去动手尝试各种东西。选择你的战场。选好目标。在你的领域中挑一个对你来说有意义的目标,然后用某项具体技术去尝试解决那个问题,借此来深入了解那项技术。
如果你想学很多人都在用的东西,也许可以根据它看起来有多流行来选择技术。这些都很好。我觉得就是这些比较朴素的……我不认为有什么秘密,我不觉得有。都是些很朴素的策略,但你就是得选一些东西,做一些功课。学习这些东西没有什么一招制敌的魔法。你只能跑。就像在问”我怎么在冲刺的时候不喘气?“你做不到的。你会喘气的,使劲跑就是了。事情就是这样。
Lenny: 我很喜欢这个建议。而且我喜欢它和你之前谈的所有内容都串联起来了——找到某个人的问题,找到你能提供的价值,然后去想,AI 怎样才能有可能提供这些?我觉得这非常务实。很好的建议。
微软为何能持续创新
也许最后一个关于微软的问题。感觉微软现在所有引擎都在全力运转。感觉它已经成了最具创新力的公司之一。感觉 Satya 现在被称为最具创新力、执行最出色的 CEO,也许没有之一。至少这是外界的感受。作为内部人员,我很好奇,你觉得微软做对了什么,或者说他们的思维方式是什么,使他们能够如此创新,在科技领域发生如此多变化的背景下,依然能保持这样一个庞然大物的地位?
Sam Schillace: 有几个方面。首先,我对 Satya 评价非常高,否则我不会留在那里。我真的非常喜欢他。他从外面看上去是什么样,里面就是什么样。我和他有过很多坦诚的私下对话,也观察过他在会议中的表现。他是一个非常正派、真诚、诚实、精力充沛、充满关怀的人,有很强的共情能力。他的激励方式不具有破坏性。他坚信领导者的职责是提升整个组织的能量,而且他确实在践行这一点。我曾在一些会议上,在场的领域我自己是专家,而我简直不敢相信他参与得有多深入,对一个他明显不如房间里某些专家那样精通的东西,理解得有多透彻。所以他在很多方面都是一位非常出色的领导者。这是第一点。
我认为另一点是,说实话,微软的文化非常谦逊。我是说,从 Google 到微软,这个转变很有意思。我不想做太多比较之类的,但我觉得两家公司确实有相似之处和不同之处。而微软让我印象特别深刻的一点就是谦逊的文化。它一直在做不怎么光鲜的工作,来让业务取得成功。所以它有一种踏实干活、保持谦逊的心态,这让我非常看重。然后,说实话,那里确实有非常多真正有才华的人。
挺有意思的,今年我写了很多专利,因为业界发生了太多事情。我今年写的专利比我整个职业生涯加起来还多很多。我随口提了一下数字,我说”今年我已经写了15个专利”。我跟合作的专利律师说了这事,他说:“嗐,那算什么。移动浪潮那年,首席科学官一年写了700个。“不管你对专利怎么看——我自己也未必喜欢专利,尽管写专利是我工作的一部分——但这就是那里的人,这些才华横溢、在各个层面都拥有深厚经验的人。
我觉得这也是原因之一。那里就是有很多优秀的人。我的上级 Kevin Scott,是我有幸直接合作过的最聪明的人之一,可能也是我见过的最聪明的人。他确实非常出色。他周围的团队也非常出色,Windows 和 Office 的领导层也非常出色。所以就是有很多优秀的人、很好的态度,加上一位好的领导者,我想这就是答案。当然还有运气。运气总是有的,对吧?几年前 Kevin 和 Satya 在 OpenAI 上下了一注,他们在融资方面做了一些非同寻常的事情,对那项技术给予了支持。这也不是凭空发生的。
Lenny: 顺便说一句,那边可没有任何抓马。
Sam Schillace: 那边的事我什么也不知道。听说风平浪静。说实话我尽量离得远远的。
Lenny: 那简直是最低调的创业公司了。太不可思议了。Sam,在我们进入非常精彩的快问快答环节之前,还有什么想分享的吗?有什么想留给听众的吗?
Sam Schillace: 主要是,我想大家应该把以上这些都当作我个人的观点,而不是微软的官方立场。我只是以一个工程师的身份在这里聊天,不一定是作为微软的代表。不过,我也不知道。动手做东西,解决问题,动手做东西。就是这样,对吧?乔布斯那句话——
Lenny: 折腾嘛。
Sam Schillace: 折腾。乔布斯那句话说得对:这个世界是由和你一样的人建造的。就是这样。这是真的。你不需要获得许可,你只需要有精力。
快问快答
Lenny: 说到这里,我们终于来到了非常精彩的快问快答环节。准备好了吗?
Sam Schillace: 大概没有。
Lenny: 很好的回答。Sam,你有哪两三本书是最常推荐给别人的?
Sam Schillace: 我喜欢奇怪的书。有一本叫《看不见的城市》,Italo Calvino 写的,是一部关于城市本质和威尼斯本质的非常优美的沉思录。我大学毕业后第一次去意大利时读了这本书,在威尼斯的时候它把我脑子里所有的回路都炸开了,所以那本书让我印象非常深。另一本我会带着一些谨慎推荐的书,是一本非常强烈且令人不安的书,叫《黄蜂工厂》,Ian Banks 写的,可能是我读过的最诡异、最艰难的书,但它是一次非常有趣的心理深度探索。这两本都是小说。如果你在找商业方面的建议,那本商业类的书大概是 Steven Johnson 的《好点子从哪里来》。我很喜欢那本书,尤其是关于”相邻可能”的部分。虽然现在算是老书了,但我认为仍然很合时宜。他在里面写了一些很不错的内容。
Lenny: 最近有没有特别喜欢的一部电影或电视剧?
Sam Schillace: 我现在最喜欢的,我的罪恶快感,是看 Gary Oldman 在《慢马》里演一个完全令人作呕的过气英国间谍,特别好玩。另外我还挺喜欢那些复古怪兽类的剧,比如《帝王》之类的。看着挺有趣。我不知道怎么说。我在媒体品味方面就是纯垃圾食品级别的。
Lenny: 就喜欢那些恶心的怪物和虫子。
Sam Schillace: 对。
Lenny: 我太喜欢了。
Sam Schillace: 各种打打杀杀、疯狂爆炸的科幻玩意。我这个人不深奥。
Lenny: 对了,你看过《拾荒者统治》吗?在 HBO 上。你会喜欢的。是一部科幻动画,里面有各种诡异、黏糊糊的外星生物,画面极其精美。
面试中的好问题
你有没有一个最喜欢的面试问题,喜欢在面试候选人时问的?
Sam Schillace: 我有一个在 Google 被禁掉的问题,我很喜欢。到现在我仍然觉得这是一个很有趣的问题:100 的阶乘末尾有多少个零?我喜欢它的原因是……对,你做了那个表情,对吧?
Lenny: 准备去问 ChatGPT 要答案了。
Sam Schillace: 嗯,是这样的。不过先别去查,因为我问这个问题的原因是——它看起来像是一个不合理、不可能回答的问题。但如果你坐下来稍微想一想,我不打算告诉你推理思路,但你确实可以在几分钟内找到答案。所以我问这个问题的目的不是要得到答案,而是想看看当我给人们一个看似不可能、不合理的东西时,他们会怎么反应。
有些人会退缩,拒绝去应对。有些人则会说:“我不知道,让我撸起袖子看看能推进到什么程度。“如果你这样做了,你确实能做出来。我觉得这很有意思,因为这就是做产品的过程,对吧?这是一个很好的信号——当有人告诉你不能在浏览器里写一个文字处理器、不能做协作功能时,你是撸起袖子去解决它,还是直接倒下了?
Lenny: 再说一下那个问题?确认一下。
Sam Schillace: 100 的阶乘,在十进制下末尾有多少个零?
Lenny: 那它为什么被禁了?
Sam Schillace: 我觉得这个问题传开了,后来……实际上,有趣的是,他们一个比较资深的主管——现在应该是 SVP 了——Dave Bezeras,他来面试的时候我面试过他,我就问了他那个问题。他的回答是:“我不做数学。下一个问题。” 所以我没让他通过。我投了否决票。他们有这样一个制度……我们现在已经是朋友了。他们有一条规则:拿到一个否决票其实是个好信号。如果你作为候选人引发了争议,他们会更认真地审视你。所以我的否决票可能反而让他排名更高了,因为我说:“我不知道,他不回答这个问题,对我来说是红旗,所以别招他。” 结果他是个非常优秀的人。所以这个问题可能本来就不是什么好问题。
Lenny: 这又回到了你之前说的另一个经验——最好的想法总有一些人极力反对。
Sam Schillace: 对,对,对。
Lenny: 一切都串联起来了。下一个问题:你最近发现的最喜欢、真的很爱的产品是什么?
电动肌肉车
Sam Schillace: 我的岳父和小舅子都在美国汽车公司工作。但我从来没开过美国车,一直开日本车。因为我在底特律长大,特别讨厌那种汽车文化。但最近我们买了一辆福特 Mustang Mach-E,就是电动版的福特 Mustang,我简直爱死这辆车了。我也不知道为什么,它就是一辆非常好开、非常有趣的车。居然会有一辆美国肌肉车——而且是电动的美国肌肉车——让我这么喜欢,真的很出乎意料。所以这大概是目前我最享受的产品了。
Lenny: 下一个问题:你有没有一个最喜欢的人生座右铭,经常对自己重复,觉得很有用,会分享给朋友或家人,无论在工作中还是生活中?
人生座右铭
Sam Schillace: “从错误中生出美德”(virtue from error)。对,我觉得这已经成了我的一个座右铭,而且我越说越喜欢。我就是很喜欢这个理念——你一定会搞砸的,那就从中创造出点什么,用你的错误来做点有创意的事。我非常喜欢这个想法。所以这至少是我当下的座右铭。
Lenny: 再说一遍,让听众记住。
Sam Schillace: 有很多不同的说法,但我喜欢最简洁的方式——“从错误中生出美德”(virtue from error),或者”错误中见美德”(from error, virtue)。
两百磅血肠的故事
Lenny: 最后一个问题。显然,你是唯一一个既把一家公司卖给了 Google——我看你已经知道我要问什么了——既卖了一家公司,又卖了两百磅血肠,两样都卖给了 Google。给我们讲讲这个故事。
Sam Schillace: 这个故事?我有一个朋友,退出了科技行业,在旧金山创办了一家叫 Boccalone 的公司,在 Ferry Building 里做手工 salumi。他会做血肠之类的东西。那时候 Google 校园里还有非常高端的餐饮,大概 2005 年左右——不对,抱歉,大概 2008 年左右。有一位非常疯狂的厨师。我朋友 Mark 来了。那位厨师 JC 把”鹅肝”(foie gras)两个字纹在了指关节上。他就是那种人,超级棒。
Mark 来找 JC 谈采购一些产品的事,因为我是那家公司的投资人。他带着一袋子血肠来了,说:“你把这个放到冰箱里吧。“我说:“为什么?“他说:“它在滴。“我说:“没事,没关系。“他说:“它在滴血。“因为那会儿他们的包装还没做好。我们把血肠拿给 JC 看,他煎了一些尝了尝,非常好,他说:“太棒了,我要买两百磅。“所以严格来说,因为我是那家公司的投资人,我既把公司 Writely 卖给了 Google,又把两百磅血肠卖给了 Google。我觉得这是一项独一无二的成就。我非常期待能遇到同样做到这两件事的人,我们可以一起开个派对。
Lenny: 那两百磅血肠拿到股票了吗?
Sam Schillace: 没有,我没有拿到股票或任何其他东西。那个人很疯狂,有一次……讲一个关于 JC 的小故事。有一次 Google 租了一些山羊来啃对面山坡上的草。JC 是一个非常政治不正确、不太在乎那种觉醒文化的人,他不喜欢 Google 里的那些调调。所以当那些山羊在对面的山坡上时,他从别的地方买了一具山羊的尸体,架在烤叉上整只烤了,然后扛着整只烤山羊穿过午餐队伍端上去给大家吃,就为了气那些人。那是 Google 的另一个时代了。
Lenny: 美好的旧时光。
Sam Schillace: 是啊。
联系方式与结尾
Lenny: 太精彩了。Sam,就到这里了。我们完成了。最后两个问题。大家想跟进这些话题的话,在哪里可以找到你?然后,听众怎样能帮到你?
Sam Schillace: 我有一个 Substack,叫”Sunday Letters From Sam”,大概每个周日写一封信,已经写了大约十年了。不完全是那个……实际上,从我在 Box 担任工程负责人时就开始每个周日给我的工程团队写信了,我想那已经十二年了。最初只是为了让自己保持自律,大家喜欢,我就一直写了下去。现在我在公开平台上写。我也会在 LinkedIn 上转载,可以在那里找到我。如果你想的话可以在 LinkedIn 上给我发消息。我不太想公开我的个人邮箱,因为这期节目会被很多人听到,我不想被垃圾邮件淹没。
Lenny: 聪明人。
Sam Schillace: 最后讲个有趣的故事。在 Google 的时候,不知怎么我的邮箱地址泄露了,有一个发垃圾邮件的人……在 Writely 早期我还稍微有点名气,所以那个人用我的邮箱地址做了一次 Joe job——就是用别人的邮箱地址作为回复地址,大量发送伪造邮件。几亿封邮件发出去了,全部退信弹回来。所以有段时间我有自己的 Gmail 前端服务器来过滤这些退信,持续了好几周才消停。
Lenny: 希望没人现在对你做这种事。
Sam Schillace: 哦,是啊。
Lenny: 你还没回答最后一个问题。除了不要给你发垃圾邮件之外,听众怎样能帮到你?
Sam Schillace: 哦,我想我感兴趣的是——人们在我之前谈到的那个产品愿景方向上做出有意思的进展。独立行动、UI 部分、生成 UI、消费 UI,所有这些方面,我很好奇。当然也包括有趣的创意。任何让人觉得值得关注、令人惊喜的东西。我一直在接纳各种奇怪的想法。我尽最大努力去接纳怪想法,践行我所宣扬的。所以如果你觉得自己在做的东西真的很打动人,想让别人关注一下,可以联系我。我会看一看,会尽力。但我不看渐进式的、无聊的东西。必须是有趣的、有颠覆性的才行。所以我不在乎——我不会去看第 27 个接入 Outlook 的写备忘录的 AI 聊天机器人之类的东西,我真的不在乎。
Lenny: 我觉得这是一个很好的最终要点,也是检验你是否在做创新事情的一个试金石——这可以说是这整场对话的一个很好的主题。
Sam Schillace: 是的。告诉我会让我生气的事情,那更好……
Lenny: Sam,非常感谢你来参加节目。
Sam Schillace: 我的荣幸。
Lenny: 大家再见。非常感谢收听。如果你觉得这期节目有价值,可以在 Apple Podcasts、Spotify 或你喜欢的播客应用上订阅节目。也请考虑给我们评分或留下评论,这真的能帮助更多听众发现这个播客。你可以在 lennyspodcast.com 找到所有往期节目或了解更多关于节目的信息。下期再见。
术语表
| 原文 | 中文 |
|---|---|
| 37signals | 保留原文写法 |
| adjacent possible | 相邻可能(创新理论中的概念) |
| adoption | 采用(指产品被用户接受和使用) |
| Agile | 敏捷式(软件开发方法论) |
| Ashton-Tate | 保留原文写法(公司名) |
| Bard | 保留原文写法(Google 的 AI 对话产品) |
| Boccalone | 保留原文写法(旧金山手工肉制品品牌) |
| Box | 保留原文写法(云内容管理公司) |
| Bucks | 保留原文写法(餐厅名) |
| Calvinist | 清教徒式的 |
| Charlie Munger | 查理·芒格 |
| CI-CD | 保留原文写法(持续集成/持续交付) |
| Dave Bezeras | 保留原文写法(Google 高管) |
| debugger agent | 调试器 agent |
| earnout | earnout(合同约定的业绩对赌条款) |
| Eric Schmidt | 埃里克·施密特 |
| first-principles thinking | 第一性原理思维 |
| G Docs | Google Docs(Google 的在线文档服务) |
| Gemini | 保留原文写法(Google 的 AI 模型) |
| Gmail | 保留原文写法 |
| GPT-4 | 保留原文写法 |
| growth mindset | 成长型思维 |
| IE | 保留原文写法(Internet Explorer) |
| Infinite Chatbot | 保留原文写法(项目名称) |
| Invisible Cities | 《看不见的城市》(Italo Calvino 著作) |
| iPhone | 保留原文写法 |
| Joe job | 保留原文写法(邮件伪造攻击手法) |
| jQuery | 保留原文写法 |
| Kevin Scott | 保留原文写法(微软首席技术官) |
| luthier | 乐器制作师 |
| Michael Arrington | 保留原文写法 |
| Monarch | 《帝王》(Apple TV+ 怪兽剧集) |
| multi-agent | 保留原文写法 |
| Nate Torkington | 保留原文写法 |
| north star | 北极星(指引方向的目标) |
| Office | 保留原文写法(指 Microsoft Office) |
| Palm | 保留原文写法(早期智能手机设备) |
| product-market fit | 产品市场匹配 |
| Roadster | 保留原文写法(Tesla 车型名) |
| SaaS | 保留原文写法 |
| Sam Schillace | 保留原文写法 |
| Satya | 保留原文写法(指 Satya Nadella,微软 CEO) |
| Scavengers Reign | 《拾荒者统治》(HBO 动画科幻剧) |
| Scotty | 保留原文写法(《星际迷航》角色) |
| SEC | 保留原文写法(美国证券交易委员会) |
| Semantic Kernel | 保留原文写法(Microsoft 的 AI 编排框架) |
| Seth Godin | 保留原文写法(美国知名营销专家、作家) |
| Slow Horses | 《慢马》(Apple TV+ 电视剧) |
| Spectre | 保留原文写法(数据库名) |
| Star Trek | 保留原文写法 |
| stochastic agent | 随机性 agent |
| Stripe Press | 保留原文写法 |
| Substack | 保留原文写法(内容发布平台) |
| TechCrunch | 保留原文写法 |
| The Wasp Factory | 《黄蜂工厂》(Ian Banks 著作) |
| VC | 保留原文写法 |
| Warren Buffett | 沃伦·巴菲特 |
| Waterfall | 瀑布式(软件开发方法论) |
| what-if questions | ”如果可以呢”问题 |
| Where Good Ideas Come From | 《好点子从哪里来》(Steven Johnson 著作) |
| why-not questions | ”为什么不”问题 |
| Writely | 保留原文写法(产品名) |
此文档由 AI 分片翻译(translate_long_document)