The engineering mindset | Will Larson (Carta, Stripe, Uber, Calm, Digg)

Will Larson:I think that we often treat engineers a little bit like children instead of giving them the responsibilities and ability to actually thrive as adults and so like the engineers won't want to do that work well that's actually not good for the engineers to kind of be sheltered from what is important and so I actually like one of the I think highlights is that I think we're coming back this moment where we can actually treat engineers like our peers and put them into really senior leadership roles and not have this kind of baseline assumption of like oh we have to coddle them or hide them from the real problems and this is how they're gonna get the opportunity to grow as well

Lenny Rachitsky:Today my guest is Will Larson one of the most requested guests I've had on this podcast Will is currently CTO at Carta he's been a software engineering leader at Stripe Uber and Calm he's the author of two essential books for all engineers An Elegant Puzzle and Staff Engineer and he's releasing his newest book The Engineering Executive's Primer in February he also publishes regularly on his blog at Lathane.com which is a must read for every engineer and Eng leader In our conversation Will shares advice on developing your engineering strategy and strategy in general how to improve the relationship between an Eng manager and a PM how he finds time to write while also working an intense full time job how he recommends approaching measuring engineering productivity how to develop your company values an amazing story about his time at Dig and so much more Will is such a gem of a human and leader and I'm excited to bring you this episode with that I bring you Will Larson after a short word from our sponsor Today's episode is brought to you by DX a platform for measuring and improving developer productivity DX is designed by the researchers behind frameworks such as DORA SPACE and DevX if you've tried measuring developer productivity you know that there are a lot of basic metrics out there and a lot of ways to do this wrong and getting that full view of productivity is still really hard DX tackles this problem by combining qualitative and quantitative insights giving you full clarity into how your developers are doing DX is used by both startups and Fortune 500 companies including companies like Twilio Amplitude eBay Brex Toast Pfizer and Procter and Gamble to learn more about DX and get a demo of their product visit their website at getDX.com/Leni that's getDX.com/Leni Today's episode is brought to you by Oneschema the embeddable CSV importer for SaaS customers always seem to want to give you their data in the messiest possible CSV file and building a spreadsheet importer becomes a never ending sink for your engineering and support resources you keep adding features to your spreadsheet importer but customers keep running into issues six months later you're fixing yet another date conversion edge case bug most tools aren't built for handling messy data but Oneschema is companies like Scale AI and Pave are using Oneschema to make it fast and easy to launch delightful spreadsheet import experiences from embeddable CSV import to importing CSVs from an SFTP folder on a recurring basis spreadsheet import is such an awful experience in so many products customers get frustrated by useless messages like error online 53 and never end up getting started with your product Oneschema intelligently corrects messy data so that your customers don't have to spend hours in Excel just to get started with your product for listeners of this podcast Oneschema is offering a $1,000 discount learn more at Oneschema.co/Letty

Lenny Rachitsky:Will thank you so much for being here and welcome to the podcast

Will Larson:Thank you so much super super excited to be here

Lenny Rachitsky:So many people have suggested that I bring you on this podcast you have a lot of fans out there and I am excited to be digging into engineering topics which we don't do enough of on this podcast so thank you for making time for this

Will Larson:No thanks I hope to be a good early engineering guest before you pivot entirely to engineering at some point in the future wow I love that

Lenny Rachitsky:How cool would that be I was an engineer actually when I started my career how interesting would that be if we come full circle anyway I thought it'd fun to start with just what is changing in engineering it feels like there's been a lot that has changed over the past few years especially from kind of the zero zero interest rate era to today's market which is very different what have you seen most change from an engineer's perspective and then just also what are you telling engine leaders about how to handle all this change

Will Larson:I think it's a pretty a pretty strange time in the market so I think that you know I I started working and you know the right before the 2008 kind of crash and so the first few years there were were not so good when I joined Yahoo there was a layoff basically every like four months there was a layoff of some sort it's pretty chaotic but then we got into the last decade and it was just smooth right and so you know numbers went up revenue went up headcount went up and people started learning how to build really large teams people started learning how to hire a lot when I was at Uber some days I would do like six like interviews back to back I would just be in a conference room and at some point you can't even remember who you're talking to because you talk to so many people one after another after another you just have some like scrambled scrambled notes you're trying to like decode afterwards pretty different now like a lot of engineering managers are spending half their time or more hiring like eighteen months ago and now they're doing like two interviews a month or less maybe zero interviews a month so there's a real shift in just the amount of time people are putting into hiring instead like all these different competencies that kind of become more critical where a really great engineering director might have just been spending their time hiring hiring really well and that could be like a top performer now that person can actually demonstrate like what they're great at and so that person might be perceived as a a low performer if they're not also like figuring out how to like lead the team getting deeper into details and also like you know sometimes getting into the figuring out like what what is the right allocation like what is the right sizing of engineering teams right like this is stuff that we weren't talking about much or maybe if you if you really pissed off the CEO maybe an infrastructure you just grew a little bit slower next year but but know different different ballgame at this point where teams are actually disappearing teams are getting cut down teams are getting consolidated and that that's just something that we've kind of avoided for that zero era that now we become like a core part of a lot of the job

Lenny Rachitsky:It feels like also engineers have they received they used to have a lot of leverage over companies inside companies imagine that's also changed in a big way

Will Larson:I think that's true I think this actually has been bad for engineers in some way like one one of my like hobby horses is that I think that we often treat engineers a little bit like children instead of giving them like the responsibilities and ability to actually thrive as adults and so like oh the engineers won't want to do that work well that's actually not good for the engineers to kind of be sheltered from what is important and so I actually like one of the I think highlights is that I think we're coming back this moment where we can actually treat engineers like our peers and put them into really senior leadership roles and not have this kind of baseline assumption of the go we have to coddle them or hide them from the real problems and this is how they're going to get the opportunity to grow as well but that's like a highlight for me the shift recently

Lenny Rachitsky:I've definitely worked with that and experienced that where you don't wanna piss off engineers and so are you saying that because that's changing leaders maybe don't have to worry as much about upsetting engineers plus you just generally think we shouldn't treat engineers that way

Will Larson:Well yeah I think a little bit of both but I think there there's been you know in this previous era where hiring and retention were kind of like one of the biggest ways you evaluated kind of middle management losing team members was huge huge issue right and so you started to coddle a little bit which is actually bad again bad for the engineers bad for the teams bad for kind of everyone bad for efficiency of the organization but now like something that I love is we get to like give engineers real hard problems and we get to we get to actually hold them accountable and that means we can put them in senior roles so one of the things that I've been pushing on wrote my last book Staff Engineer about like what is the career path for senior engineers one of the challenges is if we aren't comfortable holding engineers accountable because we just want to retain all the engineers that we can't put them in senior roles and so I think we're actually seeing a bit of a shift where we can actually hold them accountable which means you can put them in senior roles which means the engineers can actually get what they've been trying to get the entire time but we haven't been able to because we've been coddling them a little bit too much

Lenny Rachitsky:Okay so there's a few directions I want to go and I'm just going to poke around and see where we go the first is your big advocate of systems thinking we're chatting about this before I think a lot of people have heard this term systems thinking and there's books about it it's like sounds great I wanna be a systems thinker what does that actually mean how do you find you apply it in your work how do people get better at this way of thinking

Will Larson:A lot of the least successful but smartest people I've worked with were really strong systems thinking advocates and so I do want to say like every every kind of like framework has like a lot of downsides there's no framework that people can apply consistently universally and get good results and so briefly on this one I what I see is like often people will find a spot where their system and reality are in conflict and they'll be like reality is wrong and so what's a concrete example at Stripe we worked on incident management and so Stripe pretty important company where our API is available if the API is down you lose money and if you lose a lot of money you leave Stripe because you're pretty upset about that like the number one thing businesses need to do is like to collect money successfully for for the the service that they're selling right Stripe's super important and so we did a lot of analysis on incidents trying to understand why things weren't working know what we could do better but we got like so caught up in the analysis that we sort of lost track of whether we're actually improving things and it took us a while to figure that out because we were so stuck in the systems thinking model and it's not like the team was wrong it's I like was wrong like I was I was caught up in that model myself to realize like hey we weren't actually prioritizing improvements we were just prioritizing measurement and you can't you can't keep measuring know there's like measure twice cut once like sure but you don't measure in infinite infinite times and never never get to cutting you do have to cut at some point to actually make impact but but we just got caught a little bit there and I think a lot of people who get too far in systems thinking make the same mistake where they they think reality is wrong and reality is never wrong reality is always right your model is always wrong if it's in conflict with with reality but that that conflict that that gap is really interesting and that's where you can learn and so I had this model it's it's really clean it represents you know your hiring pipeline that moving through different steps it represents your your incidents and how you remediate incidents it can model almost anything pretty quickly when you get good at it and then understanding how reality is in conflict with that you start to understand where your mental model is wrong then you can go educate yourself and improve the model and just keep doing that and at some point like the model is close enough and you can stop doing that and go like actually do the work so the the biggest thing I tell people is this is a great way to learn but you also have to do things you can't just learn that's not our entire job

Lenny Rachitsky:To make it a little more concrete how would you best describe this idea of systems thinking what's a good way to just like okay I get I get what you're talking about

Will Larson:This is probably a better place to start right versus a rambling anecdote about using it

Lenny Rachitsky:We can we can go in backwards it'll be great

Will Larson:Systems thinking is basically you try to think about stocks and flows so stocks are things that accumulate and flows are kind of the movement from a stock to another thing and so what's a simple example of a stock a stock could be the number of fish in a lake a stock could be the number of people fishing in a lake and so a flow between those two could be the number of fish in a lake will decrease at some rate based on the number of people fishing in that lake so if there's a ton of fishers then the the stock of fish will go down faster if there's only a couple of fish it will go down slower but also then there's these flows which kind of dictate that where if if know we got much more efficient fishermen the flow of of fish out might go down but also the fish do reproduce so then there's another flow going back so based on the current number of fish and the reproduction rate the current number of fishers and their fishing rate you start to see how these can evolve over time and a lot of this so first always recommend thinking in systems by Donatella Meadows really phenomenal book and a lot of her work is also kind of referencing the work in Silent Spring by Rachel Carson which talks about how small amount of kind of carcinogens or something low in the ecosystem or the food chain rather as they get consumed by predators further and further up get concentrated and that that's like a kind of a classic systems thinking problem to think about where you wouldn't think a small amount of carcinogens in like a small fish actually matter but as they go up the food chain they start to concentrate it and an unexpected change can happen

Lenny Rachitsky:So then going back to your example of the hiring pipeline let's come back to that to help connect this definition which I've never heard which is awesome very clear to how you actually implement it say in hiring

Will Larson:The first thing to do is to get a model out there of any sort and so you think about your your stocks and so in a hiring pipeline you might have you know potential candidates and this could kind of basically infinite and then you have a couple of inflows you could have like sourced you could have like outreach you could have referrals and then you have like the you know candidates and those you know how many people get sourced is probably a function of number of sources you have dedicated to a role so you'd have another stock of like how many sources you have and then that would impact the rate going from potential candidates to actual candidates you sourced and then from candidates that you have in that box you'd have like a conversion rate for people who pass the first recruiter screen now move into step two of your process and then from step two you have me be like a hiring manager screen and that would be another conversion rate so you can see like over time how the candidates of the infinite potential candidates like wandering around linkedin posting about their deep thoughts of convert convert into actual people your first your second your third and then as you get deeper into it you start to actually see interesting things and so for a lot of candidate or a lot of pipelines the biggest issue is hiring managers don't want to extend any any offers because the hiring managers can't get to confidence on any candidate and you'll see in this pipeline you'll see a ton of candidates getting to offer stage but almost none of them converting from a potential offer to to actually offer and then you can say hey hey here's the problem you need to go work with the recruiter and the hiring managers on like getting conviction about who they should hire classic problem with early early managers right here's a second problem manager wants to extend a ton of offers they do extend them and none of them actually accept and so that focuses you on the second problem but there's a a third potential world where actually you're just like not getting enough candidates in you're actually doing a great job of making decisions great job of closing candidates there are just not enough candidates coming in and so by looking at this and you can you can build this model then you can go to your your applicant tracking system like greenhouse or whatever and pull the historicals and you can just see how the historicals work versus how you'd expect it to work and you can see the drop offs and this helps you figure out where should you go try to fix things first think we've all worked in companies where you roll out kind of like big changes with no data behind them it's like oh it feels like we're not not hard enough and how we evaluate candidates or something you go change a bunch of stuff but often the real problem might be that the hiring manager is making offer extensions to people who never passed loop anyway it's just the the managers are issuing too many offers because they're they're panicking less true now but a decade or not a decade like two years ago hiring managers panicking to get offers out that was a real thing that that happened a lot and this just helps you take a complex kind of abstract problem and turn it into something can actually like work in a systematic way

Lenny Rachitsky:I feel like product managers will not either naturally do it things this way already because a lot of them think in funnels and it's interesting to hear this version of it of this idea of just like following the stock through the flow of the different steps awesome another thing that I know that you are very passionate about and spend a lot of time thinking about is engineering strategy I think you have this kind of feeling like engineers don't think enough about the end strategy every other function has a strategy and engineers often don't talk about what you find there and what your advice is around that

Will Larson:First I sort of question whether any any function has a strategy in most companies I I my general experience is that there's very rarely a written strategy for any company sometimes like a value statement it's like we build the highest quality products you're like good okay like what is what what what do I do with that you're like build a high quality product you're like okay I don't know what that means engineering often has this problem where I think people will make comments like in their kind of their their culture amp or their quarterly surveys or whatever it's like the strategy is not clear or where's the engineering strategy and the biggest thing I I tell people when they complain and then engineers complain about the product strategy like the pms don't have any strategy or the the business has no strategy and the the reality is like product eng and business always have a strategy it's just often not written down and so I really like the first thing I wanna do is like I push people like not to get caught up on like the fact that there's no template out there which is like product strategy that someone's like forked and like filled in doesn't mean you don't have a strategy you do have a strategy it's maybe like a little bit hard to like articulate and maybe it's like applied inconsistently across different like layers of the product like reporting chain because it's not written down but like it's never true that there's like no product strategy there's always a product strategy sometimes it's bad but there's always one and true for engineering as well there's always an engineering strategy it's just sometimes it's bad and the the first rule of strategy is that if you write it down then you can like improve it right if it's not written down it's to say like if if this pm is just like not a good pm or if they're trying to apply the strategy that they misunderstood or if they actually are correctly applying the strategy from the header product that's just not appropriate to the problems they're working on how do you debug any of that if you have a written document even if it's like not a super compelling strategy at least you can start debugging it's okay the header product should improve the clarity of this document hey this theme actually isn't applying it correctly hey the strategy actually isn't appropriate for this one business unit where it makes sense for the others so that's kind of the first thing I think about but the second kind of big theme on strategy I think about is that often good strategy is so boring it's hard to talk about and so for example on the engineering side of thing a common strategy that's really good but very boring is we only use the tools we have today so you know a lot of times you'll get engineers that want to introduce new programming languages new databases new new cloud providers and a really good strategy for almost all companies is like we just use the standard kit we already have today and at Carta when I joined of the engineers Eric Vogel wrote the standard kit and that is our strategy of the tools we use and you know what some people are really frustrated by that and I feel for them it feels like they're losing control but the power of these boring strategies is that it focuses people's energy on the problems that we value as a company so and it is painful coming into alignment if you're slightly misaligned over time but boring strategies that tell you what actually matters and aligns you with what the company actually cares about are really good for you even if they're a little bit annoying at a time and I can expand on this idea a lot but I won't I won't ramble indefinitely on it

Lenny Rachitsky:Well maybe what might be helpful is what are some other examples of engineering strategies that you've seen just to give people even more just like yeah maybe this should be part of

Will Larson:Our strategy so first what is the definition of of strategy and the the best one I've ever seen is from Richard Remold he he wrote good strategy bad strategy amazing he also wrote the Crux I think came out like this year sometime which which I also read and I think both both great and just like a phenomenal thinker who has so much depth I think one of the challenges of of writing about strategy is you're like I've seen two things and I I write the book but I think the thing that's impressive about Richard Grimalt is he's seen so many different scenarios that he's able to really operate from both like the the particular but also like the general and data set in a really interesting way another book with similar characteristics is how big things get done by I forget the the authors but but really amazing dataset of how mega projects kind of succeed and fail but anyway Richard Vermault definition of strategy is basically three components there's a diagnosis like what is the current status quo like what are the things that are real today there are guiding policies which are basically based on the diagnosis like how do you want to address them and there's actions and actions are how are we going to implement this guiding policies and he talks a lot about actions because he's concerned about this idea of like inert strategy where you have like like we're going to deprecate our old product features we don't use but no one deprecates any of them so he's really concerned about this like non implementation kind of like useless strategy that doesn't do anything on engineering I'm a little bit less worried about that I think strategy is more more interesting on engineering in terms of kind of clarifying how we make future decisions and so what are a few examples of that at Uber we only used our own data centers we didn't use the cloud and this has changed since since the era I was there but in the like 2014 era no cloud and we had a a strict no cloud policy and this was annoying because we had to indent everything ourselves or run copies of everything ourselves but it also meant that we're able to spin up in China in like literally three months and some like surreal stories from that we we couldn't fit our racks into the data centers we had like take the roof off the data center and like lift like the racks in with the crane just like tons of stories and like all this got done in three months and truly truly phenomenal and Uber wasn't in China for very long so in some ways you're like we did all that just to leave but but they left with like a nice a nice steak dd Qiad and not not not a not a bad outcome overall but I think that strategy we run everything in data centers we don't use the cloud meant we were able to move in and out of different geopolitical constraints and companies that relied on cloud presence simply can't they're they're fully constrained by where AWS or Google cloud or Azure have built out so that that's one good example another good example at at Stripe was this idea of we run a Ruby monolith and that's like that's what we did and that's evolved a bit since then there's more there's more Java in in the Stripe of 2023 than there was in the the Stripe of of 2016 or the 2012 or whatnot but that policy really focused the engineers on building innovative features for our users rather than building kind of different tooling to support different programming languages and so in both cases both the Uber policy around running our own data centers and the Stripe policy around Ruby monoliths a lot of engineers hated these but the goal of good strategy is not to appease everyone the goal of good strategy is to dictate how we invest the limited capacities we have or the limited capabilities we have into the problems we care about and I think both of them were really effective towards towards that end

Lenny Rachitsky:A common theme across all these examples is essentially constraint deciding we will constrain our options to move faster and focus on the things that really matter

Will Larson:And solving the constraints is to me I think the most interesting thing that that strategy really does and I think when we talk about bad strategy it usually is because the diagnosis is bad and it's usually because people are sort of exerting what they want to be true on constraints where it's like hey we can do all of these projects at once and often that's just not true but it's hard to convince people that when they're the CEO or they are really committed to believing it but almost all bad strategies basically come down from a willful disbelief of what an accurate diagnosis which means then your your guiding policies are kind of incoherent to begin with

Lenny Rachitsky:Awesome I'm excited for that episode of Richard where we're gonna go real deep into strategy but maybe just as a lasting topic around there if someone listening wanted to get better at say end strategy specifically or a strategy in general is there anything you recommend they do is it read these books is there anything else

Will Larson:If people wanna get good at strategy there's there's a lot of different types of strategy right but here here are some things I'd really recommend first I think the the Richard Rumelt book I think Good Strategy Bad Strategy is probably the right starting point I think The Crux also also quite good but but maybe I would read that one second great overview of how to think about strategy I also think thinking in systems I mentioned that before related to systems thinking a big part of strategy is being able to model the reality so you can improve your diagnosis and so I think that one's really quite good as well if you get into the engineering side of things there's a lot of interesting books here there's technology strategy patterns by Evan Hewitt there's the value flywheel effect by Anderson MacCon and O'Reilly the Phoenix Project by Kim Burr Stafford which is kind of a modern rewrite of The Goal by Goldrat but I think there's like still the missing canonical book is kind of missing on this one so I I took a stab at strategy in my upcoming book which is coming out the engineering executives primer coming out early next year I also took a stab at it in Staff Engineer my previous book but I still think there's like the missing book here so I I sort of am like dreaming of writing like a a engineering strategy book for my my next project although we'll we'll see we'll see if if that actually comes together

Lenny Rachitsky:Well let's actually follow that thread of writing something I was definitely hoping to chat about you write a lot you've written two three four ish you're writing a new book how many books you've published two books and there's a third one coming up

Will Larson:So I have two books the first one was Stripe Press the second one self published and the third one with the Riley coming out in like two months effectively

Lenny Rachitsky:Okay and then also many many blog posts for many many years and I asked a few people what to ask you and this came up a lot Gurgel Oroz and Alex Zhu from Byte Byte Go both asked just this question which is how do you make time to write as much as you do and then I also I'll ask this too and just answer it either first or second it's just what impact does writing had in your career why has it been why do you keep doing it

Will Larson:I feel really strongly that you can write a lot more if you write what you want to write and so this is one of the reasons I don't I don't write for for financial gain and I don't write don't write very much on like on a schedule so I've done like a few pieces for for magazines etcetera but I I find that actually really draining to be you have a topic you have to agree on the topic if if the topic starts like miss missing like your what you want to write about you can't fix it a lot of the time and you're also like on this deadline you're like oh I'm like I'm screwing up I need to ship this needs to be done tomorrow and I just find that really draining we're conversely like when I when I own the schedule when I get to like write about hey like I'm writing on something so I started writing this infrastructure engineering book a couple years ago and I just like it just wasn't there I just couldn't get it to come together and so I just stopped and I'm not writing it anymore maybe I'll come back to it at some point but probably not to me the the biggest the biggest strength of writing what you want is you get to write where there's energy and you don't have to write where there's no energy which takes you like really really negative and this this also ties into how I write books which is that I I basically write the entire thing before I start working with the publisher and if you are I think diligent and good at anticipating what their concerns are gonna be you can you can mostly reuse the content that you're trying to write this is also easier in the sorts of books I write I think harder to do in like a really technical introduction to like MySQL or something you can't just like re sequence those chapters and pretend it's gonna work those chapters like built in like a different way than the sort of like business book that I write does but yeah that writing the stuff that's energizing and just giving up on the stuff that's not energizing that's how I write a lot and how I've been you know I've been writing for sixteen some years and and the way I keep doing it is just by writing what's energizing and what I'm thinking about now and I don't write what I'm not thinking about and I don't write for any audience just write write what is interesting to me and you know that that means some people don't like it and that's great like that's totally fine it's it's not it's not really for them it's for for people who wanna follow the ride and and that that's where I focus

Lenny Rachitsky:This episode is brought to you by Vanta helping you streamline your security compliance to accelerate your growth thousands of fast growing companies like Gusto com Quora and Modern Treasury trust Vanta to help build scale manage and demonstrate their security and compliance programs and get ready for audits in weeks not months by offering the most in demand security and privacy frameworks such as SOC Two ISO 27,001 GDPR HIPAA and many more Vanta helps companies obtain the reports they need to accelerate growth build efficient compliance processes mitigate risks to their businesses and build trust with external stakeholders over 5,000 fast growing companies use Vanta to automate up to 90% of the work involved with SOC Two and these other frameworks for a limited time Lenny's podcast listeners get $1,000 off Vanta go to vanta.com/lenny that's vanta.com/lenny to learn more and to claim your discounts get started today okay there's a lot more I wanna dig into here how many posts have you written do you think over the sixteen years

Will Larson:I would guess about a thousand like that that would roughly be my my assumption I I think there are a few years where I wrote you know hundreds of posts and so if you do that like three years it's not that hard to get to 1,000 from there

Lenny Rachitsky:That's incredible especially because you've had intense jobs for all of those years or most of those years very high pressure fast growing hyper growth companies somehow you find time to work so first let me just double click slash cosign your advice here around paying attention to what gives you energy and working on things that you're actually curious about this is exactly the advice I give to people a lot of people start this like full time writer creator life and they're like what do people want what do people want me to write about what's popular what's gonna inspire go viral and that's easy to do like a couple times but then you end up creating this job for yourself that you don't want don't be spending all your days writing about AI if you're not that excited by AI or whatever's hot these days and I find that what I find is important is almost like 90% of what you have to what you write has to be stuff you're excited about and then maybe there's a bit of here's what I know people really want here's what I know is gonna do really well because otherwise you just burn out you create a job for yourself that you don't want

Will Larson:Why would you do that yeah I just a 100% agree with that I think the the other thing is that everyone converges on the same thing that they think people want so it's like it's crypto two years ago it's like AI right now or it's like counter AI like AI is gonna like ruin the world it it's just like it's hard to say something very novel because one like everyone's trying to like say something about it two like it's almost certainly not what you're that knowledgeable about where if you just stick in your lane I think the biggest risk to writers is quitting a little bit like the forty year career idea the biggest risk to content creation of any sort is quitting soon because you get burned out the biggest risk is not that you grow too slow initially there's this there's always a sense that like you've missed the wave like it's too late to join Substack there's already the top writers are already there you'll never be a top writer it's too late to podcast there's too many podcasts you'll never make it it's too late to join Medium like you'll never make it there's too many Medium writers but it's just like not true if you just like keep writing good stuff you'll build an audience over time and you can take that audience from platform to platform what really matters is finding something you can actually keep doing for the next decade that's way harder than doing it for one year

Lenny Rachitsky:We have the same exact advice on this this is exactly the things I tell everyone when I joined Substack thought it was too late I was like man it's over and when I started this podcast like oh man there's a billion podcasts how is there ever gonna work so I so agree and I also so agree on the fact that this whole thing is such a it's a long game there's a lot of people I always say it's easy to start a newsletter hard to keep it up nobody actually keeps it up because people are gonna come and go the thing that really separates success from not success from failure is just people that can keep at it and there's not like an endgame to this right it's an infinite game and it's about being able to sustain that over long term yeah I totally agree with that unless there's so many and then all of sudden what happens is the bar just gets higher which is good because then people get better stuff and that's fine and and that's happening anyway just the bar continues to increase because there's more and more content out there and to me that's like the ultimate thing you gotta get right is just the bar you just gotta be at a high bar for anyone to care about anything you're writing about and to your point to do that well you have to actually be excited about writing a bit about it and have background and have something to contribute the way I'm just ranting here but the way I think about this is you need to add something new to the conversation for anyone to pay attention because there's so much fluffy superficial stuff and to get anyone to care is you need to say something new that no one's heard before or share new information they haven't seen anywhere else

Will Larson:So yeah the head of the our chief product officer Bruce Ali and I like spend a fair amount of time like calibrating together and making sure like again there's cases where there's like an exception right because there's like clear clear issues happening for someone but on average that is what's happening and I think people know that's what's happening because we've told them that and I think that that's pretty powerful

Lenny Rachitsky:That is so interesting I've never heard of that approach that is definitely solving that problem of EMPM or

Will Larson:Yeah the incentives the incentives are shared now which doesn't which isn't perfect it's still hard to balance them they can still like make the wrong trade offs but at least they understand the incentives are shared which I think is a pretty powerful idea

Lenny Rachitsky:That is really interesting I imagine some companies might even want to include design managers in that take it another step

Will Larson:You know the role and the the primacy of different functions in different companies like vary so much that it's hard to have like a one size like you could also imagine where you want like a staff engineer and that or not and so I think very company specific but but yeah I think design could absolutely be involved particularly for a design led company like an Airbnb or something like that

Lenny Rachitsky:Wow so interesting maybe just as a final thought there if a PM is having challenges with their EM what do you think PMs maybe don't think don't realize their engineering managers are finding important or maybe are stressed about that they're just like oh wow I never thought about that

Will Larson:I think one of the biggest challenges I've historically seen particularly in the last decade is this idea that engineering managers have the job of giving their team interesting work and I think that can put you often see this in growth teams where the growth team's like hey we just need to do a ton of experiments and engineers like I wanna build something brand new and the the engine managers in between those try to figure out like need to ship 50 experiments that are pretty boring and they wanna do something brand new like I don't know how to do solve this and so that it's a tricky a tricky moment and good good EMs kinda like find the way to balance but that that's like the biggest source of kind of ongoing friction where the EMs have been told by their teams they need to do something that the PMs just have no visibility into and it makes the EM seem like totally unreliable partners because they're trying to solve these in little bit of these invisible constraints and that's where I think pushing further to understand like hey like keep prioritizing this like rewrite into a new programming language to me that seems like completely idiotic thing to be doing like what's going on and then once you understand you might not agree with them but at least you can have an honest conversation on how to navigate those constraints versus just like man you won't believe what my EM partner did today like this this bozo did like blah blah blah and having like sort of this like victim villain kinda mindset about your peers

Lenny Rachitsky:An adjacent topic that I wanted to spend some time on is measuring engineering velocity productivity I think it's probably one of the most common and also maybe the most annoying questions engine leaders get is just how do I know if my engineers are moving as quickly as they can how do we help them move faster what advice do you give to engine leaders for and engine teams just for how to measure productivity well

Will Larson:This is a question that's coming up even more right in a moment when we're kind of reducing a lot of the size of teams the industry when the the venture capitalists that are on the board for these like venture backed companies are pushing on the efficiency of engineering engineers are trying to figure out like how do we represent this how do we prove that we're appropriately productive for the amount of headcount and funding that we have as an organization and man that's hard and so the first way that people kind of focus on trying to answer these questions is just like benchmarking by like the amount of funding that you have and that's pretty straightforward to do it's a mechanical exercise you look at like you get a data set from your venture capital funds or whatnot and you figure out like okay how much should we be spending an R and D how much should we spending engineering how much should we be spending on you know infrastructure engineering in R and D and you can like benchmark this all out and figure out what the correct numbers are there the problem is this is like a very mechanical and operate like insightful driven way like it'll get you a defensible answer it's like the old like no one gets fired for buying IBM which definitely hasn't been true in my my career ever but you know this idea that if you just have the right benchmarks like DCs won't judge you for spending too much in engineering but this doesn't actually help you get to the right place it just helps you get your board to be less angry at you that's which is useful because it's hard to do good work when your board is angry at you but it's not useful in the sense that it doesn't actually help you run your your organization effectively so then there's like the the much harder and mediator problem of how do you actually know if your R and D team or engineering team is like effective and what I find is a couple of things first if you're a good leader and you talk to the engineers they will tell you like the engineers know if their teams are effective or not and if they're not they they'll also tell you why not and their diagnosis can be wrong but there's like a a crumb you can start like picking up and you can trace trace the crumbs to figure out what's wrong like often you'll have more experience to analyze the the the complaints to figure out what kind of the contributing causes are to to them but but yeah if you just go talk to the team on an ongoing basis you will know if they're effective or not and you can go work to to solve those specific problems but again you can't tell like your board oh like it's fine I talked to the teams that they're good my intuition spot on because how do they know if your intuition is good or not right they're they're dealing with like a huge portfolio and some of their leaders are talking to are good and some of them have terrible intuition how do they actually assess I think it's tricky and what I've what I've tried to do is basically two things one aligning engineering evaluation to the business and product goals so I want us to be wholly accountable as the product goals not a well we're we did a good job products like screwing up over there obviously lot of companies find comfort doing that but like really we're here to support the products to support our customers in doing like something interesting we're not here to like build novel systems unless it supports the customer and the end of products and so first try to align heavily there second I think just showing the roadmap of the valuable things you've done in the last six months is really powerful because I think sometimes people are like don't have anything to put there and you're like yeah that's that's a real issue or if you have a ton of stuff to put there that's great and I I really find that if you just commit show the number of meaningful needy things that have impact that you're doing and you can explain the impact people will kind of step back and give you space if you can't populate that list people will have concerns and like rightly so like they they should be they should be concerned about that

Lenny Rachitsky:Is there any metrics or tools or anything like that that you find useful too because these are all awesome piece of advice but I imagine everyone's always just like give us this number we're tracking give us this dashboard see what engineers doing

Will Larson:So one of the most influential books in the last decade in kind of software engineering leadership and and kind of infrastructure is Accelerate by Nicole Forsgren Jean Kim and I believe there's a third author on that one I'm forgetting right now really really phenomenal book and it kind of comes up with these like four four metrics it comes up with like lead time comes up with like incident remediation time comes up with failure rate and the fourth one of some sort and there's like at least 50 different startups out there that are selling you dashboards kind of instrument these pieces of data and they want you to just evaluate your team on them the challenge is these are really good diagnosis metrics and so hey our deployments are slow why is that how do we speed them up but your deployments being slow doesn't make you a good company or a bad company it just tells you where you should focus on improving it doesn't actually change how how how you are and similarly if your lead time is quick or slow it tells you where you should invest but doesn't actually tell you if you should like fire your engineers or something like that that's like way way more detail specific so people do like to see these metrics just like they see uptime metrics a lot of engineers report on like sprint points or stuff like that to their board which are just like totally totally fake thing to be like reporting on but people get some comfort on it so my my biggest thing here is when people measure things this isn't an engineering only problem but when people measure they they take on the perspective of an expert and they can tell you why not to measure everything they can tell you why every measure is wrong or inaccurate and they rule everything out so they measure nothing and they go to someone who's not an expert and like well actually there's no accurate measure to give they're not an expert is like you're you don't know what you're doing and so you just have to get comfortable measuring something that's not perfect but you can actually measure and reporting on it and then the measure that's imperfect as people ask questions that's like an an opportunity to educate people on like why the measure is imperfect what are some things it misses or kind of the lies from the conversation metrics are about educating the people consuming the metrics about the reality of the rich data underneath they're not about this perfect data set that shows everything starting with something mediocre and the DORA metrics are really helpful for diagnosis but if you have to they can also be a good enough starting place to start reporting to your board or your CEO or to the other executives and then you're like oh there's all these problems with them yes there are all those problems with them but that's this place you start and you educate people up from there to help them understand the nuances and that's how they become more sophisticated understanding engineering not by refusing to give them anything they can possibly measure

Lenny Rachitsky:Awesome I'm glad that was your answer because we had Nicole on the podcast and she talked through DORA and all the frameworks that she recommends and she even actually shared some benchmarks that she points people to that give you some sense of just like are you in a good place roughly or not so we'll point people to that episode to dig deeper awesome I'm glad that you're a fan okay just a couple more questions before we get to our very exciting lightning round one is around values company values org values you have some really good advice for people for how to think about coming up with values what do you share what do you recommend to people that are trying to figure out what values they should define for their org and their company

Will Larson:I mean values are really interesting right and different companies talk about values in different ways I once worked at a company where the execs went to visit the Facebook campus they saw the values written up on the wall and they took the Facebook values and wrote them up on our walls and and that that didn't do a whole lot it it it maybe undermined people's confidence in the critical thinking of the executive team that just took these written up on Facebook walls and replicated it but I think those values did work well for Facebook and those values were meaningful for Facebook and so it's first thing you can't do is just like steal values value cargo culting man users first great Amazon value right a lot of companies aren't users first and that's okay but what's not okay is when you put hey we're users first and then you actually show like the decisions you're making and you clearly aren't users first and so one of the things I think about is just like honesty and so good values have to be honest and so any value can be honest or there's no universally honest values right like you can say something like we're thrifty or we can say something like we spend as much as we need to get the best value those are totally different and good companies are run both ways so the first rule I think about a lot is honesty you actually do what you claim you do and the value the second one is like applicability like you have to have values that you can actually figure out how to apply to to your work and so one of Stripe's values was no longer a value believe but it's like optimized globally and so optimized globally is a really interesting problem because sometimes you'll have something you wanna interesting value sometimes you wanna do something you know like hey I want to introduce a new programming language because that's better for my team but like for the organization overall this is actually much worse for the organization so I'm not gonna do it Uber didn't have this as a written value but implicitly Uber's Uber's value was do what's good for your team and ignore everyone else because that will slow us down and so the two different companies had opposite values but they're both very applicable like how should we navigate decisions should I optimize for my team or for the organization and then so those are applicable to real problems and they were honest where Uber was just like don't worry about other people like make it work for your team and that's how they move so fast because they just didn't worry

Lenny Rachitsky:Wasn't there a value of note of toe stepping encouraging toe stepping something like that

Will Larson:Let builders build toe stepping there there there were a number of values that that could be interpreted in different ways and and sometimes they got weaponized in various ways as as as all values do but but these these are both interesting in different ways and so number one is honest and two is applicable three is i i think the last thing for a good value is this idea of reversibility mhmm so there's some values that aren't actually usable and so here's a good example we build good software okay but why would you ever not build good software that make sense or we solve customer problems that matter good would ever say they're what company doesn't think they're solving customer problems that matter and so there are certain values that just you can't apply i think of these as identity values these are really just you describing who you want to be we care about our customers but who would say they don't care about that there are certain values that i think of it just like identity values and and they're not like they're not wrong to have identity values they just aren't very useful you can't actually use them for anything and so i just always push people not to spend too much time on these because they they feel good when when you're an executive team kind of debating like what are these identity values it's like we're we're kind to other people or you know sure that that sounds good like we're a family like sure that that sounds good that one i guess is a little bit reversible as you know there's netflix which is like we're a team like a sports team we're not a family

Lenny Rachitsky:And so a little

Will Larson:bit reversible but but not perfectly but yeah but these are the three that i found really useful for any value like is it honest is it applicable and can you reverse it and if not it's probably actually not helping the team make decisions

Lenny Rachitsky:These are great it reminds me a lot of i was there during airbnb's period of coming up with values something that i would maybe add and maybe fits into one these buckets is it needs to be clear who doesn't like there needs to be a group that doesn't quite fit because if everyone fits then some you're not doing anything useful here what's the point which feels weird to say it like right why would not not everyone fit in our big group of awesome company but it's clear like who is not a good fit who doesn't belong basic it's kinda like a cult a little bit like who's not in our cult who doesn't belong

Will Larson:but but i agree like if if if it doesn't apply to anyone then like why bother saying it it's like it doesn't doesn't mean anything and you could say it's actually like a hiring filter where there are people who you've explicitly chosen not to hire because this wouldn't apply to them then i think it's useful because it helps you actually figure out who to who to like bring in but if it doesn't apply to anyone you're hiring or anyone that you have in the company then it just sorta like isn't worth having because you already have too many values you are already trying to get rid of values so you have like 17 and you need to get down to like four where people can remember them so if it doesn't apply to anyone like why why bother having it at all

Lenny Rachitsky:yeah like integrity is a common one integrity like everyone has it nobody wouldn't want integrity like what does you need

Will Larson:we're the non integrity company like we we're the company that like thinks integrity is bad like that that that's like not a real thing

Lenny Rachitsky:the other one else i'll add to is honest so at airbnb we had six values initially one of them was simplify and a year or two later everyone just realized we're not actually good at this we want to simplify but we're not great at this skill and value should describe who you are not who you want to be and aspire to be so they cut two values including that one and there's like let's just do these four because this is actually who we are let's be honest with ourselves okay final question i wanted to visit failure corner something that i've added recently to this podcast where people share a story of failure and you have this amazing post about your experience with dig and the rewrite that you all went through i think it was the version four of dig can you just tell that story and what happened and how much of a mess it ended up being

Will Larson:yeah big v four is is i mean still still something i have a lot of like fond memories for there there's one picture that i that i've kept and it's a picture of a lot of the engineers around this table in the middle of this giant office and they're serving sushi we had waiters caterers come in that day they're serving sushi we have plates with champagne flutes on it there was a full bar and we're all around this table because the site's not up and so dig before essentially what kevin rose or the board or some combination there realized is that dig was losing to the social networks and that this idea of aggregated news was gonna be outcompeted by the the twitters the the facebooks etcetera if we didn't find a way to move to have a social component for it even outcompeted by reddit long term was kind of the the fear although at the time that that was far from from obvious and so we needed to move to support kind of social functionality and the previous version we simply couldn't get it to work and so the decision that was done like i think two and a half years before i joined and and this this shipped about six months after i joined was they need to do a complete rewrite in order to get there this is a decision that never works out for anyone and so i think like as someone's more experienced i could have predicted this wasn't gonna work out but i was earlier in my career my my pm counterpart at yahoo das gopinath he he went to to yahoo and he's like come to dig worst case you'll make couple 100,000 in a year worst case probably really great outcome anyway that that's not what happened the worst case was a little bit optimistic but but so we go and you know the the ceo got fired two days before i joined so the current ceo left and then kevin rose came back for for about six six months something like that and we're just on this death march trying to get this thing out and so we we push really hard this is before the cloud for the most part so we we wiped all of our pretty much all of our existing servers to reimage them to the new the new software we try to bring the site up and just keeps crashing and so it basically takes us a month to get it fully functional again and so that day sitting around that table with champagne and sushi that's just like day one and and by you know thirty days in most people aren't even trying to get the site back up anymore there there's maybe like five of us who are still trying and you know we did and and i think like that that was like a really powerful moment for for me and i think in the first two days like myself and rich schumacher like one of the other engineers we had to write like a caching system from scratch which got us like half the way up really a terrible way to do software on a side note like i'm not recommending this to anyone this was like a series of anti patterns cludged into like a launch but we we got it partially up but we had to restart it every twelve months basically every server sorry every twelve twelve hours every server had to be restarted even with the caching mechanism and then about three weeks after that i finally figured out what the core bug was that was bringing us down every twelve hours and it was this incredibly simple issue that had just been hard to debug basically related to the way that python initiates variables used as default parameters and it's something like super super silly and we just had someone who hadn't written python before who was working on the api code so they didn't realize this gotcha then no one else caught it when when it was reviewed and and it just took a long time to debug because it was like such a non obvious it didn't break anything it was just doing a lot of extra extra load on the servers and and we finally we finally figured it out and and it was just really remarkable experience pulling through and and you know what the company still still went to zero and so we we had this bad launch think we did this heroic heroic stretch to get it working a couple weeks after that like a new ceo came in did a round of layoffs this is back i think like 2032 the team nine months after i started was down to like 30 people from about a 100 and it just went it went downhill from there from a from a business perspective but we launched a lot of functionality has really learned just a tremendous amount and it kind of shaped like what i think about in terms of earlier career getting learning and going into a company that is maybe having a rough time like i became a manager like two and a half years into my career three basically running the entire engineering team there because everyone who had a lick of sense quit or got laid off and it was just like complete idiot me like trying to like be the manager for for the engineering org it wasn't qualified and no one would have given me that job but i was the only one dumb enough to take it at that point and and i learned so much and i really like that that's like the the kernel that like turned into like my entire career was that opportunity even though at the time that was it was pretty pretty grim

Lenny Rachitsky:that's an amazing story i feel like a lot of these experiences were in the moment it's just like what is going on this is so bad and hard end up being the most interesting in looking back end up being the most biggest teaching experiences the ones you like bond over with people you work with like apple it always comes to mind where it's just like steve jobs has drove people like crazy then they look back that was that was the best moment of my career

Will Larson:you would never like voluntarily take on a lot of these really challenging things but sometimes like when they show up like you're with a group of people you really respect you you love working with and you like wanna overcome together and that that's like that's really powerful experience even if like uber china was similar where like if someone had been like hey do you wanna go work on this uber china migration i would have been like absolutely not but like no one asked they're just like get this done and so so we did and i think these things are like pretty pretty remarkable and just

Lenny Rachitsky:to be clear so dig was down for a month basically during this period is that

Will Larson:So it basically didn't work properly for for much of the month it was it was like read only was back up in about three days but the vast majority of the actual user functionality just wasn't working properly for for pretty much an entire month and it was not not that good I mean like not not great but you know that wasn't the biggest problem dig had at that point but it was one of the biggest problems they had at that point and and it wasn't it wasn't a real sign of of things likely to go well for us but you know like I said you you learn from those and I'm really proud that like we and the team like got it working got it got it running even if like ultimately like we still went to zero and like ran out of money and kinda sold sold for parts do you

Lenny Rachitsky:Think dig could have made it there was a world where dig would have been a hugely successful business or do you think it's just way too late and it's the wrong product

Will Larson:The thing that really killed dig is the the change it was an SEO driven like so monetization was from ads dig was the well many companies including dig claimed that it was the first in kind of stream in feed like advertising company like where twitter has like ads within like the tweets or Facebook does but but dig did that before Facebook or Twitter really innovated the ad format but the vast majority of our monetization was on these we call them permalink pages which is the page where we then article we crawled and the vast majority traffic for that was driven by Google search and so there was an SEO change which really is like the the thing that started creating the urgency for us to launch this migration SEO change traffic started going down monetization was driven by that and so we were already on fire by the time we tried to launch this but I I do think that I still want something like what digg was trying to become today social news based on like what my friends are actually reading and liking merged with like a global index of kind of similar users who are interested in similar topics so still a product that I think google reader had has some kind of similar components to it these are both interesting products solving interesting problems that have not for whatever reason been successful as businesses and I do think there's a gap there still but there's a lot of people trying unsuccessfully to fill it and and there must be a reason why people struggle to fill it despite so many people trying

Lenny Rachitsky:Awesome Will is there anything else you wanna share or leave people with before we get to our very fast lightning round because I know you have to run-in about five minutes

Will Larson:I think we've covered a lot of it new new book coming out new book coming out in February Engineering Executive's Primer O'Reilly but that that's probably it

Lenny Rachitsky:Awesome where do people find that I know it's on O'Reilly you can look at a preview of it even today right

Will Larson:Yeah O'Reilly you can see that the early copy you can you can order it on Amazon as well but it won't won't be shipping until February

Lenny Rachitsky:Okay and then just to be clear who is this for it's for engineering executives by the sound of it

Will Larson:It's for engineering executives but but more so like anyone who wants to be one anyone who's trying to figure out how to work with an engineering executive so I think if you are struggling to understand why your CTO keeps doing boneheaded things or if you wanna side manage them you're the head of product and you can't get the CTO to stop complaining about the engineers need more interesting projects to work on this might be useful for you

Lenny Rachitsky:Too amazing okay ready for the lightning round

Will Larson:Let's let's

Lenny Rachitsky:Do it what are two or three books you recommended most to other people

Will Larson:So I talked about thinking systems of primer I talked about good strategy bad strategy but I'll give you a third one which is don't think of an elephant by George Lakoff it's a really interesting book about framing things and conversations that has really changed how I communicate

Lenny Rachitsky:Amazing favorite recent movie or TV show you've really enjoyed.

Will Larson:I don't watch much TV or many movies anymore but something I do still watch is Top Chef with my wife. She's a Top Chef super fan and there's something just very relaxing from these formulaic structured shows where you kinda know what's gonna happen. There's no real consequences that matter too much and just kind of escaping from real life to these formulas can be pretty pretty useful.

Lenny Rachitsky:No one's ever mentioned Top Chef before so that's fun. Do you have a favorite interview question that you like to ask candidates that you're interviewing for a job?

Will Larson:A lot of my interviews now are trying to help people decide if they actually want to join a company and so my favorite question I ask now is like, hey we really love you, you're gonna come through. I think you're gonna get a lot of offers from other companies too. I bet you'll have three or four really compelling offers because you're a fantastic candidate. How are you gonna figure out really specifically which of those options are right for you? And I think it forces people to tell you what they want and then you tell them why you have that more than anyone else and then you can actually pitch them on what matters versus pitching on things that don't.

Lenny Rachitsky:Love that. You have a favorite life motto that you often come back to share with friends, find useful either in work or in life?

Will Larson:No mottos but I can think of two things I thought about a lot. Yeah, at Uber something I talk to people a lot because it was a challenging time for much of it was, there's no way around just through and that was like, hey we're not gonna dodge around this, we're gonna gut through it and we're gonna get to the other side and then we're gonna be there. What I think about a lot more now is will anyone remember what we decided in six months because I think people stress out about a lot of decisions but I increasingly believe most decisions people stress out about aren't that important. So I'm like, well anyone care in six months what we did here and the answer is no. Just do something reasonable and let's move on to the next more important thing.

Lenny Rachitsky:I love that. You've done a lot of writing. Is there a piece that you've written that you feel like is underappreciated that no one really totally got and hasn't spread and you're like, oh I'm so proud of that one?

Will Larson:Maybe the piece I'm most proud of from last year was like Hard to Work With. So Hard to Work With is basically I see a lot of people who are incredibly talented but they try to hold their peers to a high standard and then they're viewed as combative or difficult to work with. And this one comes from a core struggle of my early career where I kept trying. I thought it was holding people accountable but people were just like you suck to work with and I was like, but I'm just trying to have a high standard, isn't that what we want? Every company talks about honest values every company is like, we have high standards and you're like, well let's do it and then they're like, we don't have high standards here, like you suck. So that one's really transformational to me. I think it hits some people hard because I think a lot of people go their entire career without figuring this one out, and they're some of the most talented hardest working people you'll ever work with and can't quite land this one idea that's holding them back and they care so much and they're often despised because they care so much. And I think this is one that I hope more people will read over time. I think there's a really important lesson for me in there.

Lenny Rachitsky:Well, we will link to it in the show notes and help more people discover it. Two last questions: where can folks find me online if they wanna reach out and maybe follow-up on questions and how can listeners be useful to you?

Will Larson:So find me online lethain.com, lethain.com. All my writing, my books, everything links there. The biggest thing I'm thinking about right now is just strategy. So really curious for folks who are thinking about strategy, who've done product, business, or engineering strategy. I'd love to hear from folks what they're thinking about, what's actually worked, and maybe what are the lies that have not turned out to work that they thought might work earlier in their journey.

Lenny Rachitsky:Amazing Will, thank you so much for being here.

Will Larson:Thank you so much. This is really fantastic.

Lenny Rachitsky:Same for me. 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 lennysodcast.com. See you in the next episode.