The state of the 2023 tech market
source link: https://changelog.com/friends/23
Go to the source link to view the article. You can view the picture content, updated content and better typesetting reading experience. If the link is broken, please click the button below to view the snapshot at that time.
Transcript
Changelog
Play the audio to listen along while you enjoy the transcript. š§
Weāre here with our good friend, Gergely Orosz. Gergely, itās good to see you, good to hear from you. Itās our annual, I guess, conversation. It feels like just really recently, Jerod. Doesnāt it feel like it was just maybe a month ago or so that we talked to him? I feel like this year has gone by so fast.
It feels recent, but a lot has changedā¦ And it has been a year, so welcome back, man.
Itās been a year. Letās start an annual tradition. Iām really glad to be back. Always good to catch up. It feels like a year goes by really quickly, but then a lot of stuff just completely changes as well.
I do want to highlight the phenomenal title, Jerod. I feel like we have to, for the last oneā¦ Because the second to last one was āThis insane tech hiring market.ā And then it was basically the same title, but with a bang in front of insane, so āThis not insane tech hiring market.ā We got a lot of feedback about that title saying āJust genius titlingā, you know?
The nerds loved it, yes. Well, we negated that suckerā¦ It worked out well. What kind of operator should we put on it this year, Gergely? Is it an insane tech hiring mark, is it a not insane, is it average? Whatās your feeling in the end of 2023?
Well, I feel itās the question mark hiring market. Everyoneās trying to figure out where are we, but more importantly, where is it going to go? Do we have an option to go back at some point to that more insane? I think people are hoping for that. But in reality, thereās a bit of a worry, and a reality that maybe weāve seen the very best hiring market as software engineers, and the best opportunities, the best demand around 2021-2002, or even between 2010 all the way to 2022. And thereās a bit of a worry that maybe thatās not going to returnā¦ Which honestly, it might not be a bad thing. Itās change. I feel itās the quickly-changing hiring market.
Well, we have a lot of relationships inside of large tech brands, and I would say that we see those relationships change frequentlyā¦ So as an organization who interfaces with many large brands in our tech space, seeing the folks that we work with closely change pretty much year after yearā¦ Itās either theyāve moved on, something happened internally, I send an email and suddenly weāre coordinating something, and then that email is returned by the bot that says āThis email doesnāt exist anymore.ā It is far more frequent, and I feel like every time I talk to somebody, I have to say āCongratulations, youāre still here. Please catch me up if you think things are changing.ā And they may be experiencing a surprise, to some degree, but I feel personally that the change has been very frequentā¦ Even today. But I kind of see a positive uptick. I donāt see us going back to the sheer volumes of 2021, 2022. I think that may have been the best of whatās to come, maybe for the next decadeā¦ I think maybe the big player might be the way that this hype cycle of AI has really happened big this year. And Iām not suggesting we go deep on it, but I think when we talk about what may happen with software development, thereās a lot of things happening around artificial intelligence that aid a development team. And in what way does it help them? It helps them with observability, it helps them with database schema, it helps them with all these unique things that just was not there last year, really. Itās a brand new thing, and Iām wondering how it will augment teams, and whether or not that actually changes hiring practices.
Well, and [unintelligible 00:03:56.19] second-guessing, and I think this is a little bit hard to explain, and hereās why. Around, letās say - I wasnāt there, but I read the story of what happened when the compiler first came out. Before the compiler, people would just write their machine-generated code, often onto a card, and they would feed it into this big mainframeā¦ And it was a lot of work, and that mainframe was very expensive, so there was time sharing, and developer time was very valuable. The compiler just sped that up; it was a 10x improvement, literally. And what you would have expected is āWell, you need fewer developersā, right? Like, one developer can do 10 times as many. But curious enough, the number of developers has exploded since then, because there was moreā¦ And then after the compiler, we still had lower-level languages; we had these higher-level languages come off, like letās say we had C, and then we had obviously C++ā¦ But C# or Java, which are more productive line, which is arguably ā which, again, it would have meant that you need fewer developers, and it just kept exploding. So again, with AI - of course itās gonna make us more productive. Iām not saying this is even 10 times as productive, but it will be easily 20%, 30%, 50%, who knows, depending on who weāre talking to. And the logical thing would be we can do more with fewer developers. Like, we would need fewer developers. But again, looking back at the past, what always happened is we just had more developers, because now what happened every single time is a lot of businesses that couldnāt afford developers or development had this.
A good example is website builders. Back in the ā90s, you all remember, you had to hire a webmaster to build a web page, and they had to maintain it, and it was expensive, and so the bigger companies all had their webmasters, but small businesses didnāt have it. Fast-forward to today, you can just click and put together a website, but thereās strangely not less demand for people building or tweaking websites, other specialists who are tweaking WordPress etc.
[00:05:48.17] So my sense is that as long as technology is still spreading across the world, we will still see a demand rise. And technology is still not everywhere, and Iām kind of thinking that we might see a little bit different ā until now, itās kind of gone into big tech, and these amazing positions where which pay very well and generate a lot of valueā¦ But here I am, sitting, frustrated with my utility company, for example; how just awkward it is to pay. Or public transport - again, technology is there, but itās not particularly good. So I wonder what might happen if we see a lot of technology and technologists, and good software engineers end up at these places as well. Maybe work conditions improve, compensation improves, and our kind of quality of life improves. Because Iāll be honest, my quality of life is not really driven by Facebook or Netflix, itās kind of driven by the more mundane thing; how easy it is to reach the local council. I can now do this online.
And so I wonder if that part of like these businesses that are still not really digital - are we going to see in the next 10 years a boom there of software, maybe AI-assisted, going there, and obviously, software engineers building that? So Iām kind of optimistic that thatās going to happen. AI makes everyone more productive.
For example, small businesses until now, building an app - how much did it cost? $20,000, $50,000, $100,000 to build a custom app. It was not affordable for a lot of small businesses. My trainer at the gym was telling me he really wants to build this app, heās got this dream of doing it, but he cannot afford it. All these people might be able to do it through, again, a little bit what happened with websites becoming point and click, it might happen with all sorts of apps.
So I think itāll be super-interesting, exciting. There will be demand increasing in a lot of these areasā¦ I donāt know about the rest, but so far, Iām not seeing [unintelligible 00:07:35.01]
However, one thing I will say - this is the first time Iām seeing software engineers becoming worried about our jobs. Until now, letās just be honest, what we did is we kind of automated other peopleās jobs, right? Like customer support - at every single company customer support teams have been going down in headcount as software engineers. Not me, but I saw teams at Uber were building more and more efficient ways to do it, adding machine learning, adding helpful tools so one human could serve more people. And we were very proud of this, and it was cost-saving. And now itās the first time where software engineers are asking āAm I going to automate myself out of a job?ā And I hope the answer is no, but weāve never asked this question before. So I think this is a big, big change.
Yeah. I think thatās on point, and I think that specifically around the proliferation of ideasā¦ I mean, if weāve seen whatās even happened this year, weāre very much still in demo land, of like āLook what this can do.ā Thereās very few production-grade rollouts of these things in scale. Thereās a few, and theyāre impressive, but what Iāve seen is just huge amounts of new ideas, so much so that like Iām holding on for dear life just watching the demos scroll by. And Iām like āWow, look at all these new ideas.ā This is not going to create less software; this is going to create more software, and weāre gonna be working at a higher level, weāre going to provide our value at a different place than we used toā¦ But I think that example of what the web did, moving higher up the value chain, and the abstraction level - maybe LLMs are the next compiler for the next 10 years, and weāre going to be way more productive, and thatās just gonna bring so many more people to the table who previously were just priced out. And I think itās net positive.
I also feel like ā letās say these website builders [unintelligible 00:09:21.28] because thatās a great example. I think thatās been really commoditized. If you want to do a website today, and you have no technical knowledge, you can absolutely do it, right? But then what happens when you built a website? Youāre a small business, you employ a few people, letās say you run a barber shop or something, and you built your own website, you clicked it together, and youāre now starting to grow. Business is good, youāre spending less on technology, and itās bringing you a lot more value. You start to expand, you want to do custom stuffā¦ And suddenly, that point-and-click thing doesnāt work. You now need to hire a professional who understands how your thing works, and how the leading industry stuff worksā¦ So you bring in those people.
[00:09:59.06] So I feel thereās a little bit of ā one thing that is a really big commodity these days is plumbing, for example. Plumbing hasnāt changed all that much in the past 50 years or so. And yet thereās such a huge demand for plumbers who actuallyā
Well, yes, itās not going anywhere.
ā¦can get the work done. So Iām feeling thatās whatās going to happen, if I just take that analogy. Letās assume that there is not going to be much a bigger demand increase, which I donāt believe is the case, but letās take a pessimistic view. There will still be a huge demand for software engineers who understand how these things work, who understand what is under the AI solution, who understand what is going on at the machine level, in the cloud. When you have an issue, how does the code execute? And could you just have some CPU issues?
Thereās now a new generation of software engineers who donāt really know too much about infrastructure, which is fine for the most part, except when you need to go deep. So I think the software engineers, the craftspeople who understand the whole stack, and have experience debugging and fixing issues - they will be very much in demand. And thereās this joke of calling a repairman to fix this complicated machinery, and the repairman looks at it for 10 minutes, and takes a hammer, and hits it at one point and it starts to work, and he charges $1,000. And I ask ā$1,000? Why $1,000?ā Heās like āWell, $10 for the hammer, and $990 for knowing where to hit it.ā I think thatās gonna be software engineers.
Donāt forget that with all this AI stuff, software is gonna be way more complicated. AI will generate more complicated software, so itās going to be harder to know what is going wrong. I have noticed this as well when Iām using ChatGPT or some of the code generators; they generate code, but they are often incorrect, and you need to know what you need to know.
So I feel the whole worry that weāre going to be out of a job is not true. What is true is there will be the kind of people who used to have a job, letās say from a bootcamp, from doing two months and being able to do HTML and CSS. That is no longer going to be marketable. You will need deeper skills. So my prediction is that to enter software engineering, weāre going to go back to what every other craft has - you will need to study several years. How do you become an electrical engineer? I mean, you can self-study, but you probably wonāt get a job. Most people, unfortunately, go to college. I mean, itās just the reality. They learn a bunch of stuff, that takes a lot of time, and then they enter the industry. I think that will change, and that will change very quickly.
You know, who is out of a job though? Stack Overflow. Arenāt they? I mean, Googleā¦ And Iām not even a heavy user of these tools. Iām kind of reserved in my use. But itās the first place Iām gonna go already. And itās been six months? I mean, I havenāt been to Stack Overflow in the last six months, and Iām a typical engineer. I mean, what happens from here?
I mean, we should talk about two things about StackOverflow. One is Joel Spolsky - what incredible timing, in hindsight, of how he sold it. And obviously, this is not someone who can read the future, but was it in 2020 or 2019 when he sold it for 1.8 billion to a private equity firm? Weāll have to check the exact date, but I do remember that this was before ChatGPT was even announced in preview. And shortly after the sale, it was announced in preview. People still didnāt assume it was such a big dealā¦ But anyone paying attention could have thought that this might be damaging it.
So for the private equity company, at the time it looked like a great deal, because they could obviously monetize it, maybe even take it publicā¦ Just keep growing it, because it kept growing. In hindsight, it was the perfect time to exit.
That was a great time.
If you knew anything about LLMs, then it was a great move in that sense. And now, honestly, Stack Overflow I think just has a problem where to position themselves. What I understand though, having talked to both some people there, but also following up what theyāre publicly doing, their focus is not really any more the public side. I mean, itās still there, itās still driving traffic, but what their bigger business focus is, and I understand is their biggest business income, is offering Stack Overflow for teams, for companies, which can actually serve as a very powerful AI assistant. Because what we now know, in the past month, I think, thereās several articles that these large language models like ChatGPT and Claude, they cannot make up new facts. You need to feed all the data into them.
[00:14:13.16] So as a company, I feel ā again, Iām not the biggest expert on AI, so youāll have to find other peopleā¦ But my understanding is we need to generate all that data, and so we need to incentivize data. So Stack Overflow could be in this great position that at companies they say āHey, use usā, and people will keep contributing the data that the AI cannot find, and so youāre going to be more efficient. Because we might have this data drought soon enough, that right now ChatGPT is amazing at giving coding suggestions, because itās been trained on Stack Overflow, but now no oneās contributing to Stack Overflow, so the next version of letās say TypeScript or whatever new language - it just might get worse. And then thereās gonna be this game of how do you incentivize people to actually contribute training data?
Yeah, thatās definitely interesting, especially with the open web. I mean, as a publisher, of course, you have a direct relationship with your audience, so thatās spectacularā¦ But there are other publishers who are having an indirect relationship with how they make money and their audience, and what incentivizes them in the future to crank out the news articles, to crank out the blog postsā¦ Because the trafficās just not going to come anymore, and thatās how they get their money. Soā¦
Could we go back to Experts Exchange? Do you remember what it was before, Experts Exchange? Before Stack Overflow it was Experts exchange. It was a Q&A site where you saw the questions, but to see the answers, first of all, you could pay an expert to answer, a humanās answer, who hopefully would give you a good answerā¦ And then once they answered, it was hidden behind a paywall, and you needed to pay to unlock it. It didnāt really work that well. It felt very scammy, lots of dark patternsā¦ It was clearly making money, but Stack Overflow came in to replace this model. But Iām saying this because Iām now starting to see some things going a little bit circular. Going back - because weāre now back to āHow do you incentivize people?ā People, especially software engineers, are not stupid; theyāre very smart. Every software engineer now knows that whatever you contribute to a forum or to the open web, it will be used to train the AI, including your own blog, including to GitHub. And more and more of them will ask āWhat is my incentive to do so? Iāll do it, but what do I get in return? Or what kind of noble cause am I helping?ā People will probably be fine contributing to some open source AI or something that benefits, but people are going to be a bit hesitant for private companies harvesting this data. So I think weāll see a behavior change. This is not going to be in the next six months, but I think the next five years, itāll alter drastically how much people are willing to just share their creative output.
Youāve got to think about an AI that essentially counts credits. The AI consumes knowledge we put out there, and how do you track the incentive? Well, if the AI can do it, then you say āWell, what if the AI is biased?ā Well, isnāt any pattern-matching kind of biased? Like, if you pattern-matched towards a certain skew, because you have either all the data and you can pattern-match clearly, or you have limited data and you pattern-match against what is truly not a holistic dataset, either way you have a bias. And Iām just wondering if weāll get to a place - and future Adam or somebody out there come back to this, because this might be accurateā¦ What if in like 10 years something like this gets done, and humanity says āYou know what - we are so biased as humans, because we have emotion, and we have all this humanity and all these humanistic tendencies in us that we have to program the AI to the perfect human nature, and let the bias be in the data.ā And then humanity evolves its knowledge based upon what it puts back in, and the AI creates credits of source that says āYou know what, Jerod is literally better than Adam, or Adam is literally better than Jerod and Gergely, because his contribution is so much greater, and the perfect human nature-biased AI is all knowing.ā
A real meritocracy. [laughter] I donāt know, I think I just stack-overflowed right there. You went too deep on me. [laughs]
That seems pretty plausible.
I think this sounds too simple, and the world is way more messy, more unpredictableā¦ But also, looking back, itās so easy to see the patterns. Iām always hesitant to predict what will happen, but the interesting thing about this whole AI thing is the longer Iām in tech, the more Iām realizing that technology is really interesting, exciting, and itās a fun part. But the real messy part, and the thing that is the hardest to figure out is humans.
Break: [00:18:49.05]
Letās just look at Open AI, right?
I was hoping you were gonna say that.
Theyāre right now the most hyped AI company, theyāre also the one that objectively is shipping the most visible things, and they have the most users. They passed 100 million weekly users in less than a year with ChatGPT. And yet, the company almost was in turmoil, and maybe close to ceasing to exist because of a few people and their leadership team having disagreements. And it just comes to show ā I think there was this joke of Open AI wants to align AI, make sure that it is aligned, but they couldnāt align themselves as a group of people. And I think thereās an interesting question that goes backā¦ Thereās a question of āWhat does AI do?ā, but thereās the question of who controls the AI? What is the group of people and what are their goals? And what constraints do they set, and how do they program that? And thatās gonna be just as important.
Just as messy, yeah. Yeah, what a fiasco that wasā¦ I mean, it was a good week to be an internet denizen and just watch it unfold over the course of 72 hoursā¦ What was your coverage? What were you doing during that time? Iām sure you wrote about it, but what were your thoughts throughout?
Yeah, so I was initially just following along, because like everyone, I just couldnāt really believe what was happening. It was just so shocking. I had literally finished an article talking with one of the heads of engineering, one of the two heads of engineering as ChatGPT, Evan Morikawa, who was one of the first engineers on ChatGPT; three years ago he joined this small team called Applied. There were six people in Open AI; so Open AI back then, in 2020, it was about 150, and 144 of them were research people. They were just building these really cool models, the app that would eventually become ChatGPT, and they hired a team to turn this into Applied. And he was one of the first six people, first engineering manager. In three years they grew to 150 people, and only one year ago they decided to launch ChatGPT. So they were building some new cool stuff on the side, which was very surprising to meā¦ And they apparently built ChatGPT in a few weeks. They launched it, it became very successful, they could barely keep up with demand, and they kept scaling up.
And one of the problems that they had, which we didnāt write about, but we might have a follow-up - they were short on GPUs. Even though they had all the access to Microsoft, they were short on GPUs. And there was a lot of really fun engineering challenges that I hope to come back and cover one day; itās fascinating.
I had just finished this article, which was fascinating, on how quickly they respondedā¦ And what was really interesting about OpenAI is I talked with a software engineer, and even as a software engineer, he told me āLook, this whole ChatGPT - itās kind of a black box that predicts things really well, and we know it works well, and our job is to productionize it as an engineering team.ā And what they did is they hired very senior engineers, they operated it like a startup, they moved very fast, and it felt to me that they could execute so quickly because, again, they had engineers with 10+ years experience in large-scale environmentsā¦ They just knew what to build, they were very motivatedā¦ And, interestingly enough, they worked in one office location, which apparently worked really, really well.
So thereās this whole debate on remote or not remoteā¦ From my understanding, Open AI would have never been as successful if they were not located in the same location. And they could afford this, they could pay people, they could motivate people, and so on.
[00:27:49.24] So I had just finished this, and it was just a really good example of how Open AI is moving so fast. And then as I published it, their CEO was firedā¦ Even though the company is doing extremely well. It seems no one can stop it. And I made this joke on social media, we were wondering who could stop Open AI, and itās themselves. It felt from the outside they were sabotaging themselves. As I was thinking back āWhen was the last time weāve seen such a shocking CEO firing?ā Because Travis Kalanick, for example, was fired from Uber, but it was not really unexpected, because the company was struggling.
But when do you see a company doing amazingly well just going up, up up, and theyāre about to close this 86 billion round, and I think everyone is expecting that theyāre on the way to become a trillion-dollar companyā¦ And I think the only thing that we could all think of was Steve Jobs being fired from Apple. But actually, when Steve Jobs was fired from Apple, Apple was not doing that great. So you could argue there was a bit more to it. And I think we all just like ā I followed the drama along, I figured this is the end of Sam Altmanā¦ Then there was a revolt from employeesā¦ The boardās surprisingly going quietā¦
My biggest surprise from the whole saga was that Sam was fired Fridayā¦ Satya Nadella was the one ā he felt like he was the spokesperson for Open AI.
Heās sprung into action.
Heās sprung into action. He started to communicate what was happening at Open AI, even though OpenAI was arguably a way, way smaller company than Microsoftā¦ And then on Monday, he went on a press tour. And then he came up with a solution, which in hindsight maybe it was more of a tactic of hiring everyone from Open AI, but apparently they opened ā so my biggest surprise was how Open AI is so darn important for Microsoft, incredibly important. And my biggest-biggest surprise was that they wanted to keep them independent. They would really prefer that OpenAI operate independently, they get the benefit of their research at Applied, and they donāt really want it inside of Microsoft, Iām assuming, because of the scrutiny.
So it was really fascinatingā¦ And I think this whole event just broke the image of Open AI being this unbeatable company. For example, now I think a lot of people are looking at Anthropic; their models are doing pretty wellā¦ They just didnāt have this drama. They could, everyone couldā¦ My biggest takeaway is I feel this field is super-volatile. I think until now what weāve seen is they move ā these companies move really fast, and weāre wondering āHow can they move this fast?ā, but they actually donāt have it all figured out.
Itās not sustainable, yeah.
Itās not sustainable. They will need to slow down to get stabilityā¦ Or they keep rushing, and theyāll be extremely unstable, just like Open AI is right now. So Iām assuming things are probably ā Iām not envying any of the people there. The only thing I will really commend is it seems the team has come together for a cause. That team is, as agreed, the employees that are there supporting this cause, they want to work with this leaderā¦ Which is, again, unparalleled to see almost 95% of people sign some petition that they put up together.
So it feels to me like that is a great sign for any company. Theyāre in this together. As far as I know, no one took up the offer of jumping ship for same compensation at Salesforce. So I think itās just very confusing. Thereās lots of really positive things about OpenAI, but I feel up there thereās too many questions with the leadership right now, and how stable their leadership really is, and their vision.
Yeah. Funny joke I heard, when it became clear that everything was pretty much gonna land where it had begun in terms of employer and all of thatā¦ Somebody said āI feel sorry for that one Microsoft IT worker who has to return 770 MacBooks back to Apple.ā [laughs]
Yeah, that was also crazy, right? I think we read that Satya Nadella agreed that they donāt have to use Teams to appease them.
Yeah, they donāt have to use Microsoft Teams. So many just entertainment nuggets that came out during that time periodā¦ And such a weird thing, and then just to have it all kind of land where it startedā¦ But yeah, the view of the inside of the leadership at Open AI, and the disagreements, and just the weirdness of their board, and their company structure, or whatever it is, the entityā¦ I guess, in a sense, I like it, because itās going to provide more diversity in the space. Weāre gonna take other companies more seriously. It just felt like Open AI had such a huge lead, and it just continued to, like you said, just launch, and improve, and every time weād turn around, ChatGPT was better. And they were integrating other peopleās stuff, it looked like.
[00:32:20.23] Yeah, it felt like superhuman. So even when I talked with Evan, it just seemed they did everything perfectly, or even better, and just made no mistakesā¦ Which, again, weāre all human, so I agree with you. I think it was good to see that theyāre human, they make mistakes; itās people working there, and theyāre not special. Of course, again, they came up with a really good [unintelligible 00:32:40.29] but I think because of this, again, if ChatGPT is a little bit better, or even a lot better than all the other models, the others are gonna catch up. We know itās the same people working there, same faults, same kind of approachesā¦ What weāve seen is Open AI has capitalized and they have moved a lot faster than their competitors, for different reasons.
It seems like the real winner in all this is NVIDIAā¦ [laughs] Doesnāt it?
I think a little bit of Anthropic as well. The big question with Open AI is āWhat do they care about? Do they care about moving fast? Or AI safety?ā Because this was the big argument internally.
Right. And they canāt decide.
I donāt think they can decide. My sense is theyāre moving fast. And I think, by the way, that is the right strategy for a Silicon Valley company that wants to maximize its value for the employees, and the shareholders, and its markets. And they have been moving the fastestā¦ Whereas Anthropic is really principled, saying they are safety-first, and theyāre also moving at a good speedā¦ So I think theyāre projecting a lot more stability. And Iām now seeing on social media a lot more people sharing Claude as examples of their chatbot, and Iām also just going to try it out as well to compareā¦ And just keep it in mind that thereās not only one player in the space. And Anthropic is very tempting, because theyāre also independent. Theyāre not ā like, Google has Bardā¦
Yeah, whatās their story? I donāt know much about them.
Their story is that some Open AI employees were actually unhappy with how decisions were made at Open AI, and how they felt safety was not as much prioritized as they would have hopedā¦ And so they started Anthropic, where they said āSafety is our number one. Weāre going to build on that.ā And they just took it from there. So itās almost like a fork of Open AI, with a bit of a different focusā¦ But I think weāre now starting to see that the principles do matter on the mid and long-term. On the short term, speed is everything.
And so I think itāll be fascinating to see. Iām really rooting for all of these companies, by the way. I think I will always be rooting personally for the smaller ones a lot more than rooting for the big guys, Microsoft, Google etc. to grow even bigger. So I hope Open AI, Anthropic, and other startups that come and fill this space, they will succeedā¦ But what I feel is these companies have ā Open AI has rapidly gone through the startup phase, and now theyāre almost big tech. They now rapidly have to matureā¦ Which is going to be painful, because - like, I worked at a company that went through this maturing at Uberā¦ Itās not as fun. Itās not as fun working at a larger company than it is at a smaller one. To make it fun, you need a really good founder, and thatās where NVIDIA is super-interesting. Theyāre now a big company, and people just love the CEO.
I was in the US a few weeks ago, and I caught up with a friend who works at NVIDIA, and he was telling me that, like, Jensen - people adore him. They donāt have to go into the office, but when Jensen all hands and he speaks on the all hands, he kind of freestyles the whole thing, answers every question, doesnāt come with notes, and is just very passionate. Everyone is in the office, everyone goes to see him.
So to me, the big surprise in all of this thing is, I feel ā when we talk about big tech, I used to always think Microsoft, Google, Amazon, Netflix, Appleā¦ And [unintelligible 00:35:45.10] And now by market cap theyāre like the sixth or seventh biggest company. By the growth, theyāre the biggest, but we somehow still donāt think of them like that. So I donāt know too much about them, but I am planning to learn more. And I have some friends who used to work at Google, and other places, and theyāre now at NVIDIA, and theyāre very happyā¦ And I know theyāre doing good. And as you said, theyāre clearly the biggest winners of all of this right now.
[00:36:11.23] Well, they get to benefit from it, right? I mean, they get to get their AI and keep it, too. I mean, all the GPUs out there - everything is powered by what theyāre buildingā¦
Right nowā¦ But for how long?
Thatās the question, yeah.
Yeah, nothing lasts forever, thatās for sure. But I think it does help to have a charismatic leader, that can be there, answer questions, as you mentionedā¦ Thatās ā
And I feel a very down to earth leader. Donāt forget, he was the only leader, when the stocks were going down last year, and most public companies, even Facebook gave into investor pressure to fire 10% or whatever, to show to the investors that āWeāre firing, weāre cutting costsā, and then the stock recovered, and they just rehired. Facebook is a prime example. Theyāre hiring like crazy right now, so theyāre probably gonna ā again, I donāt have the data, but my sense is theyāre probably gonna go back to the headcount, to where they were before all this firing. NVIDIA did not do this. Their stock wasnāt doing that bad, but since Jensen was there, they havenāt fired anybody. Theyāve been very clear about this. So heās been holding his bag against investors. And also, theyāre the only company that I know, in big tech ā apparently, their unofficial motto is that āOur company could go out of business in three daysā, so there is a sense of urgency that is just always there.
Yeah, I donāt knowā¦ Iāve now listened to Jensen a couple of times, on a couple of podcastsā¦ I just donāt know any other CEO who just sounds so honest and down to earth.
Yeah. He doesnāt sound like businessā¦ Which, again, itās probably ā Satya Nadella is also very relatable. Heās done a great job at Microsoft. But it does fill me with some hope that weāre seeing some of the kind of human CEOs succeed, as opposed to the ones that are just doing death marches, or sticking to a vision and not caring about anyone else.
Rightā¦ The evil villains.
Break: [00:38:06.00]
Well, you mentioned Facebookās hiringā¦ Are you paying attention to the uptick or neutral hiring processes of, I suppose big tech at large, but something that gives us to judge the tea leaves against with the market rebounding or not rebounding?
Iām keeping tabs on it as much as I canā¦ Iām not hearing too much. What I am hearing is Facebook has started to hire a lot. Theyāre the only company whoās really ticked up their hiring. Other companies are also ā hiring just seems to be quietly coming back. I think I talked with some people Google where theyāre getting headcount. Amazon hiring is back, even though Amazon also fired a lot of people with the return to officeā¦
It feels to me that thereās no massive hiring, but thereās now ā Iām not hearing places that donāt do backfills. And these companies, typically large tech companies, in a normal year, they would lose 10% of their headcount, or 8%, in terms of people leaving and backfilling. It seems this is a bit lower right nowā¦ But still, for these companies to hire even 5% of a 100,000 person company, or letās say thereās 50,000 people, software engineers working at a company like Microsoft - Iām just guessing - 5% of that would beā¦ Itās in the thousands. So I hear a lot of backfilling and open positions for more experienced engineers, to areas that seem to be hiring a lot less. Engineering leadership, engineering managementā¦ Most companies are reducing management layers, so thereās less space for senior managers, directors, and Iām hearing a lot of directors, people who were managing managers, sometimes going back to managing teams, so taking a step downā¦ And new grads; new grad hiring seems to be very much down at a lot of these companies.
Iāll have [unintelligible 00:43:32.24] hiring will be back in big tech. Last year, at least in Silicon Valley, most big tech did not hire interns. Roblox did, for example. I spoke to a friend at Roblox, and this friend told me something really interesting. First of all, they were pretty surprised because they got all the graphs to choose from. Usually Facebook takes the best grads, and Google, and theyāre kind of left fighting for the restā¦ So this year they could just choose ā sorry, last year in 2022 they could choose from all of them.
And then this manager told me something interesting. They hired these grads in the spring, they were working with them, and he asked one of them āHey, how are you doing? What are your friends up to?ā, and this intern said āWell, a lot of my friends who didnāt get the internships they were hoping for, they didnāt get good offers, some of them just went back to grad school for another two years, and theyāre hoping the market will get better.ā And thereās been a few people, apparently - this is Silicon Valley - who have just changed from software. They just took to a different place.
And another really interesting thing that Iāve heard, that I think is very specific to the US, is - one of my friends whoās a manager at one of these late-stage startups, her husband is a university professor somewhere in California. And he told me that during the lockdowns, thereās been a class that was a pass/fail. Instead of getting grades, they were just pass/fail. And what he noticed is these classes are a lot worse in terms of capability. So now heās taking some of these classes that last year there was only pass/fail, no grades, and apparently he needs to dumb down the assignments, because they just donāt have the skills. And sheās also hiring some of these grads, and sheās saying that this whole pass/fail year, which is like one or two years, is just not where it is before or after. So she was like āOh, I think weāll have a problem. These graduates may be ā for no fault of their own, theyāre not going to be as competitive.ā She said they can probably catch up, but this is just so interesting on how something seemingly so small - itās COVID, letās make it a bit easier, pass/failā¦ And on the workforce, managers notice, professors notice. Theyāre like āTheyāre not working as hard. Theyāre kind of more [unintelligible 00:45:37.26]
You ask a lot of young kids these days, they want to be influencers. Thatās something thatās cropped up as āI want to teach people something I donāt even know, or learn something enough to teach people enoughā¦ā Thereās a lot of influencing or desire to be a star in those ways, rather than, I think, put in the grind that weāve all put in to learn what we learned, to get where weāre at now. I worry, like any (I suppose) 40-year-old would worry about the prior generation and what they might come into, becauseā¦
[00:46:12.03] Well, yes and no. So I think, obviously [unintelligible 00:46:12.26] itās funny, because a lot of people aspire, young people out of college, or even in college, to become an influencer and have lots of followersā¦ Which is practically like the description of me. I have a lot of followers on different social media platformsā¦
Youāre an influencer?
Well, I mean, I could be labeled that, but I really try not to be as such. My goal is to learn interesting things, and share these interesting things, and also run a business on top of itā¦ So I think of myself as ā just like the two of you, I run a business which happens to have a newsletter component, but I think of it as education, and keeping up with the industry. I feel thereās some people who are looking at influencing as a means to an end. āI have a lot of followers, and Iāll somehow make money. Iāll do sponsorships, or whatnot.ā
I think thereās always going to be those people. And this path is open to more people. I think itās a lot harder than you think, and my honest view is I think itās a terrible thing to only aim for having these large following numbers, because it can be taken away anytime from you. These platforms, every single social media platform, youāre at the mercy of whatever the algorithm is thereā¦ And you can just have a big disappointment.
What I think works a lot better, what Iāve seen multiple times, is people who have a business, a small business, or something that they do that is generating some most of their revenue, or some of their revenue, and then on social media they share something around that. Thereās a lot of, for example, photographers who do this on YouTube - they shoot stuff for clients, and they often record how they do, what gear they useā¦ First of all, they donāt stress as much about how many views they get, or how much ad revenueā¦ Itās kind of nice when it blows up, but they donāt care too much. Theyāre also just a lot more authentic. Thatās also what I try to do. I have a lot of followers on different platforms, especially Twitter and LinkedIn, these text-based onesā¦ But most of my output is just me ā Iām researching my articles, so my goal is to figure out whatās going on in the market, and I often just get input, I sometimes share draftsā¦ I donāt really care if tomorrow the algorithm deprioritizes my views. I enrolled in Twitter ad revenue, because I wanted to see how much I can make, but I donāt really care if itās zero or not. So itās just a nice bonus for me, but itās not my main thing. So thatās what Iām trying to do.
But stepping aside from this, worrying about the next generation, I actually asked this engineering manager in Silicon Valley whoās hiring new grads, about Gen Z, āWhat do you think about Gen Z?ā The gaduating generation, or maybe the one after, growing up with iPhonesā¦ And she was saying āI love them. Itās amazing, theyāre just so engaged. They come into the workplaceāā and I think in software engineering, we product managers, especially - sheās more of a product-focused engineering managerā¦ Thereās always been a problem of you hire a software engineer, and you really want to get them interested in product. Talk to customers, understand themā¦ And sheās like āWell, with this generation you donāt have that.ā They come in, theyāre like āAlright, Iāve used the app, Iāve tried out our competitorsā¦ Hereās what I think we should be doing.ā Sheās like āAmazing.ā So theyāre super-engaged, really bright, really good at context-switchingā¦
Apparently very protective of their time as well. So they will say āOkay, itās like 6pm. I need to meet up with friends. Goodbye. Yeah, Iāve done all my work.ā Like, this is a generation that doesnāt do any more of that āOh, well, weāll put in the hours, weāll just stay here for face timeā¦ā And also they just ā I have some friends in Europeā¦ Europe is a little bit slower-paced in a lot of companiesā¦ Theyāre complaining that theyāre quitting, because theyāre bored, and they donāt see advancement. In Silicon Valley, my friend is like āTheyāre amazing, because itās a startup, itās fast-moving.ā They love it. Theyāre growing with it, and theyāre growing really fast.
So I actually feel like yes, of course, thereās influencers, but I feel like thatās a very small subset. I hear very good things about this new generation who is entering the workforce, and I think it might give a run to the money for us people whoāve been there for a bit longer, which is, I think, a great thing.
[00:50:02.07] Yeah, theyāre sharp. Well, thatās good to hear, that theyāre coming into the workforce saying āI tried the app, I compared it to the competitorsā¦ Hereās what we should do.ā I mean, that usually comes after a year or twoā¦ Because youāve gotta come in super-green, like āI have no idea how business works.ā And I think that entrepreneurship has been pushed down to the younger generations to be like āI should be doing something in my teens, not because I want to make money, but because I want to understand the way the world works at an earlier age, so that when I do enter the workforce, or I consider it, I have a more clearer understanding of directions I should go, or whatnot.ā That also happens with wise parenting, and involved parentsā¦ So not all that is on just simply society, but I think it is a societal thing where entrepreneurship is essentially more accessible to the younger generation.
I mean, I ran papers when I was a kid. That was the most common thing you could do when youāre young. I did not sell lemonade, I didnāt have a lemonade standā¦ I would have done that ā I shoveled snow for folks, I cut grassā¦ Me and my friends had a grass business called BAD, because it was Ben, Adam and Donaldā¦ So we were the BAD crew, you know? In a good way. So thatās kind of fun. But I mean, we did stuff like that as a young age. Now itās different. Now itās like, you can literally create a business as a young person, and invent something because you have 3D printing accessible in your household. That just wasnāt a thing when I was a kid. You can invent a thing as a young person and make a small fortune.
Yeah. And also, we talk about AIā¦ Itās a really interesting question, we now have people coming into the workforce who are out of college, or maybe theyāre self-taughtā¦ Who is the biggest beneficiary of AI tools? Letās just take Copilot, or alternatives; Sourcegraph has Cody, thereās Tabnine, thereās all these other ones. Clearly, it can speed you up. And my view is that thereās two groups who will benefit hugely from it. One of them, surprisingly, is the really experienced developers who master it, who already know exactly how it works, because they will spot immediately when it makes a mistakeā¦ And it does hallucinate. It helps them context-switch a lot faster. I havenāt developed anything in TypeScript for ā I havenāt really touched it for much, but I built a website on the side and I just use ChatGPTās generator. I knew what I wanted, I just didnāt know the syntax, and it just really helped me.
Simon Willison, whoās either the inventor of Django, or one of themā¦ Heās a very well known software engineer. He has gotten into AI, and using ChatGPT, and doing a lot of cool stuff in the spaceā¦ And he actually said that he feels heās about 20% more productive, which is a huge deal, because heās a very productive software engineer. Heās now independent. And he said it just makes him a lot more daring. Heās now using [unintelligible 00:52:36.10]
So senior engineers who master AI tools will be at a huge advantage. But also, I think people who enter the workforce, who are kind of AI-native, and figure out how to make the most of it to just get up to speed a lot quickerā¦ And not just fully ā like, learn from it as well. I think these two groups are going to ā potentially, theyāre gonna overtake the group in the middle, who was like āWe have some experience, but weāre not sure about this AI thing.ā
So if youāre a software engineer and youāre not playing with AI tools, of how you can be more efficient, where its limitations are, you will be left behind by this younger generation who is starting. Day one, they literally start with ChatGPT, or whatever else they do, and theyāre gonna very quickly put together ā build, mockā¦ Okay, theyāll make mistakes, but theyāre going to learn. And again, donāt forget that this generation is ā theyāre sharp; they context-switchā¦ And theyāre also thinking in terms of problems, product problems - not necessarily software engineering problems - and they want to solve for it.
Now Iām afraid that theyāre gonna take our jobs, you know? Heck with the AI. The kids. The kids are gonna come take our jobs, Adam.
Isnāt that how it works, you know, anyways?
Thatās how it should work, yes.
That is how it works. Well, letās not bury one of the leads here, which is your bookā¦ I havenāt had a chance to thoroughly read it, but I did catch some of the chapters youāve put out there. So thank you for putting that Look Inside document out there, because I was able to glean a little bit. Who is this book for? Whatās the title? Because I didnāt say itā¦ Who is it for, and how long did it take you to write it?
[00:54:08.21] Yeah, so the title is āThe Software Engineerās Guidebook.ā And this was the book that I had been writing for a very long time. It was a book that I wanted to write while I was an engineering manager at Uber. So the story of this book is that I spent 10 years growing as a software engineer from the entry level positions. I was at some point a principal engineer at Skyscanner, and then a senior engineer at Uber. So thereās the different levels. And at Uber I was a manager and I helped people get promoted to the senior level, staff engineering levels, and so on, on my team. And I just learned a bunch of stuff the hard way.
For example, promotions work, but the expectations are at a company like Uber - and Uber has similar expectations to the likes of Google etc. And I was just mentoring people a lot on how to become better software engineers, on how to go for example from a senior engineer level, where youāre expected to not only code, not only mentor, not only get things done and unblock your team, but at the next level you need to think more of how the business works, you need to coordinate teams, you need to lead with influence, and not titles, and so on. The time I decided to write this book is when I became a skip-level That meant that my team was too big. I now had a manager report to me, who had another, I think, six people. So I had a team of maybe 20. 14 at the time reported to me, and six people to this person.
And I was having a one-on-one with one of the new joiners, who was now my skip-level. And this person asked for some advice on āIād like to potentially get to the next level.ā I think he was a software engineer too, and āHow do I get to senior level?ā And I was thinking to myself, āI could really help this person, but Iām no longer their manager, and itās not really appropriate for me to step over my manager. Thatās their job.ā But I do have a lot of expertise and experience to share.
And especially when I talked with people outside of Uber, I sometimes had lunches with developers - they were just really confused on what it means to be a senior. I got into an argument with a person who said āNo, mentoring shouldnāt be part of what a senior does.ā Iām like āYes, it does.ā At a place like big tech, or even a larger scale, itās absolutely a part of it. And so thereās a lot of confusion.
So the book - Iād like to think itās for most software engineers, but in reality itās a very good book for entry or mid-level software engineersā¦ And also more experienced software engineers who have not worked at big tech. The book is probably the least useful for people who have worked at big tech or large-scale startups, and theyāre past the senior level. Itās not really for people who have a really good mentor, who can navigate them on what it takes to grow at these companies. If youāre at a company that has a really good career ladder, and a good manager, and you get good feedback, this book is not gonna help you as much.
And a third group that itās for - itās for managers who want to help grow their engineers. Because the interesting thing is ā now, this book is kind of partially written as my experience as a software engineer, kind of going up the ladder, and just becoming a better professional. And as you become a professional, you also grow on the levels, you get a fancier title, you get a bit more money, you get more responsibilityā¦ But then as a manager, I was on the other side, I was helping people get to there. And honestly, I just kind of wanted to document āWhat does it means to be a great software engineer at a mid-to-large-sized company?ā What I think it is; what are important things that you should know aboutā¦ For example reliability. What are some of the basic concepts, like the percentages, p95, p99, p50ā¦ What are some on-call practices that you should just know about, that again, if you worked at a large company that has these, itās gonna be like āYeah, I know all of this. We deploy with feature flags, we have automated Canary deploys, we have all these tools, we have platform teamsā¦ā But if youāre not, I kind of tried to collect the things that were all a-ha moments to me.
So I hope this book will democratize a little bit of how those cutting edge tech companies work. Think of Amazon, think of Uber, think of fast-moving companies like Stripe. And the interesting thing is as I was writing this book - it took four years to write. I started in 2019 at Uber, even before we had our first Changelog recording. I left Uber in 2020. I wanted to finish it, but it took a bit a little bit longer. And then I started my newsletter in 2021.
[00:58:07.16] And as I wrote my newsletter, I really got involved with the industry, talking with different companies. For example, I did a deep-dive on Facebookās engineering culture, on Amazonās engineering culture, and I started to see these gaps that āOkay, this book should cover this, and this, and this as well.ā
So in the end, I found it really hard to balance between how deep to go into the book, and how broad. I ended up going really broad, and leaving a lot of breadcrumbs, like āHereās things that you can go into.ā I tried to go over everything that is relevant for a software engineer. And so far, the reception has been very good.
The biggest criticism comes from people who are really experienced in big tech, who have been working at startups. A CTO, Will Larson, who worked at Uber, worked at Stripe, worked at Calm, heās now the CTO of Carta, he thinks this book is really good for entry-level and mid-level people, and people who have not moved around a lot of companies. And I think heās right. If youāve been in the industry, if youāre a veteran, if youāve been at these companies, itās more of a refresher and a framework. But the feedback from people who have been in this industry for 20 years, working at consultancies, working at (I would say) mid-level tech companies, they said āOh, my gosh. I would have needed this. This would have sped up my career with yearsā¦ Because now I finally know what I need to know to get into these higher tier companies.ā
For sure. Guidebookās a good name. I like chapter 18, āStakeholder management.ā I think this is like, if youāve been there, you kind of knowā¦ But what does it mean to have stakeholders? What does it mean to work with them? Do you have to have a good relationship with them? And I think what youāve done is kind of normalized what is the expectation of being in software development in a large organization. Because you may not know how to deal with a stakeholder. What inputs do you need from them? What kind of relationships should you have with them? Whatās the ultimate goal of the relationship? And I think when you kind of answered that in this particular chapter - you need that. If you havenāt been there, this is like āHey, when you are a senior engineer or a software engineer in a tech company like this, these are expectations of what youāll experience. And it may not be exactly this, but a version of this, and this is normal.ā
It was my goal. And one thing I wanted to do is - Iām going to be honest; I worked at a lot of different companies, which maybe helped my view. So my career was in Hungary I worked at a local consulting company, and then moved to UK to a small local consulting company. I then moved to JP Morgan, which was a bankā¦ But it was not really a tech company, but it was a more prestigious company. I was at Skype, which was somewhat of a late-stage startup at the time, and then it became Microsoft. I went to Skyscanner, which was a 700-person scale-up, a European scale-up, and then I worked at Uber, which was just very high growth at the time, and it was a true Silicon Valley company. When I joined in 2016, it was the place where people were declining Google and Facebook offers, probably until like 2017. So there were some really good people joining there.
And especially when I got to Uber, I just didnāt understand a lot of things. There was a lot of vocabulary that I didnāt understandā¦ I was sitting in meetings and I was just making notes of āI need to look up what does this thing mean.ā Even simple things like a one-on-one, which is a one-on-one meeting with your manager, which at most companies, or at better companies, you have it every week. For example, I had that at Microsoft, but it was really weird. My manager didnāt know what it was, we just did it, and I didnāt realize that, well, to do it well, you as the employee should come prepared, and say āAlight, hereās, manager, what Iād like to talk to you about. Iād like to talk about my career. And hereās the stuff that I did.ā You should showcase your work, and then some of these tips.
So for example, with stakeholder management - again, I worked with stakeholders for many years, especially as a manager, and I just saw what the great engineers did. And so Iām trying to give the vocabulary and the structure of how you can think ā you donāt need to follow it, but it gives you ideas. For example, with a stakeholder, most people ā the stakeholder can be your product manager, or the legal team who youāre working with. And a very simple thing is, especially when youāre at a staff level engineer, itās kind of an expectation, you just sit down and ask them, āWhat do you do?ā Like āHey, you work in a legal team. Can you tell me what part of legal do you do?ā Because then theyāre gonna explain it to you. And again, most people get around to this after a while. You ask them about their challenges. You just do some small talk with them, like āHey, do you have any kids?ā etc. And again, these are things that usually it just takes a while to figure out, because you start talking business. And if you do these things, youāre gonna be way more efficient.
[01:02:20.20] A tip that, for example, I learned very late is asking them to shadow on a meeting. Like, literally, the legal team you would ask ā theyāre having a meeting, and like āAs a software engineer, can I go into your meeting and just sit in the corner, see what you guys are talking about on the legal meeting?ā It makes no sense, because youāre a software engineer. But when youāre at the staff level, and you need to understand the business, this is one of the best ways you can do it. And again, most people, including staff [unintelligible 01:02:43.13] just donāt do it. And if you do it, youāre gonna be way more efficient.
So Tanya Reilly, who wrote a really good book, āThe staff engineerās pathā, which is an amazing book for staff-level engineers, she just told me after she read the book that she just feels that this book will democratizeā¦ When people sit in meetings, they just ā āOkay, I know what this means. I know what that means. Or if not, I can look it up. But I feel that I know whatās going on in this space.ā And I tried to kind of break it down. Everything I felt was intimidating, or I didnāt understand, I just put it in the book.
So the book is 413 pages across 27 chapters. How do you know when youāre done? How did you know like āThis book is finishedā?
So hereās the thing, Iāve self-published. I originally wanted to write for the publisher, I had some ideas, I had some ā I really hoped that I would work with one of these really big brands that I look up, and some authors [unintelligible 01:03:35.02] back in 2019 I submitted this topic, and I had like three kind of top book publishers in mind. The first two rejected it, and they rejected it in a nice way. It was a close call, and they just felt that they had a competing title, or something like thatā¦ Which is how it works in the book industry, by the way. And then the third one said āYes, we want to publishā, but with this third book publisher, they actually turned out to be pretty opinionated, and they wanted to make it a little bit more beginner-friendly, and they wanted to put some structure in place that I just didnāt like. So in the end, I was like āOkay, I feel like Iām gonna fight with my publisher, and itās just a lot of energy. Iām just gonna write myself.ā
Self-publishing is great, because I can do whatever I wanted, and I was pretty opinionated, so I wasnāt worried about whatās rightā¦ But you never know when youāre done. This was a problem. And I was writing on and off for about three and a half years, and finally, I just ā and I was writing my newsletter as well. And I resisted the urge to recycle things from the newsletter. Thereās very little overlapping content. There are a few chapters; of the 27, I would say maybe thereās four chapters that have been published in some form in the newsletter, and theyāve been reworkedā¦ But I just kept it separate. Because a newsletter is very āwhatās happening right nowā, and I want to write his book as the stuff that is going to be irrelevant in five years, and then I will have to come up with a new revision.
So three and a half years I had a lot of stuff written, and I decided Iām just going to give myself a deadline, which is what a publisher would give you. At the end of the deadline I had about 500 pages, or 550 pages worth of textā¦ So the last month I spent cutting it off, and I just decided to put 100 pages as bonus chapters, which are available for free, because I wanted to keep the book at a reasonable length.
And by the way, these 400 pages - I did a trick. I did the largest print I could, because this book is about twice the length of some of your engineering management books that youāre used to. Itās about the same length as āDesigning data-intensive applicationsā, except āDesigning data-intensive applicationsā, another popular book, it does have a lot more formal things. Mine is a bit more on the soft side.
In the end, Iām happy with it. Itās not a book that youāre going to sit down and read the whole thing. Itās more of a reference book. Itās like you open chaptersā¦ You know, Stakeholder Management. Team Dynamics. āI just became a software engineer, I need to get things done.ā Thereās a lot of getting things done. Iām trying to make it really practical. Itās pretty much the advice that I gave to people at Uber, so Iām kind of hoping that if you pick up this book, itās a little bit like I was your mentor a little bit. Itās not as good as if you have a mentor. Please, try to get a mentor. Thatās the best you can do. No book will do it justice. But I just hope it gives some structure, it gives some ideasā¦ And so far, thatās what Iām hearing.
[01:06:12.29] And by the way, if youāre reading this book and you have feedback, or criticism, also just shoot it over. I donāt really plan to write another book anytime soon, but I do plan to improve this further in a few yearsā time, so Iām going to collect whatever might be missing, Iām just hoping that this is going to be on peopleās desk, and theyāll reach for it and say āYou know what, itās given me a couple of good ideas. And I tried them out, and it just saved me a month or two, or even a year of me figuring this stuff out.ā
So no new books soonā¦ Anything else coming down the pipeline or youāre working on, in addition to āThe Pragmatic Engineer newsletter? Anything we can look forward to?
For now itās the newsletter, and Iām just going to chill a little bit for the next month or twoā¦ Because writing this book and writing a newsletter was a lot of work. And I want to get some other version of the book out. Right now itās only paperback, which might be very surprising; but because Iām self-publishing, doing a Kindle version is a bunch of extra work, which I am going to doā¦ And doing an eBook and an audiobook is also on my plate. I just wanted to see how the paperback goes. And Iāll be honest, one of the reasons I did the paperback first is Iām really hoping that this is a reference book that can be on peopleās bookshelf.
I know people have a strong preference for Kindle, so thatās the next version thatās comingā¦ But Iāve yet to go to someoneās house and say āOh, I really love your Kindle collection.ā But I have gone a lot of times and said āWhat is this book? Can I borrow it? Can I take it home?ā So Iām kind of hoping thereās gonna be a little bit of this with the book.
So Iām just now making sure that thereās more printability, and in countries where ā Amazon is printing this right now, but now itās also on IngramSpark, which means that individual bookstores will be able to order itā¦ Iāve learned a lot about self-publishing. I plan to write a post about that, hopefully help other people who are thinking about that.
Yeah, thatād be a good one.
For sure. Well, speaking of Satya Nadella, whenever he was on that road show that you mentioned, back to the Open AI conversation, I noticed a bunch of books behind him, and I was like āManā¦ā I paused it, and Iām like zooming in, and looking at all the books, and Iām like āIāve read a couple of those.ā And one day maybe your book will be visible in a future roadshow, saving Open AI, or a version of it, in the future, behind the scenes of CEO of Microsoft, you know?
Thatād be cool.
Even if not Satya Nadella [unintelligible 01:08:22.00] Itās already on a lot of peopleās bookshelves. I was very happy with the reception. And there was a little bit of a validating thing about this whole thing, because in the end, to publishers, this was different. This was four years ago. I wasnāt as well known. Iāll just be honest. I did have a blog; people were reading The Pragmatic Engineer even back then. But two publishers ultimately said āInteresting, but itās not for us.ā Basically, what they said is āWe donāt believe this will be great business for us.ā Because book publishing is about the business. And the third one said āWe think this will work, but you need to make a lot of changes to it.ā And I just stuck to my guns, and I wrote the exact same book as I pitched. The introduction is the same as I pitched four years ago. I did make some changes, but it was just validating to see that itās doing very, very well, both in terms of sales. It jumped to number one in software engineering for a while. I think it still might be there. For six days it was the most sold book in the Netherlands across Amazon; Amazon Netherlands, above all childrenās books and everything. In the US it went up to number 30, or number 32 on launch day, which is across all books sold in the US, which is, again, a big deal.
So it was just very nice to see that yes, there is demand for this. And I have been getting feedback that people do like that I havenāt simplified a lot of things. I havenāt made it a bit more verbose, which was a suggestion that a publisher gave me. And again, they had really good intentions. This is what theyāve seen sell. And since then, Iāve had one of the biggest publishers in the world [unintelligible 01:09:46.23] Penguin Random House wanted to talk after a launch, and they asked me āWould you be open to writing a book for a bit more generic audience, for soft skills?ā And I said āNo. I just want to write for software engineers.ā And they said āWeāre not interested in just publishing for software engineers.ā Iām like āI know, but Iām only interested in writing for software engineers, and Iām only interested in writing stuff that is not a beginner.ā Iām trying to give stuff that is more advanced. And I believe thereās a market for this, and I think now there is.
[01:10:13.03] So itās been nice to see that this book that I deeply felt was missing for me - itās nice to see that other people feel the same way. So itās one of those thingsā¦ So Iām really grateful for all the readers, who are both buying it - because a lot of them just bought it honestly blindly, because I guess they knew me for a whileā¦ But now the feedback is starting to come in. And again, Iām really looking forward to critical feedback as well. I think thatās the thing that I ā I donāt like to feel that Iām done. This book, I donāt think itās done. I think thereās going to be new versions coming out of it. Iām going to improve it, and I want to keep up to dateā¦ Because the industry is changing.
This is the first book, probably. Well, maybe not the first, but the first wider-sold book. AI coding tools are inside of it. I made that change six months before publishing, because right now, multiple chapters, āHow do you grow as a software engineer? Well, you pair, you get mentorshipā¦ā You use AI coding tools as well. If youāre not using AI coding tools to improve, youāre already left behind.
I mention things like cloud development environments, which are now spreading in big tech, and some of these other things which are pretty newā¦ And I want to keep the book later on updated as well with the new technologies that are spreading pretty decisively. Developer portals, like backstage, which are common across the big tech, are again, inside the book. So right now Iām really proud of it, because I think it really describes what is cutting edge across large tech companies.
Yeah. Well, if anybody can keep it up to date as things change, Gergely, youāre the man, because you are always up to date with whatās going on in the industry. If you need help with the audiobook, I know a guy whoās got a good voice. And itās not me, itās Adam. So maybe we could ā thatād be cool. A collab on ā would you narrate a book, Adam? Would you ever do that?
Would you lend your voice to somebody elseās words?
It sounds like a neat idea.
I could do it any day. Tomorrow.
Just an idea. Planting the seed for future collabs.
Anything else, Gergely, that we havenāt covered, that you want to talk about before we tail out here?
I think thatās mostly it, really.
Adam, are you still with us or are you just daydreaming right now aboutā¦?
I was thinking about whatās left. Thereās one thing I want to leave for Plus Plus, that I think we should at least touch on. That way itās a smaller audience evenā¦ So hey, if youāre not in the Changelog++ arena, it is cooler there, because your next to that cool Changelog metal, as we sayā¦ And thatās where you want to be.
Thatās right.
So weāre gonna ask Gergely a question here in a secondā¦ But for now, Gergely, itās been fun catching up with you. Itās been fun seeing you again. I canāt believe itās been a year since the last time we spokeā¦ And I always appreciate your perspective. And I think just your genuine nature to find the truth in whatās happening, and to share that. I love the way you produce your newsletter. I think you do it very honestly. Thereās a lot of newsletters out there who have just ulterior motives, that just are strictly financial in some cases, or audience growth in some casesā¦ And I think that your commitment to being real with your community, in the community, is refreshing, and we like that.
Absolutely.
Itās why we have you back. We love talking to youā¦ So thank you for that.
Yeah, thanks for that. It was great being on.
Until next year.
Yeah, until next year.
Bye, friends.
Changelog
Our transcripts are open source on GitHub. Improvements are welcome. š
Recommend
About Joyk
Aggregate valuable and interesting links.
Joyk means Joy of geeK