Screaming in the Cloud - Engineering a Tech-Driven Newsroom with Jeremy Bowers

Episode Date: February 26, 2020

About Jeremy BowersJeremy Bowers is an Engineering Director for the Newsroom Engineering team at The Washington Post. Previously, Jeremy was the Senior Editor for News Applications on the Interactive News Team of The New York Times, where he led a team focused on writing software for elections, Congress and the Supreme Court. Jeremy was also a news applications developer on the NPR Visuals team and a Senior Newsroom Developer at The Washington Post.Links Referenced: Twitter: @jeremybowersLinkedIn: https://www.linkedin.com/in/jeremyjbowers/Personal site: jeremybowers.comCompany site: https://www.washingtonpost.com/

Hello and welcome to Screaming in the Cloud with your host, cloud economist Corey Quinn. This weekly show features conversations with people doing interesting work in the world of cloud, thoughtful commentary on the state of the technical world, and ridiculous titles for which Corey refuses to apologize. This is Screaming in the Cloud. my podcast, so here we are. Their platform is designed to help SaaS companies understand how and why their costs are changing, and lets you easily measure by things SaaS companies care about, like cost per product or feature. Also, they won't make you set up rules or budgets in advance, but will still alert you before your intern spends $10,000 on SageMaker on a Friday afternoon.

Go to cloudzero.com to kick off a free trial. That's cloudzero.com. And my thanks to them for sponsoring this podcast. This episode is brought to you by DigitalOcean, the cloud provider that makes it easy for startups to deploy and scale modern web applications with, and this is important to me, no billing surprises. With simple, predictable pricing that's flat across 12 global data center regions and a UX developers around the world love, you can control your cloud infrastructure costs and have more time for your team to focus on growing your business. See what businesses are building on DigitalOcean and get started for free at do.co slash screaming. That's do.co slash screaming. And my thanks to DigitalOcean for

their continuing support of this ridiculous podcast. Welcome to Screaming in the Cloud. I'm Corey Quinn. I'm joined this week by Jeremy Bowers, who's an engineering director for the newsroom engineering team at The Washington Post. Jeremy, welcome to the show. Thanks, Corey. It's good to be here. It's great to have you on the show, but it does feel a little surreal.

Oh, you work at an actual journalistic outlet. Cool. You want to come on my nonsense podcast and talk about computers for a while? And it first took a fair bit of courage to ask that question. The response was, absolutely, which, oh, great, you folks are human beings, too. You put your pants on two legs at a time, just like the rest of us. You know, you keep making these assumptions about pants, and I'm not 100% sure that you've

ever put them on. Oh, yes, right. Of course we put them on two legs at a time. Oh, yes, true thought leaders jump into their pants. Exactly. Yeah. So you're an engineering director, which most people can wrap their heads around what that means if they're listening to a show like this. But for the newsroom engineering team, what is that?

Well, any newspaper of sufficient size or even media company will run into these problems where they have reporters who are working on beats that are really heavy in data, or they have folks who need tools built for them to help them get to stories that they can't get to. And many organizations kind of deal with this on an ad hoc basis. When I used to work at the St. Petersburg Times in Florida, now the Tampa Bay Times, don't ask. You know, we would handle this with like a small group of like one or two people who are like self-taught programmers. But the Post, being a large news organization, you know, has an actual legitimate, you know, engineering team. And this is the kind of thing we deal with. So when we have reporters who are trying to make sense of a voter file, which is like a gigantic spreadsheet that's 384 million rows long and about 600 columns wide, or if they're looking at federal election commission data that is campaign finance disclosures, it's just not something that they can look at with their eyes and make sense of it.

They need software written for them. And so that's what the team does. That sounds like it's almost a quarter the size of a typical month's AWS bill. We won't even talk about the size of our AWS bill. One can only imagine. It's interesting because it's sort of a business-level problem that business analysts and BI folks deal with all the time. But given what you do and how you present it to the world, you take it a step beyond. It's not just about sorting through vast quantities of data and

turning that into pretty reports in PowerPoint, but you then have to take it a step further and present that as something that someone who's not in a business environment or has the context to absorb with potentially an MBA equivalent level of understanding. But people in middle schools, back when, once upon a time when I'm dating myself now, we would have to take out clippings from various newspapers for current events day and go in and give a small presentation on it. And I don't know if cow kids do that today, but there was always a, oh, have something data-oriented was periodically something they tried to wind up having us talk about. But being able to deliver complex, distilled outcomes from data to a middle school reading level has got to be a phenomenal challenge that most business folks don't, I guess, have the luxury of not having to worry about.

You know, this is one of the things that we struggle with a lot. Early attempts at data visualization from like the 1800s, you know, we're using relatively complex stuff, you know. Oh, like back in the 1800s? Yeah, legitimate 1800s. Oh, early days of Microsoft Excel. Exactly. Yes, right. So they were writing their VB script, William Playfair was writing Visual Basic to try to generate these complex charts

and stuff that for people who had exceedingly low reading levels for the general population, he was trying to get rather complex things into their heads. I think the thing that we have now is we understand that our readers might like sophisticated if they can put their full brunt of their attention against something. But, you know, we understand that we're like competing for attention with so many other things. And so the majority of what we're trying to do isn't just to distill it down to make it easier to understand, but to make it quicker to understand so that you can instantly get the thing that you need to know about the story. And typically that's what we're using the data to do. Do you have an example of a story or a series of stories that you spent a significant amount of time on?

I mean, I have to imagine you're not sitting there doing the deep dive distillation of data for, and that's, we won the Grammys this morning, or actually, even that's a terrible example, because I'm sure there's data that goes into it. There's excellent data. Yeah, and that's the enormous pile of data that talks about, oh, I don't know, something dumb some political figure said this morning. You know, I'll tell you, the thing that we've been spending a lot of time with, so my team focuses mostly on elections, and it means that we spend a lot of time with our

politics staff, our national politics staff. And a thing that we're learning is that this is a lot of reporters. It's their first time really working with large data sets on their beats. A lot of them come from other beats inside the paper and they're working on the campaign for the first time. And they're suddenly like have to start explaining things like how the shape of the American electorate or what are these trends that we should know about campaign finance donations? And like, these are legitimately difficult questions to answer, even if you're, you know, a real nerd and good at this. And it's particularly difficult if you've just come from, you know, sports or from features, and you're now you're writing stories about, you know, how campaigns are operating. And so the vast majority of the stuff that we end up working on is like trying to find ways to integrate, you know, like this sort

of complex stuff that we get, you know, from these data sets and into reporters like daily workflow. So a good example of this would be we have reporters who want to go write about the state of Texas. They would like to write a story about what's happening in Texas and is it likely that Texas is going to turn blue or purple in the 2020 election. And so in typical years, like our reporters might go down and write a story about Austin, right? Because there's a lot of Democratic voters in Austin. But we do a quick look through the voter file and we can point out that a story about Texas turning blue shouldn't focus on voters in Austin because there's not a lot of new voters in Austin. If you want to talk about Texas changing colors in the next election, you want to talk about the Houston suburbs where there's been a

huge influx of Latinx and Democratic voters that are relatively new to the state and they are not voting Republican. And so these are like trends. It's basically updating the priors that our reporters are holding to help them write better anecdotes when they go to write an anecdote. So if we're going to send a reporter to Texas, we won't send them to Houston or to Austin. We'll send them to Houston. So how does that data get intelligently surfaced to the newsroom? I mean, I can see having to like, oh, just run the following set of SQL queries is all well and good. But having had conversations with a fair few reporters over the course of my career, most of them did not spend most of their time dealing with databases. And if they did, they were really sad all the time.

Right. I mean, this is a, this is a, actually, it's probably the hardest part of the job. You would think that maybe the hardest part of the job would be like, you know, data ingestion and cleanup or, you know, the sensemaking steps that we take to sort of find trends. But if the only thing that we can build for a reporter is like a dashboard that we have to ask them to come look at and like stare at pie charts for a couple hours, they're just not going to include that information in their story. And then we're all kind of – it's a lesser version of the story that surfaces. So like the first thing that we think about is, all right, so when we find something interesting in this data set, how are we going to get this to a reporter so that they can make, so this can be like a first class part of their story? So in the majority of the time, we spend a lot of time trying to figure out how to turn it into words rather than into something visual. Although we do have a pretty good relationship with our graphics desk, but that's a whole other conversation. But for our reporters, we focused on these little newsletters that we can send them. We have a lab that's working on the elections. This professor, Nick Diakopoulos from

Northwestern University is hanging out with us for the fall. And one of the things that he's working on is lead generation. Not like what you would think from business intelligence, but literally helping reporters write leads to their stories. And so he has this little piece of natural language processing that reads in a bunch of data and produces essentially a tip sheet for a state that shows off things like, well, here's places where a lot of new voters have registered, or here's places where that voter registration is weird or interesting in some way. Here's places where the vote history has changed significantly in the last two years. And these are really great ways to like sort of get a reporter a little factor too that they can use when they go out and do their reporting. And so, you know, it's not going to be a story that

you look at and go, hey, that's a story about data. It just is a it's a standard like newspaper story that might feel anecdotal to you. It's just that it's the correct anecdote instead of the wrong one. That sounds like an incredibly challenging problem, but it also feels like it's directly aligned with a lot of the modern day snake oil. Sorry, it is more upscale than that. The modern day serpent grease that is artificial intelligence slash machine learning slash math, if you're not trying to scam VC money out of people. Yes, right. How, I guess, when I think about a newsroom and its technology,

my immediate response goes to, my immediate mental leap is to typewriters and notebooks and people who buy pencils by the gross. And it's a very old-timey type of mental image that's largely informed by comic books and cartoons when I was a kid. I'm going to assume the technology has evolved somewhat if for no other reason that I listen to, for example, a Washington Post podcast in the shower most mornings. I don't wind up opening the window and getting smacked in the face by some paper boy hurling a, sorry, paper child, paper youth as the case may be, hurling a newspaper into my face. So there's obviously been significant technical changes that have hit the journalism profession in the last 30 years.

But it's strange that I think on some level the collective consciousness hasn't really caught up with that. You know, it's funny that you should bring that up about having a street urchin hurl a wadded up piece of newsprint at you in the shower, because it sort of illuminates this problem that we have in journalism, which is that there was a lot of technological thinking and engineering thinking that went into changing how we present information to our readers, but comparatively little that goes into how we change the way our reporters actually report on stories. You know, and so, you know, you're joking about, you know, typewriters. And honestly, a lot of reporting is not that different than it was in

the 1980s or 1990s. You know, the advent of the internet is a thing that we definitely have to have in like that our reporters use, you know, LexisNexis and things like that to do searches on people. But, you know, truthfully, there's a whole lot of change that's happened in, say, in other fields where like reporting has just sort of lagged behind. And so the thing I think that I'm pretty excited about doing is like attempting to take some lessons that we might have learned in other places, you know, minus the serpent grease, of course, and try to bring some of that sensemaking to our reporters. Now, critically, I think that it's basically impossible or a poor task for us to try to replace our reporters with, say, a fleet of robots, you know, who do all the reporting work, produce a story, and then put it out on the internet. That's called a content mill. Exactly. I still think that there is, I still think that reporters have these pattern matching skills that are more or less ineffable. And I think it would be remarkably difficult, if even

impossible, to replicate. But there are certain things that they do on a daily basis that would make your head spin if you saw the wasted skills. Like when I worked at the New York Times, we had this Pulitzer Prize winning legal reporter. And every morning he would wake up and he would refresh the Supreme Court website to see if any new audio transcripts had been published yet. And so it took him about 10 or 15 minutes to crawl to all the different pages on the Supreme Court's website to see if anything new had popped up. And then he would brush his teeth and come into work and then he'd check again.

And after a couple meetings, he would check again, do a couple phone calls, check again. And I just found this to be like a breathtaking waste of, you know, an incredibly high power, highly powerful mind, you know, to spend those extra minutes like pressing F5 on supremecourt.gov. You know, so we built him a little bot and all the bot does is tell him when a new transcript has been filed and it drops it off to him in Slack and then he can click on it and read it. And so, you know. RSS, you've reinvented RSS. Exactly. That's the thing is, this is the lowest of low hanging fruit. It's practically touching the ground. There's so many of those cases where, you know, a lot of the tooling that we build, you know, it does not feel

technologically superior. It feels like some real lightweight, real lightweight stuff. But in truth, we get a lot of mileage out of just getting the lowest fruit, you know? Um, and in particular, helping reporters out with little things like that, uh, means that there'll be more trustworthy, that our team will be more trustworthy to them when we start working with them on things that are like more sophisticated and require more, uh, honestly more trust on the part of the reporter that the, that the changes that we're making to their reporting process aren't, aren't bunk. So I have to ask, given this is the Screaming in the Cloud podcast, how has cloud technology impacted what you personally, I guess, and newsrooms collectively

do, if at all? I feel like I need to throw the if at all in there, but let's not kid ourselves. I don't think there's any industry that hasn't been touched by this. There's a pretty standard story. I worked at the Post in 2011 and 2012, then took a short break at the New York Times and at NPR in between before coming back in the last April. Oh, to go to detox, those small publications. I did get to tell my friends at the New York Times that I was leaving to go back to my hometown newspaper, and I don't think any of them found that very funny. No, I imagine they would not have. The Post in 2011 and 2012, when we were running election results, we were running them on physical servers that were located inside the building, like just around the corner from my desk. I could actually go to the server room and look at it if I wanted to.

We had five servers that we ran millions of page views off of. And we did not have root access to those servers because a kind soul in New Jersey had decided that we did not deserve to have root access to them. And so on the night of the New Hampshire primary in 2012, we were running Varnish on our own servers because we didn't have access to the post CDN at that time. We ran out of file handles. And so our Apache instances that were running on those servers slowly throttled themselves to death because they couldn't open any new network sockets. And as a result, we were down for about 15 or 20 minutes and not showing results pages to the world. And that was the time when basically right after that

election was over, I slept in the next day. I didn't go into work. But the Thursday after that, I walked in and I said to my boss, that's it. We're going to the cloud. I'm tired of all of this physical server baloney. Like we can rent servers. We can buy as many of them as we want for like six hours, and then we can just turn them all off. Well, less so on the election side, but it also seems to me just from a perspective of what actual real newspapers do with protecting sources and whatnot, you're one of the few people who can say that information security does have people's lives on the line. Yes, absolutely. You know, and there are definitely cases where, you know, we step back from doing things in the cloud. You know, we do a lot of

document analysis, which requires us to do OCR. And a lot of the good, cheap OCR in the world is cloud-based, which requires us to upload potentially hundreds of gigabytes of PDFs up to either Amazon or to Google to apply their OCR software to it. And so one of the things that we worked on, that I worked on at the Times that I'm working on here at the Post, is a large local system for OCRing and transcribing documents. So that if we have things that are really sensitive, we don't have to think twice about where we're putting them. And sure, we could solve some of those problems with cryptography, but that really feels like a now you have two problems sort of situation where, you know, just having like a beefy server here with like 96 cores

that can just like rip through a bunch of tesseract and you know turn a bunch of pdfs into actual text that our reporters can look at that feels like a thing that is okay for us to you know uh to say maybe not this maybe this one doesn't have to end up in us east one it seems like it's one of those areas where there's a number of companies that oh no what we do is so secret it's our secret sauce and our problems are so special and beautiful and unique that no one can handle these as well as we possibly could. And, oh, so what did you spend the most time on last year? Oh, replacing failed hard drives. Good, good. That sounds like a differentiated thing that everyone should be focusing on.

It's a core competency. One question that I think is probably going to be on a number of folks' minds is the Washington Post is owned by a patron, for lack of a better term, Jeff Bezos, the founder and CEO of Amazon. Is there any business relationship between the Washington Post and Amazon other than one might expect for, oh, we buy pencils off of them or we use them for cloud services? Is there editorial oversight? Is there, oh, you can use whatever you want

as long as it's AWS because that is owned by the same owner? You know, I have to say, early on, you know, even before Jeff Bezos bought the company, our need to go to the cloud in some way predated that quite a bit. And actually, the guy who is now our CIO used to be the CTO. His name is Shailesh Prakash. He had come over from Microsoft via Sun Microsystems.

So he's a computing OG. And he had he basically got here to the post, looked at this like litter of hardware that was like we were running in multiple data centers, including one in our own building. And they had said, this is ridiculous. We can't be doing this. You know, to your point about changing out hard drives, like, is it really a core competency of ours to like, you know, pull servers out of racks and like, you racks and make network cables all day long. Probably not, right? So there was this big push just about the time that I was getting ready to leave the post in 2012 to move basically everything to the cloud. And at the time, basically only AWS existed. I think there were some other lightweight solutions that were available. I think there were some other like sort of lightweight solutions that were available. Like I think there was Linode and like a handful of other things like that. But, you know, AWS had basically like scaled, like large scale kinds of things that we could use, like tooling that we could use. And so, you know, we made the call, we moved to AWS. It was grand. I left to go to the

New York Times. The Times, of course, made a huge move from AWS to Google while I was there, I think in 2017 or thereabouts. I have to tell you, that was one of the most breathtakingly difficult things that I have ever worked on. It's like trying to learn all the new idioms. It's basically the same thing, only slightly different between two huge cloud computing giants. When I come back to the Post, it was different between two huge cloud computing giants. When I come back to the post, it was basically like going home a little bit, getting comfortable again with AWS.

But to answer the question more specifically, I think that if we ever felt the need to move along, I don't know that we would ever have a problem with that. Our relationship with AWS is basically, please fix this problem. We're a really good AWS customer and not please fix this problem. We are going to call Jeff in the middle of the night and have him nuke it from orbit. You can threaten almost anything you want, it turns out. Only I could, yes. You can threaten anything. It doesn't mean you actually have to be able to follow through on it. I swear, my entire escalation point for most of these things is simply Twitter. Start tweeting something obnoxious and there you go. Well, we beg our AWS reps when they come to town. We have our punch list of things that we're

desperate to get fixed, you know, web sockets and a handful of other little things like this. But, you know, they know when they come here, they're basically just going to deal with us the same way that they deal with like any other customer, except I think we're slightly nicer to them. Well, from my perspective, that seems like a relatively low bar. Yeah, fair, fair. I don't think that they get like the, I don't think that that's a job that I could ever do, you know, travel between clients. Basically, all you ever get to do is hear them, hear folks be upset about like, what is it? You have no direct impact on anything that needs to be fixed because effectively, hi, I'm your account manager. I'm here for you to abuse more or less. And you, it turns out that

my misunderstanding of how large companies work is fatally flawed. I'm a five-person company, so I assume every other company is too. I assume when Jeff Barr, AWS's chief evangelist, isn't frantically writing blog posts, he's building the service he's about to write a blog post on. I assume everything he writes about he built himself single-handedly because why wouldn't he? It turns out companies don't actually work that way. So you're very often having to cajole folks internally to get status updates, to get information about what's going on in a timely manner to customers and in a way that doesn't inspire you know blind panic uh yeah so it turns out that region's down and we're not entirely sure why and when i called the team all i heard was screaming and then the line got

disconnected so i don't really know what to tell you huh why is our stock down 40 percent yeah it's it's one of those areas where messaging is important, message discipline is important, and outcomes are always challenging. On some level, I feel like that is aligned in some ways with the role that journalism has to play. In many cases, the journalist does not get to become part of the story. It's about understanding of the other players and telling a story about what's going on. Storytelling is one of those, I think, disappearing arts as we sort of descend down the well into page views and clickbait and trying to

drive outrage, more or less, instead of actual journalism. Given the hot button issues that a lot of your reporting tends to focus on, collectively as a group and what you're working on specifically, that extra care has to be taken not just to avoid conflicts of interest or anything in that vein, but rather the appearance of conflicts of interest. Yeah, absolutely. So a thing that I – I wasn't a journalist in high school or college. And honestly, until maybe even a few years ago, it was really hard to even think of myself as like working. I know I worked for journalism organizations, but I did not like self-identify as a journalist. And even now, I think like my team, even though they literally sit in the newsroom and we literally work on election results, you know, there still feels in my head like there's a little disconnect.

But in truth, you know, we're working on something that is what I would consider to be like the family jewels of the Washington Post, right? Elections coverage is practically sacrosanct around here. I'm lucky. I was blessed with this team of engineers, many of whom have never worked on elections before, but who are just working on side projects that were so clearly demonstrated an interest in this that I had no choice but to go thieve them from the directors that were running their teams. But they have all decided that, like as a team, we decided that we were going to follow the newsroom standards for how we conduct ourselves during an election year, which means everyone on the team, you know, we don't vote in primaries, we don't give money to political campaigns we don't put up yard signs and we do this not because it's like because we think that that'll make us fairer we do it

because it's critical that we understand our place and how people perceive the work that we do and so you know it's not fair it's not it's not fair to say to my engineers, you can't have an opinion about how our world works, but it is fair for us to say about ourselves, you know, we're going to do the same. We're going to treat ourselves the same way that like, say a political journalist would, and we're going to use those same standards of objectivity. I think that it is really great on the part of my engineers to sort of take this on. And I think that it's kind of great on the part of the post to trust us with those parts that are so clearly important to the organization. This episode is sponsored by ExtraHop. ExtraHop provides threat detection and

response for the enterprise, not the starship. On-prem security doesn't translate well to cloud or multi-cloud environments, and that's not even counting IoT. ExtraHop automatically discovers everything inside the perimeter, including your cloud workloads and IoT devices, detects these threats up to 35% faster, and helps you act immediately. Ask for a free trial of detection and response for AWS today at extrahop.com slash trial. One of the hardest things that I tend to see on the internet is whenever the Washington Post or the New York Times or someone else does, has a columnist or an op-ed or writes an article in a way that the internet, by which, of course, I mean Twitter, it's all the same thing these days, perceives to be a bad take, then suddenly the entire world goes nuts with, I'm canceling my subscription. Okay,

I get the outrage and I get wanting to send a message, but by the same token, there's an awful lot that's good. And everyone writes a bad article or has a bad episode once in a while. Not this one, mind you. But at some point at some point okay wow with all these people canceling all the time where do they ever have any customers and they must be out of business then you do a little digging wait a minute this is the third time this year this person claimed they're canceling their subscription so on some level it feels like it's a bit histrionic and it also feels on some level that you folks can't actually win, where no matter what you write, you're going to wind up irritating some folks. Part of that is

the element of speaking truth to power. And part of it also increasingly feels like we're living in a—we're all living in our own bubbles, however we tend to surround ourselves. And I'm as guilty of that as anyone else. I tend to assume every random person I pass on the street has an active Twitter account that they absolutely care about weird nomenclature from giant web services companies, and they have an unhealthy aversion to people mispronouncing acronyms. factually true. And every time I encounter someone where that shakes me out of that viewpoint, I have to stop and reevaluate. I do wonder how that winds up evolving over time, because it doesn't feel like it used to be quite this, I guess, partisan. Then again, I feel like we've been saying that forever too, and maybe it's just the world isn't changing. I'm just getting older and my perspective's shifting. Right. You know, I was just reading Poisoning the Press, which is a book about Nixon and his relationship with the press in the 1970s. As you assume that everybody that you meet has a Twitter presence and cares about the mispronunciation of acronyms, I assume that

everyone that I meet has a pile of unread books that's approximately three and a half years old and that they're barely making their way through them. So I have finally made my way through this book. And the thing that struck me about it was it described a time that felt, you know, not super unfamiliar to our current one, you know, felt like an adversarial relationship between the president and the press corps. It felt like real similar levels of discontent. And one of the things I thought that was really intriguing about that is that those times in the mid and late 1970s was also like a really difficult time for newspapers. There was an advertising bust around that time that made things really difficult. It was leading people to question their business models. I recall that the New York Times around

that time made the decision to break out a real sports section and focus on their features desk after a series of reader surveys indicated that these are things that people were interested in. And so the thing that I, while I don't think I disagree with your general feeling about this world, a thing that I'm sort of intrigued by is like the sort of instant feedback that we get from people, histrionic or not, you know, that's definitely a feeling that somebody has. And, you know, especially when it comes to things like elections, I really enjoy getting more or less instant feedback on things that we're trying out in various election nights. You know, we had this Virginia election very recently where

the state of Virginia, you know, just held elections for its House of Delegates and its state Senate, which are now controlled by Democrats for the first time. The House, the Senate and the governorship are controlled by Democrats for the first time. The House, the Senate, and the governorship are controlled by Democrats for the first time. And this was an interesting opportunity for us to get to test out some new elections features. And it wasn't the whole internet that was excited about it. It was a lot of Washington Post subscribers in the Virginia and D.C. area. But still, it was really nice to get that sort of instant feedback about what like what people liked and didn't like. You know, we take some of it with a grain of salt. Like, obviously, there's folks who are going to be like pretty upset.

But, you know, on an election night, it's not like we're getting people who are like, we're going to cancel our subscription because your tables were right aligned. So the stakes feel like they're a little bit lower for us, honestly. But it is it is interesting, and it's a difficult, you know, elections are one of those places where, you know, a newspaper has like a real opportunity to kind of reach out and connect to readers, you know? Everybody wants to know what happened or what's happening and who's winning and what's happening in our democracy. And it's one of those places where we just have like a real opportunity to show trust and reward people's trust in us. And honestly, it's, for me, it's like the most exciting time to work is like on an election night when the results are flowing in and you're sitting in that room, like watching all the, all the stuff happen. It's just like, it's great. It's like being at the nerve center of a democracy.

It really is interesting seeing how some of the sausage gets made to some extent. I mean, I happened to be in New York City last weekend at the time of this recording, and I walked past the New York Times building and holy crap, that thing's big. And then you remember there are bureaus scattered around the world as well. And huh, I never really, it never occurred to me that that newspaper that shows up or the apps that constantly get refreshed have more than a couple dozen people working there. Turns out it's kind of a massive undertaking for any of this stuff at scale. That's absolutely the case. You know, and the funny thing is like, you know, even here at the Washington Post, you know, we had like an election night for us, you know, involves something like

30 people, you know, about 15 or 20 of us huddled in a room, about 10 or 12 working remotely or via Slack, you know, and that's just for an off year election in our backyard, you know, a Virginia election and a Kentucky governor and a Mississippi governor, you know, and then if you can scale that in your head to what, you know, Super Tuesday or the New Hampshire primaries are going to look like, then you start to get a picture of like what it's like to be working in a place where the entire organization is just laser-focused on a single night and a single experience. It's just crazy. The energy in the building is palpable. I mean, you wouldn't want to drag your feet on the carpet because you'd probably set off a

thunderclap. It's a little crazy. But honestly, to me, it's just one of the reasons why I enjoy working for a newspaper. Because it's rare that... – normally if you were going to go work at a newspaper, you would have to work at a place where you were not using a whole bunch of your technological savvy, so to speak. A thing that I thoroughly enjoy about working at The Post is that I get to do like work on hard engineering problems, but I also get to do it at a place where on an election night basically everybody stops and is watching to see what just happened. Who did we just elect? What's happening? How are we going to explain this to readers? It's a good time. It really does feel like it's something that has a little bit more permanence than a lot of other things. I'm not trying to talk smack about various software as a service companies or startups or whatnot, but the Washington Post, the New York Times, these are societal institutions that have been here since before most of us were alive

and will presumably be here long after we're gone. We hope. I mean, you folks do have the permanent headline at the top of your website, Democracy Dies in Darkness, which is just, you know,

it's taking a while and the article isn't done yet, but I'm sure it's in progress. It's so goth, you know. It basically, the New York Times is all the truth that's fit to print. And the Washington Post is, you know, democracy dies in darkness. It feels like, you know, these are the two sides, the yin and yang, if you will.

I reward my employer for having the guts to go with the slightly grumpier, slightly gother tone. You know, we're scrappy underdogs. Democracy dies in darkness. Exactly. Get your t-shirt. Well, Jeremy, thank you so much for taking the time to speak with me. If people want to hear more about your thoughts on these and other topics, where can they find you? Well, I'm on the tweets because I am a person that you have bumped into before.

So naturally, I have a Twitter presence. I am at Jeremy Bowers on Twitter. Excellent. We'll put links to all of that in the show notes as well. Thanks again for taking the time to speak with me when you could have been doing literally anything else. That's okay. When I'm done with this, I'm going to go to a meeting about making sure that we have the right campaign finance data available for reporters when the filing deadline hits in January 31st of next year. Exciting. And I have no doubt you'll get there on time Jeremy Bowers engineering director at The Washington Post I'm Corey Quinn this is screaming in the cloud if you've enjoyed this episode please leave a five-star

review on iTunes if you've hated this episode please leave a five-star review in iTunes this has been this week's episode of Screaming in the Cloud. You can also find more Corey at ScreamingInTheCloud.com or wherever Fine Snark is sold. This has been a humble pod production stay humble