Companies because I use Google as an example. If I went in maliciously as an SRE there, I don't think I could take down google.com for one full hour. So if I can't do that maliciously and intentionally, how much upside good am I going to be able to do in a place like that? Or is it a small business like, oh, if I screw this up enough, we don't have a company anymore? Right, that's the level of impact I want to have. Not that I'm setting out to screw up, but I want to be able to materially change the fortunes of places I work if I do it right.
Hello, hello, hello. Welcome to episode 32 of the Metacast podcast. So today we have a special guest with us, Corey Quinn. Corey is well known in the cloud computing circles. He writes the Last Week in AWS newsletter. He has a Screaming in the Cloud podcast and awesome Twitter. If you don't know, go check it out at quinnypig. We'll also list it in the show notes. But Corey, before we hand it over to you for a better introduction, I wanted to tell you a little bit about the podcast.
So Iliya and I, we worked in AWS together for five years, and then I was there for about like 12 years overall, right? So you probably know already in Amazon, there's that culture of the empty chair, which is a symbol in every meeting, the customer not being in the room, but every decision you take is supposed to be considering the customer who is not there in the room, hence the empty chair.
Yes, and then they moved the empty chair to the CMO's office for two years, but that's besides the point.
Yeah, yeah, yeah, yeah. But I think for a lot of us, when we were building services and thinking about like big customer facing decisions, that share was personified by you. And I'm not joking.
I would hope so on some level because I think that's something that gets missed is that I'm not coming up with this stuff necessarily out of thin air. I'm also usually not the only person who experiences something and has a certain reaction to it. I just have no filter. That's really all it is.
Yeah, and before I forget we have Ilya, my co-host here. We were supposed to record yesterday, but thankfully we didn't have to because he would have been recording from a car. That would not have been nice for anybody. Well, I'll...
I don't mind if he records from a car, but could he at least slow down on some of those turns? My god, I'm starting to worry.
Let's do an actual formal intro of yourself. Let's tell us who you are, what you like doing.
Actually before we go there, I think we have to say that we were looking for words for how to introduce you because I guess a lot of listeners for this podcast will come from you when you hopefully retweeted this. But our listeners who like are native to our podcast, not all of them will be familiar with your work and what you do, or even AWS, because they might be, you know, Oracle customers or, you know, and they may be on threads, so they may have been missing on your persona altogether. But yesterday, we asked Chad GPT, do you know who Corey Quinn is? And surprising?
Actually, Chad GPT, well, maybe not surprising, Chad GPT knew who you were. And he gave us a very good answer, saying that Corey Quinn is a well known figure in cloud computing community, particularly in AWS, yada, yada, yada, it was very on point, like exactly what you do. And then we asked it to introduce Corey in Corey Quinn's snarky voice in town. And that's what came out of it. I don't know if you want to read it.
Sure, yeah. Ladies, gentlemen, and all you lovely misconfigured S3 buckets out there, get around for a dose of sarcastic wisdom that's more cutting than an IAM policy gone wrong. And there's more, but we'll stop there. And I think it's very on point, yeah.
It definitely makes an attempt. I don't know that quite gets there most of the time. Believe me, you're not the first person to try to get it to say things in my voice. It misses some of the nuance. And you know, that's honestly I think that's probably a fair failure mode. I got to learn from history as an engineer who just finally got tired of not saying what I really thought and getting fired for it. And figured, alright, what if I lean into this?
And embrace all the things that everyone who's known me for longer than 30 seconds has said would be my ultimate downfall, my personality. And seven years in, I consistently surprise myself by failing to get it too wrong. But there's always tomorrow.
Yeah, but I think your personality is actually and the authentic personality that like you said comes out on Twitter on these blog posts and everywhere. That's actually part of the main draw. I feel on the contrary of like not losing your job.
Partially that. It's also because my largest customer is a single digit percentage of revenue. I don't have any external investors and I'm not an AWS partner. So really, no one can fire me. It's kind of great.
Corey Quinn
of Cloud Economist.
Sure, I talk to large companies that have expensive AWS billing problems, either that the bill is too large or that the bill is inscrutable or that it's hard to predict. And I believe firmly that costs and architecture in the cloud are one and the same. So there are engineering optimizations that get made and for minimal amounts of work can cause significant levels of cost savings. There's also ways to do allocation between different internal initiatives and figure out what projections are going to look like. And this becomes key, not just for your own internal budget.
But also for negotiating with AWS and those long term contracts. That negotiation takes up about half of my consulting day. It's kind of fun.
So is the economist, I guess, a bit of a misnomer because you actually do a lot of technical investigation that results in cost savings, right?
What I started doing is no one else was really in the space in the same way. So I combined two words that are often misunderstood. Cloud, meaning a bunch of other people's computers, and economist, meaning someone who claims to know a lot about money but also dresses like a flood victim. So you smash the two of them together and suddenly no one knows what the hell I do or how to categorize me and that was convenient for a time. When people didn't know what to call you, it became much more difficult to bucket you and thus dismiss you in some ways. And early on that was important.
Now you've established Cloud Economist as your own thing because for me at least, Cloud Economist is almost synonymous with Corey Quinn because I don't know any other cloud economists. So well done.
I found multiple companies hiring for the role with that title. There's now an AWS cloud economics division, which is kind of fun. I also met someone who apparently had a PhD in very similar aspect from 2012 and he's like, there's another one I thought I was the only one and if I'd bluffed just a little bit more I probably could have wound up with a book deal out of it. But you know, here we are.
So take us back a little bit. Like you said, you started as an engineer. How did you land up in that space? What were you doing? This is, I'm assuming, before there was a cloud.
Oh yeah, I started off as a grumpy Unix systems administrator. And the job never really changed. People argue with that, like, no, I'm a DevOps engineer, or I'm an SRE, or I'm a platform engineer. It's all the same job. The tools change, the methodologies change, but the responsibility doesn't. It's keep the things that plug into the wall, whether it's your wall or someone else's, running. Keep the site up and ideally be able to predict the future and not get it wrong.
If you get everything right, no one knows you're there. If you get it wrong, everyone blames you. It's really the definition of a no-win job. That's infrastructure. And one thing led to another. I focused in different areas at different times of my career and finally wound up dealing with cloud. In my last job, I had yet again to deal with a surprise AWS bill inflicted on me by other departments. And I wished there was someone I could pay to make that expensive problem go away.
Couldn't find it. So when I decided to hang up my own shingle and figure out what I was doing, this seemed like a decent business problem to take a whack at.
That's very interesting. So were you employed by somebody in that job?
My last job was at BlackRock. My personality did not fit in as well as you might think in a large regulated finance company. Why did you choose to work here in the first place? You bought the startup I worked for. I was not consulted on this.
Is it BlackRock or a Daedalus company? Is that the company?
They are a giant finance company with a bunch of different divisions. The answer to everything is it depends. One of the challenges I've always had with very large enterprises has been the lines of communication internally are challenging at the best of times. I'm not entirely kidding when I say that half my job these days is introducing Amazonians to one another. Yes. People are hard. They're always the hardest part of anything. Computers are at least deterministic, ideally. I understand.
Yeah, we often talk about this ourselves now that we are outside that world. We can see all those like where those shim lines are fractured. Like let's say we're using something, I forget what it was, Ilya, like something from Google the other day. And the signup form was like completely unlike any other Google signup form that I've ever seen.
It was a Google Play Store form that could not recognize a phone without dashes.
Yes, if I put the phone numbers with the dashes, it's like no. And I'm like, this is Google? It was very strange, but it was immediately clear that these are like two different, completely different divisions. They don't know about each other and all that. Yeah. And this is normal in a big company.
Oh yeah. The forms drive me nuts on all these sites. One of the things that drives me nuts is AWS is on stage all the time talking about the value of data and personalization and understanding and knowing your customer. But I have to fill out the same giant pile of mandatory fields every time I want to attend an event or a webinar and I've experimented with this. It does not matter at all what I fill out in those forms. I do not get any different outreach or treatment. Once I said I was
a Swedish VC, I have gotten no startup pitches in Swedish. I don't know what to make of any of this. The problem that I have is not just the annoyance having to fill out the form, but it just feels disrespectful of if you're going to ask me a question, do me the service of listening to the answer, at least acting like it. But no, here we are. Marketing's hard.
It's essentially like data gathering going on, right? And somebody has probably thought, okay, we'll end up using this data someday somewhere, but nobody has used it yet.
That seems to be a big push by the cloud providers. You need to collect all the data you can. Why? Because we charge you for gigabyte per month on that. So yeah, store all the data. That's probably not the actual reason, but it's hard to argue against it.
Yeah, so Corey, one of the topics we like to talk about with our guests is, because we recently jumped a big tech corporate ship to do our own thing. And you did the same thing seven years ago, you went from BlackRock into around your own consultancy. How did that shift happen? And yeah, we'd be curious to understand like how the transition went.
Oh, I got fired because of my personality. And at that point, I couldn't stomach the idea of doing the same thing yet again for different companies. Because let's face it, when you run the systems, it doesn't matter nearly as much as you might think what the company actually does. Are you selling boxes? Are you streaming bits? That's basically as far as it goes. What are the tolerances and constraints around things? And that problem just started to look the same again and again and again.
What made me a good consultant was the same thing that made me an amazingly great employee for the first three months. While I come in and everything's on fire, it's great. I know how to fix this. And sure enough, I would get to a point of a lot of these problems being fixed. And then, all right, now maintain it forever. That didn't work very well. So instead, I found myself in this unfortunate place of having to figure out just what the hell to do now.
And then finding the fun problems involved me straying into other people's lanes. Not recommended.
So did you like find your, I think, co-founder, right? Business partner.
Yes, oh yes, we've been friends for 15 years at this point. I think that people take the idea of business partners far too lightly. You see this with lightning dating for founders and whatnot. Realize it's like a marriage in a lot of different ways. You are forming a legal entity together. You are able to financially just ruin the other person. And you don't want them to be too much like you or you basically have two people doing the same thing.
Mike and I are nothing alike in most ways. My first language spoken at home was sarcasm, whereas Mike's love language is Microsoft Excel. But where we wind up aligning is that there's no daylight between us when it comes to our values. And that's one of those important things. Because if one of you believes in treating customers well, the other one believes in taking every penny you can get from customers this quarter, because who knows, tomorrow may never come.
You're going to have an awful lot of friction. It's important to be aligned on a lot of those baseline things before there's money on the table.
So what was Mike doing at that time? How did you like convince to jump on and start this thing?
He had just written a book on... practical monitoring was the title for O'Reilly. And he was an independent consultant focusing on the observability world. It turns out that if you really want to hate what you do for a living, write a book about it. That convinces you that you want to go find something else. So I convinced him to come run this place. And we wound up merging a new entity and here we are.
Hold on, what is it in writing a book that you hate what you do? Because I've just written a book.
So common misconception is that people want to write a book. I don't think that's actually true I think they want to have written the book So I wound up figuring out one day that I want to do it I don't have that kind of attention span or that kind of that number of words in me And I believe that for a long time until last december for the holidays Mike and my wife wound up and a few friends wound up pitching in and buying me a series of books
Uh leather bound called the complete works of Corey Quinn. This is just a printout of my tweets from 2021 That's a multi-volume set and it's it's psychotic. It's a list of all the stuff I put out there It's over a million words. I guess I did write a book
Can you open a random page from the book on your desk and read it?
Absolutely do that, it doesn't matter. There are times I wind up putting all kinds of weird pictures up there. They turned out the retweets, all the rest. Everything I put out there, like pick any random day and look at me on Twitter, no we're not going to call it X. And that's what it turned into.
Weird stuff. Do you know that book Shit My Dad Says? Or Shit My Dad Tweets or something like that?
Oh yeah, it's funny in fact the picture I just pulled up there was a picture that I captured that someone had a... This is back in the early days of the mid-days pandemic, it was a Rorschach's blot on a mask and my caption was, Ooh, I can buy a mask with an awesome image of my parents fighting on it. Yeah, you know, full stuff. Just the psychotic stuff that pops into my head I never had an outlet for before. Well, I do now. At least I did until a jackass bought it.
I love this idea of the book. Is there a great essay? Are there like, love, likes and retweet counts and all that on there?
No, I pay almost no attention to that because my jokes are for me. I care a lot more about, is this humorous to me? Does it work? It's not for everyone. And I think that's something that a lot of people miss when they're trying to find their voice is they try to appeal to everyone. In doing that, you're going to appeal to no one. Also, no matter what you do, some people are going to just flat out not like you. And if you're a people pleaser, that gets unfortunate, but you also can't
avoid it. So you have to figure at that point, you want people to dislike you for who you actually are, for who you pretend to be. For me, it wasn't much of a choice.
So do you have a setup like Greatest Hits there for yourself?
Not particularly. I'm not one of those who sits alone at night chortling at my own jokes and how clever I've been.
So I'm curious when you set out to write a joke. So first of all, is there a routine or do you just like randomly do it whenever an inspiration comes or you have like a specific time and day or something?
I tend to take the perspective of just like whenever I wind up working on some particular area of sitting down to write a blog post or whatnot or coming up with anything, a blank page is scary. It doesn't give you a whole lot. Most of my best humor comes from riffing off of other people. My humor, I think, took a bit of a dive during the pandemic just because I wasn't in social settings with people for a while where things come up. Like you say something and someone misunderstands like, sorry, I thought you said that Route 53 was a database. It's like, that's good. It is now.
It's the weird one-off conversations that hit and you just jot the idea down and investigate it later.
That might have been actually one of the first few things where I knew your name before because I was already working at AWS, but that the Route 53, like use it as a database post that got me hooked into the whole like the podcast and all the newsletter and all that. That was awesome.
I mean, at some level I realize that anything can become a database if you hold it wrong and take a sufficiently liberal view of what a database is. I mean, hell, you have information I can ask you about. Congratulations, you're a database.
Yeah, we should link to that episode from our show notes. Because I remember that was the very first episode I listened to because you just kept talking about Route 53 as a database on Twitter. And like, what is he talking about? Yeah, because you were not explaining what it was, you were just using it as a side thing, right? This is
This was a common problem by the way early on because when I started independent consulting seven years ago I had something like 1500 Twitter followers and it had taken me seven years to get there
These were mostly people who'd seen me talk at conferences and who I knew. They had the pre-existing context where I could say something and have the implicit under assumption that everyone reading this already knew who I was, my shtick and the rest and at some point you grow beyond your own sphere when your audience gets big enough and no one has the
Context. People look at it now and think this guy's a fool if he thinks that's gonna work as a database. Well, no kidding I found that out when I had to break character and I saw a reddit post once upon a time when someone's saying yeah
I get he's kidding, but this actually doesn't sound like a terrible idea. It's like hi Let me explain to you exactly why it is because we did this when 2009 in production to figure out which rack servers lived on and which server a VM lived on there are better tools for this job What is the problem you're attempting to solve for?
Yeah, I guess basically what happens here, right, is that your early followers who have been with you for years, they know all the context, and they know what's joke, what's not. And they also kind of get the jokes, I guess, at a deeper level. But then some of the newer people, they actually might come to you almost like a TikTok consumer, right? You just like saw a tweet, laughed a bit, swiped, go next. Do you think this changed? What kind of engagement you have? I don't want to use the term engagement in that kind of TikTok, Instagram way, but more like, just to give you an example, podcast usually gives a lot deeper
engagement with people because they would write you like a very long email. They get very intimately connected with you. Whereas tweets might be more kind of ephemeral. I'm curious if you see different level of that kind of connectedness with your early fans versus people who came after that first 1500.
Oh yeah. I've had my DMs open for almost the entire time. And because it's, if people are actually curious about something, I am thrilled to wind up entertaining anything. It still surprises me. When, again, talking to Amazonians from time to time, like, I really didn't like that thing you wrote, but I didn't want to say it because I didn't want you to drag me in public. It's like, you've got how many years of history now to look back on and never see me doing that unless you're Larry Ellison? What exactly? Yeah, that's it. Some random software engineer, and some for some service.
And AWS, yep, that's what I'm going to cash all this in for just to make you look like an idiot. Yeah, that's going to go well. No. And the biggest problem I run into with most of the stuff I put out is silence. I don't get a whole lot of feedback on certain things. If you hit reply to my email newsletter that goes to something like 32,000 people now, it goes straight to my inbox because less than a dozen people do it on any given week. It's easy to work with. The podcast, for the first six months that I ran that thing, I was half convinced I'd forgotten to turn on the microphone or something. Then I went to a podcast.
I was at a conference and a lot of people talked about it and it was, okay, that's interesting. My theory is that anyone can send an email, so at least it's easier to hit reply. And I send emails too. I'll give him a piece of my mind. But writing to someone about a podcast feels different. It feels like you're calling into a radio show. And who does that? Well, lunatics. And most people who aren't on Twitter don't self identify as lunatics. So it wasn't the way to go. Why do you think they're lunatics? Have you listened to AM radio shows? My God, the people who call in sound deranged. Like, I don't want to sound like that.
I'm just going to roll my eyes and keep on driving down the road.
What I think we've seen with our show, but also with some of our previous guests, is people who actually do write an email for a podcast host, it's almost like a letter. So it's almost like you read the book, especially if you listen to multiple episodes, and then you finally decide, well, I have something to say to this author, and I'm going to write a letter and it ends up being like two pages long email full of praise or like some questions and all of that.
And it feels super weird like you're taking a risk to do that as well. Yes. And because no one does it. There's also the question having before I started this stuff I used to do that myself from time to time. And the vast majority of the time I got no answer to it. And it was well okay at least I got it off my chest and said what I needed to say. Always polite. I've never felt the need to reach out to someone and tell them I hate what they're doing. That just what am I possibly going to achieve by doing that. I'm just going to be a jerk and make someone's day a little bit worse. But I find that it's an effort investment.
I'm going to be doing a response. Oh God do I sound like a lunatic sort of parasocial creeper or something if I wind up sending this. And the answer for most people is absolutely not. One thing I feel in terms of engagement that I have noticed as well is the way that I engage with large brands with the Internet etc. is not something I would recommend because I got it wrong a lot when I was starting out down this path.
The failure mode of clever is asshole as John Scalzi is famous for saying. And if you get it wrong in this direction you're just being mean. I've had a number of times I had to jump in to DM someone like hey heads up. I just want you to be aware of how this might be coming across and you probably don't see it because I didn't the first time I made a joke very similar. Let me learn from my experience so you don't have to make these mistakes on your own. And people usually respond to that pretty well. If you call them out in public they feel the need to double down and get defensive which is never the right move.
Now we had a few questions like this later on, but let's just like get started here. Did any company or any HR or somebody get back to you saying like, take this down or edit it or something like that? If so, like, how did you handle that?
No, I don't work for them. HR is only focused internally on employees. You have people who meanwhile will reach out and say, I don't know if that landed the way you think it will. But I've never gotten a letter from someone's legal department threatening me for anything that I've said. I have a pretty good sense of where the line is in that respect. And it stops well before we get into the defamation territory. So instead, it's much more about people reaching out as
individuals. I did have, for example, Forrester DM me once when I was talking about some stuff in one of their reports that someone had posted. And they said, well, posting screenshots of this is a violation of the agreement that we have for publishing these. It's great. One of your customers with whom you have an agreement published this. And I certainly didn't agree to anything you say. It's on the internet. Screenshots are generally fair view for this. So have a nice
day. And they didn't bug me again. Right.
Did you ever have to consult legal when you were about to post something? Like, how do you know where that line is? Because legal is very complicated.
I've noticed that. So I have been for the last 15 years I have that my wife, that's why I've known her. She's been an attorney the entire time. She used to be a litigator. She's now in corporate litigation as well as a lot of contract work. She's one of those high powered lawyers, like much of one of those schools that people like aspire to go to Obama was one of her law professors. I'm just some schlub with an eighth grade education, but she's actually impressive. So you got to pick up a lot of this stuff osmotically. You have a lot of conversations over dinner about a lot of
your main topics, assuming you're both interested in that sort of thing. And it helps identify at least a framework to think about these things. For actual questions I've got on these things there. We do have attorneys that we wind up paying for this sort of thing. But I almost never need them at this stage because if it's close enough to the edge that I need to get a legal opinion, don't go with it. The joke is almost certainly not worth it. Find a better one.
Right. Let's go back a little bit. You were talking about how before we got here, like how you knew Mike already and he jumped on, you decided to. I'm really curious about the name Duck Bill. So what's that from?
The duck-billed platypus is the mascot, Billy the Platypus for the newsletter and the rest. And I came up with that before I came up with the name for the company because it amused me. I wanted a cartoon animal I could have fun with and not get yelled at for. And the platypus is an inherently ridiculous animal, so why not? And it's as good as any other when it comes to these things. And then when Mike and I decided the future might be us joining the company, it's like, well, what do we call it?
Like, we're in that kitchen. He just looks up as the duck-billed group. Soul works for me.
Awesome. And where are you at today? Like how many employees do you have? What do various people do?
Number seven just signed an agreement, more to come on that in the near future. We've been as large as 12 or 13, but we're in that fun place where we're completely bootstrapped, which means we have this ancient old timey business model where we make more money than we spend every month. And that means that we have to be able to grow organically. And if you build out a services team in the wrong ways, you start having serious concerns. It comes down to utilization. We never bill our clients by the hour, for example.
We have to be paid, like, you know, payroll and our AWS bill and whatnot. And at some point you need to be able to project these things. And people sitting around idle for too long becomes a problem for the business. But that also means if you don't have the capacity, you can't take work that suddenly comes thundering. And it's a constant balancing act. I think this is part of the reason that people love SaaS so much. Scale is trivial and easy to solve for, right, until it's very much not. But at least there, there are clear-cut solutions.
solutions.
I considered doing that at the beginning. I thought that's the direction I was going in. But the problem that I've discovered across the board when it comes to at least AWS billing optimization is that it is not something that is nearly as math and API driven as you would think. It is overwhelmingly dictated by psychology. It's a people problem. And there is no API, legally, that you can install in people that makes them all behave a certain way or you can address them programmatically.
It's things that make logical mathematical sense. Like, okay, this is less of a problem these days, but earlier on, we talked to a customer. Great. You should buy $18 million of reservances this month. Go. But they sit there and they'll hesitate because it's more money than they'll likely see in their career. And if they screw it up, that feels like a dangerous thing. So they don't.
But they should because they're going to be spending that money one way or the other. You might as well do it at a discount. But the psychology of it is fascinating. One of my personal favorite areas is the sense people have around loss aversion. People would much rather have 95 percent coverage of their spend than 102 percent.
Now, 102 percent is mathematically going to be better for them regardless, but they are going to freak out about the overspend far more than they will about not having spent enough. It's kind of weird.
Is it the prospect theory that Daniel Kahneman was talking about? I think he got the Nobel Prize for this, where it's like people are... forget the number, but almost like 10x more prone to be risk averse. I guess the risk affects them 10 times more.
Boss aversion, specifically. Yes, it's partially that. It also flips in its head in some ways for corporate entities. If I told you right now that you need to come up with $1,000 for something and you can either make it additionally or you can cut your cost to come up with it, people will by and large buy us very heavily as individuals toward cutting costs. It's easier to cut Netflix and stop eating out than it is to wind up getting another job or negotiating with your boss for a raise or coming up with a side hustle. Companies, on the other hand, are optimized inherently towards growing your business.
So, growing revenue, entering new markets, building things and boosting the returns. Cost optimization is one of those things you feel like you should do on some level, especially when the market turns and people suddenly care about it a lot. But you can spend all your time optimizing costs or working on security or buying fire insurance for your building, but it's not going to move you one inch toward your next actual business milestone. It might keep you afloat long enough to do those things and you need to do those things, but it doesn't generate value directly on its own.
Don't do something correctly that businesses are inherently optimized for maximizing revenue and growth. Whereas people who work in those businesses, they are wired to be loss averse. And that creates kind of that problem.
As individuals, yes. For example, there are two reactions and it's differentiated entirely between is this corporate money or is this personal money. Fifty dollars for a hamburger. Go screw yourself. Fifty dollars for a hamburger. I'm going to need a receipt. People treat company money differently as they should. One of the most toxic pieces of advice I've ever heard people give around expense reporting and are you the corporate card. How do I decide what to treat what to spend this. Treat it like it's your personal money. You understand not everyone
comes from the same background right. There are people on food stamps and there are people with trust funds that you're telling this to and they're going to have very different approaches to how this stuff works. In my case it is my money is the question what pocket it comes out of. But you want to building mental frameworks to think about these things and reason about.
So you said that you are fully bootstrapped and we admire that we are fully bootstrapped, all of our previous guests were fully bootstrapped. Well, we are more than fully bootstrapped.
We're more like socked up right now. We're not bootstrapped yet, but we're hoping to be bootstrapped. Yes. Yeah
Have you ever had in your seven years running Duckbill, being with Duckbill, cases where you're like, you're not going to make it, but then you still kind of somehow made it?
The hardest part of running a company, bar none, is managing your own psychology. A lot of people say they want their executives to be very transparent about everything that's going on. They'd be terrified by the end of week one in an awful lot of shops if that happened. When I was independent just by myself and didn't have other mouths to feed, it was in the morning I'd be flipping through benefits services available just in case. And then something would happen in the afternoon, I'd be flipping through a page in the yacht dealerships just in case. And it's the constant peaks and lows.
I think it's a good idea for death because you're going to die lots of times if you do. Very few things are as critical long term as they feel like they are in the moment. There's an immediacy tied to it. And companies are disturbingly resilient in some ways. Like if I quit this company, it's screwed is never true in almost any case.
The only one I'm actually honest of an open question right now is like since so much of our media side of the business is tied up in me, if I quit the company that I own, the company is going to have some problems on that. But that's also always been a bit of a way that I would select roles.
I don't like big companies because I use Google as an example. If I went in maliciously as an SRE there, I don't think I could take down Google.com for one full hour. So if I can't do that maliciously and intentionally, how much upside good am I going to be able to do in a place like that?
Or is it a small business like, oh, if I screw this up enough, we don't have a company anymore? Right. That's the level of impact I want to have. Not that I'm setting out to screw up, but I want to be able to materially change the fortunes of places I work if I do it right. Which meant that long term infrastructure was probably not for me.
I think it's very good to have a framework because like the company protects you from downside, but you also don't get any of the upside pretty much unless you're an exec.
Even then, the world of difference between how much founders make in exits and employee number one. It's a universe apart. Take a look at founders versus like look at the compensation publicly stated clearly between Andy Jassy who's making hundreds of millions of dollars spread over a 10-year span, and Jeff Bezos who was for the time the wealthiest man alive. And it's orders of magnitude difference.
So talking about that, you said you focus a lot on the media side. What are the other people there? Who's doing? Yeah.
Not many people ask that question. I think there's an implicit assumption that despite all the faces on our teams page and whatnot, everyone else's job is to sit there and clap as I do all the work. That is not true. We also make extensive use of external contractors. And I want to be clear when I say that that is not we just find people and then miscategorize the shit out of them because we don't like laws. No, I'm talking companies that do things. For example, HumblePod does all of our podcast production work start to finish. They do a lot of the production of producing work on a lot of the other things.
The video work that I get involved in from time to time. And as a result, I could finish a recording and then not think about it again because they take everything there and move on with it. So for me, at least, the goal has been on media to anything that doesn't need to be me personally doing a thing probably shouldn't have me doing it. I should not be updating copy on the website. I need to write the newsletter, but I shouldn't be the one putting in the email system and scheduling the send for 730 Pacific on Monday. There are a bunch of different things that I've learned to let go of as a result.
Yeah, so actually, I want to dig a bit into how you produce your podcast. So when you first started, did you do everything yourself? Or did you hire someone right away? Never.
I started off hiring someone right away because I knew going in that I have no background in audio editing and if I was going to learn this for a period of time, I was not going to be as good as someone who actually loved to do this. Same with graphic design. I never designed our own logo either for any of the previous iterations of logos I had for the stuff I was doing. It's stuff that you can outsource relatively inexpensively and then get back to doing the thing that makes you special. Your audio engineering is never going to be it as someone who's producing the content.
So in the very beginning, once you decided, like, we're going to start this thing, right? How did you get the first customers and what were you doing? Did the podcast already start? How are you paying for all these things?
No, it started off with I had a bit of seed money from Severance for being let go. Great, I don't have to work for four months. What do I want to do instead? And the first 18 months in business, your customers all come from your network word of mouth. People screw that up at their peril where they have this idea that I'm going to, I have this amazing thing. I'm just going to go post on Hacker News and people come flooding into my website and convert, or I'll buy a bunch of
Google ads and then people will come torrenting in. That's incremental stuff. You need to make sure that this is something that resonates with people who already are predisposed to like you. You'll do business with friends first and foremost. And in time, we still find one of our primary generators of business is both recurring customers, some of whom have transferred to other companies, or word of mouth. People who have said, yeah, I was wondering how to fix this and everyone
tells me to talk to you. It works out well. So it's hard for me to draw a direct line between the newsletter that's out there and people coming in who care about AWS billing, but people have heard of me as a result. There aren't a lot of large personalities in the cloud space, let alone the cloud billing space. And people don't talk publicly about problems they have with AWS bills, unless they want to get fired if they're in a significant portion of billing. We've negotiated
over $4 billion of AWS contracts so far and gone growing all the time. That's something that people with that scale of problems aren't public about it. So I need to be loud enough that I popped to the top of the mental SEO stack for that very specific, very expensive problem.
What's the average percentage if you track that of the money that you spent or did you have to save for your customers?
It depends because it also assumes a few things that aren't necessarily true. In many cases, companies aren't looking to actually save money by these optimization projects. They're looking to understand what is going on in their environment. They're looking for ways to think about it. And there's a reason I talk about optimization, not cost cutting. There have been times I've advised companies to spend more. Okay, you say that that S3 bucket is the one true source of all the company's data, but it's not being backed up anywhere. It doesn't have multi-factor delete turned on. And you say if it's gone, then it's going to be a good thing.
Then there isn't a company anymore. Spend more. And this is also why I only ever do it on a fixed fee basis. Otherwise, it's backups. What are you, some kind of coward? Delete those. And I add that up to the tally of what I'm saving them. I want to give the advice that I would want someone to give me or what I would do in their situation. And that's really all it is.
That's really all it is.
Sometimes yes, usually with an eye toward cost. I've gotten the hell out of the security space just because it seems that tends to derail things, though I still periodically discover problems in the security realm as a part of the bill analysis. It's a weird party trick, but I can take the PDF exploded bill of someone's month on AWS and then describe their architecture to them. It's a sad party and it's winding down at that point, but still it can be done. Architecture and cost are the same thing. We also have no partners in the space, which is intentional and by design, which also means if I recommend a product or a service to a company, they listen.
Because they understand that, oh, this is something that I believe in and not just something I'm recommending because I get money if I do it.
So what are some of these huge architectural anti-patterns, if you will, that you very often see people come in and like, okay, that's the one thing, go change that and you'll save like X percent immediately.
There are remarkably few that I can say that about definitively. There's a reason that we classify these things as curiosities. It's a lot of tooling and approaches tend to assume, oh, you have duplicate cloud trail management trails in place. That's a bug. Fix it. Yeah, 95% of the time you're right, it is. But there's also that edge case where security as an organization wants to have its own unadulterated trail that has not been touched by anything else, and then they won't share that onward.
So, okay, that is a position to take. Here is the actual cost of that decision. Now you have context to figure out how you want to make that decision going forward. What we do is purely advisory. We make no changes to customer accounts for a variety of excellent reasons. But it's almost always a mistake when you see it. Data transfer is often rife with misunderstanding. The generalized advice is, oh, buy, AWS comes up with is buy some savings plans or reserved instances.
Go ahead and right size your instances. Usually in that order, too, is the way they recommend it, which is not how you want to do it. And third, we are completely out of ideas. Oh, wait, switch to Graviton and or serverless. And the reason they love giving that advice is because those three are really the only three things that can apply almost universally. Everything else is contextual and it's nuanced.
And it's not something that is going to lend itself to just blindly run this script and it'll be fine. Like our internal tool suite, we consider to be sort of power tools in that these are the tools a professional would use with context. But that doesn't mean you necessarily want to give it to someone who has no idea what they're looking at, because they're going to take whatever the tool says is gospel and get themselves in trouble.
Ilia, do you want to move on to like the media side of it? We had a lot of questions about that. Yeah.
I'm really interested in tracing how your media and content evolved. So I guess let's take a step back. So today you have Twitter slash X. I don't know which one you use. Anyway, so you're on Elon Musk's ground.
That's just plain fun. That's not actually revenue driving. I've never monetized Twitter.
Right. So, but still it's like, it helps a lot. That's probably how many people discover you. Then you have two podcasts, Last Week on AWS and also Screaming in the Cloud. We have a newsletter, maybe two.
One newsletter comes out three times a week. Monday is the roundup of AWS news. Thursday is the same thing with the security focus. Different releases. And Wednesday is usually long-form thoughts. It's also a blog post. The AWS Morning Brief podcast is just the spoken word version of those because not everyone wants to read. Some people want to listen. I'm not like that, but I'm there for other people in the way they want to consume content, not the way I would want to consume content.
And the Screaming in the Cloud show is different. That's something else, and we'll get to that in a bit. But this all started off as just the newsletter by itself because I was already having to pay attention to everything AWS put out. Notice that there was a tremendous amount of nonsense when it came to the sheer number of things that they cared about versus what customers needed to think about.
And I wanted someone to distill it down to the interesting things. And they didn't really have anything out there like that. So I figured I was already having to look through this stuff. They got me 80 percent of the way there to catch it all. And then I figured I'd just write some stuff and send it out. I would give it a few weeks, and at some point if I didn't have at least 50 people reading it who I didn't know, then I would go ahead and shut it down because that would just be sad.
Then Charity Majors tweeted about it when I pre-announced it by two weeks, and the first issue went to 550 people, and it's been growing ever since. Oh wow. So what year was that?
That was 2017.
Oh, Charity's been great. She, in the space that I operate in, yes very much so. We knew who she was long ago and that historically was the space that I wound up focusing myself on because it's the field I come from. Those are my people for lack of a better term.
Yeah, for those listeners who don't know who she is, I think we should link one of her info Q talks about testing code in production. I think that's what it's called. That's an amazing talk about observability.
So today, like, do you do all of the... you write the newsletter, the weekly one you said, right? The in-depth one? What else? What else are you doing for the media side? Or is it all you?
All the content is me. I don't have anyone else writing any of this. I do have an editor who helps once I put out the rough draft, who then winds up helping with structure and then I go back and incorporate those changes. But that's writing. Everyone else's operation side or sales side, for example. That's what sort of happened next is a company, Datadog, as it turned out, reached out after I've been doing this for about six weeks and said, can we sponsor your newsletter? It's like, can you give me money to talk about you? Of course you can give me money. How much money? It sort of grew from there.
And it was surprising given to me the relatively small scale of the audience versus the revenue it commanded. But it makes sense when you realize that this is a specialized audience that has tremendous influence over purchasing decisions. And if one prospect becomes a customer, it sponsors everything I do for a decade and change just because the dollar figures are so ridiculous. And it was a surprising source of revenue. I treated it like its own P&L. So I was able to then take that and basically self-finance investment in the media side of the business.
My total upfront investment for the newsletter, I decided I'd give it twenty five hundred bucks because that's what it cost back then to join the AWS partner network. And is that the best way to grow my business? I don't know. I had that money. What would I do with it instead? Well, I do have that money. Let's see. I have a theory. And here we are.
The podcast came a few months later, Screaming in the Cloud specifically. I just recorded the 500th episode of that. So it's been going reasonably well. But it started off as basically an excuse to talk to people I had no business speaking with.
And because if you historically, at least to my mind, wanted to talk to someone, their response would be, how do you get into my office? But if you invite them on your podcast, they'll clear an hour off their schedule to make that happen. And at least that was my opinion at that point. It gave me an excuse to reach out rather than just I find what you're doing interesting. Can we talk about it?
And those authentic conversations with people about things they're excited about, whatever that happens to be, every episode is different. I enjoy the ebb and flow of them all. And you get to see an awful lot of the industry in a very short period of time just by talking to people who are innovating in various corners of it.
Yeah, you know, I listen to multiple of your episodes. I think they tend to be pretty short, kind of intense and focused on one or two topics max.
Roughly half an hour, uh, I expect it to an hour and do a half hour recording on it because it's, it depends on the topics and where we're going with it. But a lot of people can be interested in engaging for, on almost anything for that period of time. But if you do an hour, two hour, three hour long podcast, well that's a big time investment you're asking for people to listen to it.
And two, people start running out of steam and the things they're really passionate about right around that point as well. There are times I've done episodes that crossed into an hour and a half or so, but it's atypical.
How do you think about sourcing the guests? Are you the person reaching out? How do you go about?
I am. It turns out that it's not that hard to see someone doing something interesting or they make a great point you hadn't considered to just drop them a line and say, hey, would you like to chat about that on the podcast? And in the end, the answer is yes, great. Here's a link to the booking page. It asks the stuff I need to care about and I don't think about it again. The time rolls around. I wind up, oh, in 10 minutes, I have to go do a recording about this person? Cool. And
I suck at preparing for things. I'm ADHD personified. And when that's your failure mode, you've got to be good at improv. So great. We spend about 10 to 15 minutes beforehand talking about what it is that we're going to cover. Anything I should avoid talking about if you're sensitive about it, great. Let's dive in. And for the rest of that, we have to do about 40 to 45 minutes most times on the recording side of it. And then we wind up cutting that down as a result.
So before you start recording, you talk to them right away before you hit the record button. That's how it works. Yes.
Because otherwise we have a pre-meeting a week in advance. Great, I'm not going to remember what the hell we talked about in that level of context and depth. And once you get people there, it's good to at least figure out the direction. Even if they say there's nothing in particular I need to talk about, great, I'll take it from there.
and then you throw it over the wall for the editing and all that. You said you outsource all that.
I wind up closing the browser tab and don't think about it again because we've automated an awful lot of this.
Even the publishing and scheduling and all of that.
Well, some of its people, some of it is automated systems behind the scenes. The line is sort of blurry and these days I'm not sure where it starts and stops because I'm not building those systems anymore. I did originally. Because one of my failure modes before that was I would do a recording and then it would just sit there for months before I remember to copy it over to the editing folks and get to work on that. What tools do you use to record?
A microphone and a keyboard, generally speaking. There's not a lot of magic that goes into it. I'll use remotely.fm as the browser platform that I've been using now. I was using Zencastr for a bit but they got kind of weird and I had some audio artifacting issues. And there's nothing perfect but they all tend to use largely the same APIs. Most of them are as good as any other. I'm not a big fan of doing it over Zoom. The compression gets weird. But it works out.
Yeah, just when you said that, you know, you can forget to copy things. I just wanted to insert a plug for the squad cast that we're using right now for recording. We're not sponsored by them, but we had their founders on the show. It was episode eight, really nice guys, bootstrapped company. So there, you can actually add people to your organization. And then once you record, they can just go and download the files themselves. So which makes things easier if you outsource, you don't have to pay for extra users.
They don't have much of an RBAC story going on over at Remotely. We evaluated Squadcast back and we periodically review all of our vendors and as our needs change because we're still using this thing because it solves a problem that we had four years ago we haven't thought about since as a bad reason to keep doing something. And we wound up passing on Squadcast but I don't for the life of me recall exactly why. Some of it has to do with the reason we pick these things is mostly not me. I'm not an audio file. I do not have the ear for that sort of thing.
Audio quality is terrible. Can't you tell? No I can't. I also believe people over index on tools where they want to make sure that their production value is stratospheric. That's secondary and takes a backseat because it has to to content. People will suffer through an awful lot of crappy production value if the content is compelling. And if it's not it doesn't matter if your production value is Oscar winning. No one is going to care.
And I think you said that right, there is a minimum bar. It's easy to like get over that minimum bar for the quality, the audio quality of it. Beyond that, yeah, you could make it like a movie, but most people wouldn't care or know the difference between those two.
Depends on what I'm doing too. If I don't have a chat with someone at a sidebar in a re-invent for example, I want to have something portable on my person that I can flip up and say go and we start recording. And on some level, if you have certain production quality requirements, well now you have to convince them to come back to a studio and schedule time for that and a lot of opportunities get lost as a direct result.
So tell us like how all of this the newsletter and all that you said you have a separate like P&L line for that How does it feed into the business? How do you tell that attribution?
Effectively, it's one of those rarities in the world, a marketing department that turns a profit.
Right, but how do you attribute like, is there a customer acquisition or something like that sort of thing that you can attribute to a podcast episode or a newsletter?
No, not directly. It's almost impossible to pull that off because people don't reach out to us because they hear an ad for us somewhere. It's something that has to sit with them for a while. There needs to be a recurring, acknowledged, painful problem. It's also a very fixed point in time. Today you might not care. Tomorrow you will. And in three days you won't care again. People put in the word you hear about us.
The overwhelming answer is the nonsense that I do. Very often, like, I don't know, I've just always sort of known that Corey was there. That's kind of the point. I also don't trust podcast download statistics for crap, but I do know that for no other reason than I've had some guests who absolutely have huge platforms of their own. And the numbers don't really move even when they promote it themselves versus what I see as a standard baseline. So, OK, whatever on that front.
A lot of these sponsors definitely want to do attribution. And that podcasting in particular, and also to some extent newsletters, are not the way to get there. The only way to get there reliably is to have an in-depth conversation with the person who reached out. And that's hard to do. There are some things that you can never do attributed value to. What is the ROI on having your company's advertisement for 20 minutes on the Times Square billboard in New York?
And the answer is, I don't know, but it's cool. So you should do it anyway. People love to be able to measure these things about what's working and what's not. But an old saw on marketing is half of your marketing budget is wasted. You'll go broke trying to figure out which half.
Right. And I think you made a pretty important point on one of your episodes that your advertisers, so big companies who provide enterprise services, so the customer lifetime value there is huge, and their clients are mostly companies. So the person who heard your podcast, their ad on your podcast and made the decision might not actually be the person who is buying.
Right, it comes into meetings, it gets discussed internally and then someone decides to reach out for a trial. I had one sponsor that decided after a couple months that they weren't going to continue because they weren't seeing value. Okay, cool. Two months later they came back and bought everything I'd sell them and it turned out that they sat down and interviewed their largest customers.
I think two of the top five or three of the top five where I've heard about them on the podcast. Like, oh it works. I'm honestly surprised they were able to figure out that those companies came through the podcast. Most people just don't know where it came. Someone in a meeting brought you up.
So by the time you launched your podcast, you already had the platform of the newsletter. So your podcast essentially didn't have to be bootstrapped from zero in terms of the audience.
Sort of. I mean, just because you're going to email a newsletter doesn't mean you're automatically going to download a podcast. A lot of that doesn't convey. But it gave me a platform from which to talk about the launch of it.
So on the actual like the billing side of your business, right? Is that like a recurring model? Or do I come to you when I have a problem and I know that there's Corey?
At a fixed point in time depending on what it is you're looking to do which is a little bit challenging. It feels like it's a bit of a treadmill story where you have to run to keep up all the time. But there's something reassuring about recurring models but the way that we approach it at the moment we have not yet found a way to serve our customers well that doesn't turn into just rent extraction and that doesn't sit well with me.
We actually had a episode a couple of episodes ago on subscription pandemic, that's what we call it. So it's a global phenomena that everybody wants to charge you a subscription on personal level, you know, enterprise, I guess, retainers for companies also come to that space. So I'm curious, what do you think about usage based pricing versus subscription pricing for cloud services?
People like to do consumption or usage based pricing because they believe that it is transparent and that it's fair and that it's predictable. And in truth, it is none of those three things. It is not at all clear. If I'm going to charge you for every second that something runs that works out to this monthly fee and what you need is going to scale up and down, you run your application for a month and then figure out what it costs. It's an after effect thing.
The fact that it's transparent doesn't mean it's easy to predict and predict is the hard part on this. And as far as it being fair, I don't know what's fair. I don't think that necessarily means much in a business context. I do know that most of my customers would pay 10 to 20 percent more than they're paying now for AWS in a heartbeat if there was some magic way to be able to predict to the penny what it was going to cost them for the next three years. So basically to flatten the spikes.
Make it predictable. Spikes are fine, even seasonal, but they want to be able to predict and then reason around it and as a result then inform the rest of their business decision making process.
So for your own business, because it's not like a recurring revenue kind of story, right? In the beginning, when did you decide that, okay, we're ready to like hire one more person we need for this role? And because it's not a recurring revenue, it's a bit harder to kind of reason about that.
It became a bit easier at the start because at the start we hired someone to do sales for media. And it made sense because I didn't want to spend all my time wooing sponsors and that's wound up justifying itself super easily. The next problem was, okay, now that that's done, what do we figure out with? What do we figure out as far as having other people come in and service these things? And if so, what does that mean for
our ability to focus on the business as well? There's no easy answers in any of this. It comes down to what makes sense, what doesn't make sense. Hiring is always harder than people give it credit for being. Managing people is incredibly challenging. I no longer manage people for that exact reason. I let my business partner handle that. In fact, the only person I can actually fire at this company is my business partner who's also the CEO, so it's kind of fun.
And so are there like other technical people in the company of the seven that you said? Oh yes! And what are they doing?
Consulting delivery has always been fun. I'd also say everyone on some level is technical. Just, it's hard to wind up viewing technical through a narrow lens of, oh, they write code. They are very technically proficient at the areas that they focus on. One of the challenges as well is when you hire for someone to do the kind of consulting we do as we do it, it's basically unicorn hunting. You need someone with the background of a senior SRE because there's no way to teach scale for
someone who doesn't have that. You need them have the consulting skills, which look like a lot of varied things. They need to be comfortable presenting to clients at basically every level of the org up to and including the board. And then when you get all that done, a lot of the work starts to look repetitive from certain points of view. Then these people are often bored and then they decide what do they want to be doing instead. So there are challenges in this space. We don't
believe that we've managed to solve them all yet. We're getting closer.
So as a founder and the business owner, you're probably more tolerant to doing repetitive things because it brings you money.
Oh, personally, I can't stand when too many days look alike. But there's a reason that the type of engagements that I work on look very different than the type that other folks work on here. The long-term deep dive into every aspect doing cost analysis doesn't interest me. My attention span trips out. But let's sit down for two days and tear through one side of your bill and then down the other and find the optimization opportunities there in a quick first pass. That's always fun.
Right. I want to get back to that point you made about your first 18 months of customers being from your network and all that. I want to ask a selfish question, because we are a bootstrap business. So we will launch our product, maybe two, three months, I would say in the very best case scenario, because it takes time to build it publicly launch. Yes. Yeah, publicly launch. We may run out of money in terms of like our own savings, right? And we don't necessarily want to take VC funding unless we really have to. So we were thinking about doing consulting, maybe some product consulting, maybe consulting people like on the process.
Like product development because of our experience. Actually, what would you recommend to folks like us? Like how do you find your first customer in your network?
People do consulting in a lot of ways and I think a lot of them are misguided at best. Never charged by the hour. You're going to spend at least 70% of your time on client development and finding new clients. So unless you want to work 120 hours a week, I don't recommend that. There's also the idea of being very narrow and very focused because it's much easier when you can speak directly to someone who hears himself in the description of what you do and address the expensive problem with which they wrestle.
What I started off with was great. I am a great solutions architect. Cool. Swing a dead cat. You're going to hit 15 people in a room who look like that. And when you're trying to be all things to all people and going broad, you're now competing against Accenture who has a bigger marketing budget than you do. Instead become more and more narrowly niche because that's marketing. Positioning is marketing. And ideally you should be able to trigger Rolodex moments where people at a party ask what you do. You tell them like, oh my God, I know someone you need to talk to.
That happened a lot. Specificity and being the leading expert in a very narrow space is the way to be able to charge top dollar for a lot of these things.
So basically you were in the niche of your own for a while. You established the niche when you first started. Oh yes. So I listened to an episode of your podcast, which was probably six years ago or so. I probably don't remember the specifics, but I'll take it at face value. Yeah, I can't find it, unfortunately. It was with some VCs who invest in enterprise software or cloud software.
And you made the point that GCP has the best tech of all of the cloud providers, but they have kind of the worst enterprise sales, the worst customer service. And that triggers a lot of people away from that. So I spent time in the US, I spent time in GCP. I kind of have seen some of that internal stuff going on. What you said there really resonated with me in terms of how these companies operate.
I'm curious, have you seen much change, I guess in Google specifically, in GCP specifically, over the years?
Google's gotten much more proficient at speaking enterprise. I think that is Thomas Kurian's influence and it's done tremendous things. You no longer have the same trope of the salesperson from GCP who shows up at your office and insults you. That happened in several occasions I can remember personally. It's a... they don't do that anymore. They realize that the decision makers are not always engineers but a lot of the people driving the decision are. That's a natural evolution that works.
I also think that Google continues to be technically excellent. I think where they completely miss the ball is understanding just how much damage product cancellations do to them in the conversations to which they are not a party. At this point after they killed Google Cloud domains, how do I recommend a Google product and not look like a clown in the process if it's something that a cancellation is going to cause a problem with?
There's no database anymore, right?
No, Janice. Sorry.
Yeah, I mean we we ourselves we use Firebase right now. We have Google domain. How's those dynamic links weren't treating you? See by the time we got to Firebase it was already deprecated. So we didn't use that
Yeah. Oh good. So you're safe for this one. It's just the next cycle that'll catch you.
Yes, exactly. Yeah, but it is a hard thing. Yeah, with Google Domains too, I think all the emails and all that Ilia and I have set up, we don't know what's going to happen. And there's no clarity. We don't know exactly when what is going to get deprecated. Is there an alternative to it or not?
What's your take on this? Because this was a big decision, I suppose. The repercussions of that is pretty big because there are I think 10 million domains registered to Google domains. It's both people like small individual people just doing domain squatting to large enterprises who registered domains to Google domains. And especially some features go away. Let's say email forwarding goes away and stuff like that. People will be really pissed at Google. I'm
curious because I cannot explain it. I've not been there long enough to really understand how they make this decision. What's your take on this?
Every single Googler that I've spoken to about this, and many of these people are remarkably senior, has expressed that they found out about it the same time the rest of us did. Wow. Corporate communication chains are hard.
That's crazy, yeah. I think coming back to the other side, Amazon as a company and AWS, one part of it, big part of it, one of the best things about it is the customer service. Google, we haven't had the opportunity or the misfortune to like talk to their customer support yet. What's your take on like, is that a typical because Google, the company doesn't have a good name for customer service? Is GCP similar or?
I have to defer to what other people have to say about this. One of the side effects of being known in the space is that when my name comes in attached to a question I have it tends to get actioned very effectively. It is almost unheard of these days for me to have a poor customer service experience with a cloud provider and I'm not asking for special treatment but it also is the natural evolution of these things. I'm also at a point technically where when I start having
problems it is invariably not something that I've gotten wrong on my side. I've solved more problems in the course of my career than I care to name by starting off trying to draft an email to a newsletter to a mailing list on this or to or write a ticket where I distill the problem down into a skeleton reproduction case and then I find the problem. Rubber duck debugging is great. So at least my issues are complicated.
Can you attribute any product changes directly to your feedback?
Oh absolutely, I've been talking in terms of AWS, Google or both.
AWS like we know I think ourselves. So we'll just talk about Google, yeah.
My personal favorite, and I'm not going to name the service, but the GM was super excited to meet me and brief me on the service a week before re-invent. And they asked if I had heard of it yet, and I said, well, I did write the original PR FAQ. And their response was, no, you didn't. Ding! It's in your inbox. Let's see how far it's fallen from my original vision four years ago. And they just had no idea. And again, they weren't true to life on all this. I'm an outsider. What does it matter?
But yeah, there's a bunch of stuff that I've seen that I've caused to get fixed, that I've been noisy about, that I have asked for on Twitter and then get granted later. And there are also very good reasons why some of the things I ask for don't get done. No one shows up at work hoping to do a crappy job today unless apparently they work on Microsoft security.
So it's one of those areas where it's just, there are constraints. The scale of what they do is monstrous, and it has to work for every customer the service has. And that means that corner cases become common cases at that scale.
So looking at these three organizations, AWS, Azure, and GCP internally, where would you say they're doing a great job? Each one.
Microsoft has the ecosystem stuff working for it, but I think that they're having serious problems right now with their lack of attention to security on that to the point where I think that we're going to start seeing a lot of brand name customer references going to ask to not be references anymore as a result. Because if you see your bank as an Azure reference customer, what does that really say about you? And that is reaching a fever pitch. Google is doing surprisingly well in a lot of these things where where I don't necessarily see that there is a whole lot of people that are going to be using the platform.
There's a whole lot holding the back except Google themselves. And AWS has focused on some level toward revenue protection and honestly just being crappy to customers in ways that I would not have expected years ago. I don't know what the pressures are that are driving it, but it's unfortunate. And I'm no longer given the same benefit of the doubt or assuming the same good intent that I would have three years ago.
I wonder if with AWS, because I left AWS when Andy Jassy was still the CEO of AWS, and he was the founder of AWS, even though he was an employee of Amazon, but he was the founding figure, right? And Bezos was still the CEO of the company. And I think what in 2021, basically, AWS CEO has changed. So it was no longer led by the founder by hired executive. I wonder that I mean, I can only speculate, of course, if that might have played a role, because a founder figure might think a bit different about certain issues.
It's entirely...
Impossible. I don't know at some level what that is and I think that a lot of these factors are incredibly complex and trying to distill it down into simple talking points from the outside is a fool's errand. It's, oh what they're doing is easy all they have to do is just no no no no no none of this is easy but I do know that historically customer obsession was their guiding star and it feels like that has been lost as a
customer. Now that is what I am feeling and that was the thing that made AWS special and if you take that away there's not that much difference between them and Oracle if we're being perfectly direct. And I know a lot of Oracle customers, but I don't know anyone who likes Oracle.
What has taken that space, the customer obsession piece of it that's missing now?
I don't know that anyone has yet. I mean at least small companies it's easy to do because you know you're a small scrappy startup and you want to obviously be great and get a lot of attention to your early adopters and the rest. But somewhere between that and a trillion dollar publicly traded company a lot happens.
Right, right. No, I mean like in AWS you said the North Star, the focus used to be customer obsession. Now it's no longer that. So what is it that's taking that space inside AWS?
Well lately it seems like it's generative AI because the market's excited about it so we need to spill that all over everything we've got. And we're an AI company. We're totally not behind. It's like, me think the lady doth protest too much because you're awfully insecure sounding about that. And you haven't shipped anything really significant in that space. And you're rebranding a bunch of stuff that's been around for a while as generative AI.
And I don't want the underlying platform behind all of my infrastructure to be hype chasing. Yeah, makes a lot of sense.
So with that, we actually have a question. What's your opinion on producing content with ChatGPT, which a lot of people seem to be experts in these days, at least if you trust your LinkedIn feed?
Oh, I've been doing it for a while now. Part of what I'll do is I'll tell it to write a blog post about a particular topic with talking points I provided in the style of Corey Quinn. And then it'll spit something out. And I usually despise everything that it wrote. So I copy that into a text editor and then I spend 20 minutes mansplain correcting the robot on these things. And at the end of it I've got a serviceable first draft and it's easier than staring at the blank page.
And it comes with a structure. It comes with misunderstandings around points in some cases. It wouldn't have occurred to me that someone would believe that to be true. We'd better make sure they don't. And at the end of it I maybe have like three words in a row that were originally in the thing and I can live with that for one. And two, I'm perfectly fine with plagiarizing from the thing that's plagiarizing from me. It feels certain poetic justice to it.
By the way, you wouldn't publish whatever it output as is.
My God, no. Without review and looking at these things, hell no. I do have a colleague who has what I affectionately refer to as the asshole in email problem, who will take a message that normally they would have sent and put it through Chat Jipity and say, great, turn this into a business email. And that works super well. And I'm sure some of the other side is like, this is a five paragraph email. Hey, Chat Jipity, tell me what this person wants me to do. Yeah, it's like basically it's an API for human interaction.
And that's what the email has to convey that a Slack message doesn't.
So I accidentally did this yesterday where I asked ChatGPD to give me how to make a Mai Tai. There were some friends over, but I had my system prompts and everything set up to always say, you are a software developer, always responding like technical and precise instructions. And what came out was hilarious. I mean, it was awesome, but hilarious.
I had that too when I was trying to get it to explain, I forget the exact concept, to a six-year-old because she was curious, but being snide and sarcastic as a default tone does not work well with six-year-olds in case you weren't aware of that. So like, oh, let me fix that right down. On Twitter, I've done a bit of content with chat jippity just in terms of here's its response to the prompts I've given it. Cyberbullying it into ranking the US presidents by absorbency was a good one.
It's like, that's not a respectful way or effective way to rank US presidents. It is if I have a spill and need to select a US president to mop it up with, like, oh, here you go. All kinds of fun stuff you can ask it. Like, for example, companies are doing a terrible job of marketing this stuff in fun ways. Here's a transcript of a meeting. Pick out the action items for everyone.
No. Here's a transcript of a meeting. Find the most unpleasant son of a gun to work with because we're trying to build a layoff list.
Cool, so on this funny note, you just have a couple of parting questions. So Corey, do you listen to podcasts? Or do you read books? Can you recommend something that you... content that you can see on your mind?
This is a weird confession. I don't listen to podcasts and I don't watch videos. The way that I ingest, that I learn things are twofold. One, I can read astonishingly quickly. So I absorb and retain information through reading very well. But if I want to learn something technical, the best way I found is to build something with it. That is what really sears it into me.
So it's weird that I produce podcasts but don't listen to them. But it's absolutely true. There's a reason that all of mine have transcripts. So that you can just, you get the written version if that's what you're into. And you can absorb things a lot more quickly there. There are a wide variety of different things that you can read that point to different ways. There's trash reading you do for fun. There's technical stuff that comes out.
And then there's the occasional book that changes the way you think about things. Never Eat Alone was one of them by, I think it was Ken or Keith Ferrazzi. It talks about the value of how to build connections and network effectively. I mean some of it is obvious like just go to lunch and people will do you favors. No, no, no, no, no. Do favors for people.
It comes back around again in weird ways. And the way that you do this is you always make it a point to go out and be social with people. Like some of the best money I ever invested in my entire life was at various jobs that I had. When I would take anyone who wanted to at any time out for coffee and buy coffee and just have a talk of what they were working on. I must have spent thousands of dollars on coffee a year and it was worth every penny.
Because sometimes it turns into weird stuff, weird ways of seeing the world, weird opportunities. They talk about the problems that they have and that gives you an idea. Or hey I know someone who's focusing on that. Have you two met?
It's a small thing for you but in some cases it can change everything for them. And that's indicated every time that I wind up getting a rando DMing me, which still happens from time to time, about something I said years ago about job interviewing or having ADHD or any of the other stuff that I rant about sometimes really changed their perspective.
And then they credit that with changing how they approach something. I don't think I did much. All I did was just made an observation. They did the rest of the work. But it's gratifying to know that hearing the right thing at the right time can affect people's lives. Been years since I read it. I should definitely read it again.
Cool, yeah, we will link the show notes.
Now before we close, I had a side question from what you said. You mentioned that basically the mode that you learn things is ingest things is to try them out, right? How do you keep up with the insane volume of stuff that's coming out from AWS, Azure, GCP, like all the time?
Because AWS putting out something does not mean that it is for you. Okay, that's interesting. Maybe I'll use it for something, maybe I won't. I don't try the majority of things that come out, but there are things that I go very deep on. But if I have a spare afternoon to kill and I build Slack into my schedule for this sort of thing, all right, let's try spinning up something with Amplify and let's see what I'm going to learn.
What did I learn? I learned I'm not a front-end developer. And that was fine. But even just kicking the tires on those gives you an exposure to it. But for a lot of the in-depth stuff, talking to people, this is the value of talking to people, I might think that Aurora Serverless is crappy for a use case or great for a use case, but I'd rather talk to people who've been building production applications on top of it.
What do you think? Because if you love it or hate it, that's going to tell me something. And there's also a circle of people I trust on a lot of those things who are working in certain areas. I don't necessarily trust randos. Like, that means we'll trust Amazon reviews. I don't care. But people that I know in certain spaces who have opinions that are considered and worth hearing, I listen twice to those people.
Yeah, so Corey where can people find you?
Last Week in AWS.com What is pig a reference to? Oh, because when I changed my name to Corey Quinn in 2010, all of the usual domain stuff were taken. And Quinn is a great last name to pun with. My wife and I, for example, don't engage in hanky-panky. We have quintimacy. So, you know, keep rolling with it. And it was an experiment. So instead of being a guinea pig, I'll be a quinny pig. And it just sort of stuck.
That's very cool. All right, Corey, thanks. It was a pleasure to have you on the podcast. And I think I need to close this with, if you liked this episode, give us a five star review. If you hated this episode, give us a five star review. That's how you end your podcast every time, right?
I do, and then also leave an insulting comment and then I throw it back to something that was said during the episode. Again, this is like how I keep it fun and engaging, and I generally don't know what I'm going to say until about halfway through the sentence introducing it. It's like, great, like, you better have something. Goes well. It's games I play with myself to keep myself engaged.
Alright, yeah, thank you so much. It was awesome, yeah.
Thank you for having me, I appreciate it.