Get in touch

Show Notes

Improworsement: is an improvement that makes things worse, and Erin McKean knows all about that. She’s wanted to create dictionaries since the age of eight and this year she is celebrating the 10th anniversary of Wordnik, an online dictionary she helped create that has grown to 10 times the size of the Oxford English Dictionary. This week Paul and Rich sit down with Erin to discuss the evolution of Wordnik, from its humble beginnings in PHP to developing a full scalable API. Erin shares the challenges she’s faced, both technical and financial, and gives us tips on how to deal with failure. She also helps us expand our vocabulary and answers the difficult question: what is the best word?

Transcript

Rich Ziade What’s lovelier than a Word of the Day? Can we pause for a minute and talk about the Word of the Day? 

Paul Ford What’s better than a Word of the Day? . . . Nothing! [Music fades in, plays alone for 18 seconds, ramps down]. Rich, have you ever hosted a large API on the internet? 

RZ I ha— 

PF Maybe one focused [chuckles] on Readability?

RZ On words. I have. 

PF Hah! 

RZ Yes. 

PF How would you describe that, as a— as a thing to do? 

RZ Well, it depends on what you’ve got on your hands. Readability was widely used and [music fades out] it was, all of a sudden, expensive— like your agenda shifted to like, “How do we make them bill smaller?” [Laughs

PF Right [Rich laughs more], and there was also like you have a team on it and it’s hard to know what’s gonna happen next but it’s just terrible. 

RZ It’s hard to know what’s gonna happen next and you get emails of like, “Why—” And it was free [yeah]. It’s worth noting. The API was free and— 

PF But— but also a piece of fundamental infrastructure. 

RZ And then there was that. And so you started to feel responsibility, even though nobody was paying you. But it was interesting and cool and . . . a little stressful. 

PF So, there’s a person who we’ve called via Microsoft Skype, the preferred communications method of podcasters the world around. 

RZ Mm hmm! That is an unpaid endorsement. 

PF [Laughs] It’s— Skype is otherwise horrifying. 

RZ Yes. 

PF Who has exactly this experience. Has been through all of it. 

RZ Yeah. 

PF Started as a[says lexicographer slowly] lexicographer

[1:28]

RZ What is that?! Define that for me. 

PF Rich, I’m so glad you asked, I will define what a lexicographer does. A lexicographer makes dictionaries. 

RZ We’re going way down the nerd hole right now. 

PF Well! Let’s not go too far down that nerd hole [ok]. Let’s stop trying to define words ourselves, and bring in Erin MacKean to the show, via Skype. Erin, where are you? 

Erin McKean I am in San Francisco. 

PF What offices are you in? 

EM I’m at the offices of my extremely generous day job, Google. 

PF Oooh! Ok. We’ll get into that in a second. 

RZ Yes. 

PF I’m sure, like many from Google, you’ll be able to give us all the details of what you do in a completely transparent manner. 

EM Oh! Yeah, of course. 

PF Ok. So, let’s go back a little bit. The API we were talking about just a minute ago is called Wordnik. And it’s a giant dictionary on the internet. 

RZ Ok, explain. 

PF Well I shouldn’t explain. 

RZ No, you shouldn’t. Erin, explain what Wordnik and the history behind it. 

EM I like to joke that Wordnik is the biggest dictionary that nobody’s ever heard of. And it is literally the biggest dictionary. Full stop. And the way that it can be the biggest dictionary— like it is ten times or more than ten times the size of the Oxford English Dictionary. It can be that big because Wordnik has a different approach to what it means for a word to be in the dictionary. It turns out . . . that there was a study done in the journal Science back in 2010, where they looked at the Google Books’ corpus and they found that only 48 percent of the unique words in that corpus were in traditional dictionaries. Which means that more than half of the unique words of the corpus were not in traditional dictionaries. 

[3:06]

RZ How is— eh— I gotta ask: sorry, Erin. How is that possible? 

EM Well, English is a very Lego-like language, right? So take any dictionary that you care today, you could probably put any number of prefixes and suffixes on every single word in that dictionary and then, all of a sudden, you’ve doubled or tripled the number of words. So a lot of the words are morphological variants. A lot of the words are just really rare, really technical, one-off joke words. Like, for instance, maybe the word proud-shamed. 

PF Sure. 

RZ Mm hmm. 

PF Or— or— you know a better example is like ginormous.

EM Right but ginormous is used by a lot of people. The thing is is that writing traditional dictionary definitions is complicated; it’s time-consuming; it’s expensive because you need fairly well-trained individuals who take a lot of time to do it. So the number of words that a typical, traditional dictionary adds in a year is in the low thousands. And if you think of there’s being a backlog of millions of words, even the ones that aren’t necessarily like transparent, right? Like you would know what I meant if I said something was unginormous. You wouldn’t think that was a particularly elegant coinage but you would understand it. So even leaving aside the unginormous words . . . there’s still a pretty big backlog. But then the other thing is most of the words you learned in your life, you’ve never looked up. The number of words in your vocabulary that you actually went to the trouble to look up in a dictionary . . . is pretty small, as a percentage. Even for the nerdiest among us. So you put all those like facts together, and then you realize, “Ok. Most words aren’t in dictionaries, especially the rare or unusual words that drive people to a dictionary. Most people learn words from context, like reading them in a sentence. So why don’t we just skip the tricky, expensive, time-consuming part of writing definitions and just try to find the best example sentences we can for every single word?” And so that’s what Wordnik does. As a website, it tries to give you an example sentence and whatever other data we’ve managed to like scrape or license . . . Traditional definitions; synonyms and antonyms; people leave comments; people make lists; people leave tags. All the kind of like web two point oh user generated content stuff. And we think that if you read a couple of really good sentences about a word, you’ll pretty much understand and you may never even feel the need for a traditional definition. 

[5:28]

PF Sounds like you might’ve told people about Wordnik before. Um, first of all— 

EM I can pretty much talk about Wordnik until, like, the sun explodes. 

PF Of course, it’s ten years in, we gotta talk about that. Um most people don’t make dictionaries. So how did that happen? Like you’re— how did you end up being a person who makes an online dictionary? 

EM So, [smiling] I’ve actually wanted to make dictionaries since I was eight years old. I read an article in the newspaper that the second edition of the Oxford English Dictionary was 20-some years behind schedule and, for some reason, eight-year-old Erin was like, “That’s what I wanna do!” 

PF [Chuckles] “I— I wanna be 20 years behind schedule.” [Rich laughs

EM [Laughing] I— I think that no dictionary projects ever really come in on time. So 28 years is really not . . . that like [laughing] unusual. But— and, you know, I was a little kid in North Carolina and nobody that I knew— like nobody probably for like a 200 mile radius had ever met anyone who worked on a dictionary. The centers of dictionary production in the United States were Boston, New York, and Cleveland. And— 

PF Wait, what’s in Cleveland? 

EM Webster’s New World used to be in Cleveland. 

PF Oh. Ok. Ok. Oh! They closed it down? That’s like worst than the steel mill. 

EM [Inhales sharply] Uh yeah. And there’s no like retraining funds for [laughing] lexicographers. And then I worked on dictionaries my entire adult life,until I started working, like, more day job in tech a couple of years ago, and in 2007, I was the editor-in-chief of American dictionaries for Oxford University Press, and I gave a TED Talk. I think it is still the only TED Talk about dictionaries, so it’s pretty easy to find. And, at that TED Talk, I talked about why can’t we basically citizen science up the world of dictionaries and find all the words? And there was a lovely person in the audience who came up to me afterwards and said, “You know, this should be a start-up!” 

PF And you violently agreed. 

EM [Chuckles] I needed some persuading. 

PF Oh did you really? You weren’t— this wasn’t like a secret product pitch at TED? 

[7:30]

EM No! No! Like because, really, if you work at a large academic publisher, it’s not really a well-travelled path from that to start-up entrepreneur, but this— this person, Roger Mcnamee, who’s a lovely human being, he was like, “Hey! Well, let’s just think about like how would it work?” And since I really enjoy, like, solving theoretical problems that I think will have absolutely no effect on my day to day life, I was like, “Sure! I’m happy to talk to you. You see really nice!” And um— so we talked and talked and then we finally hit on the idea of the thing that was keeping dictionaries from scaling was the pesky definitions. 

RZ Let’s jump ahead: how do you pull this off? 

EM So, basically, we got some seed money and we digitized an amazing ten volume dictionary of American English called The Century and then we started grabbing text. So we Gutenberged it up— 

PF Meaning— meaning like Project Gutenberg. So thousands of books, freely available. 

EM Thousands of out of copyright books freely available; quite a few too many sermons; and quite a few too many, like, really racist, like, white-man-goes-and-imposes-all-his-cultural-expectations-on-the-people-of-some-place-very-far-away-type travelogue. 

PF I know! But it was free

EM It was free. 

RZ Take what you can get!

EM [Laughs] And we got a lot of texts through, like, through the Internet Archive and you could buy blog posts basically by the pound. 

PF Wow, so wait, you were buying, like, thousands of blog posts? 

EM Well— well, you bought access to a feed of blog posts. And I went to go see Chris Anderson at Wired and said, “Please, sir, may I have some texts?” And he was like, “Sure!!” And gave me, like, all this XML formatted Wired— 

PF Like on a— like on a— on a zip disc? 

EM Yeah [laughing] I think so! [Laughs] And so we just started, like, choppin’ up texts into sentences and in— So we started on Leap Day in 2008, and by . . . the end of the summer we had data for about four million words which was about four times the size of the OED, in terms of just raw wordage. And then everybody kind of said, “Oh hey! This might work.” And [chuckles] we rented some desks in an architect’s office in Dumbo cuz the other people I was working with all lived in New York. So I would like come to New York for, like, one week every three weeks, and we always had to stop all our meetings when the train went by. And, I think, the first version was like in some PHP framework because one of the people knew how to use that one [Rich laughs]. 

[10:05]

PF A true, true start-up [Rich laughing]. Like a little— 

RZ Yeah. 

PF— a little romantic; I’m sure everything’s going wrong; you’re travelling all the time; and, just, trying to make the world’s biggest dictionary. 

EM Right! And when I stayed in New York I actually sublet my best friend from high school’s studio on the Upper West Side and she would go stay with her boyfriend and I would, like, pay her 200 dollars to stay in her place for a week. 

PF Oh so you actually— you were—  you were exploring the prototype of AirBnB. 

RZ [Nasal exhale] There’s a lot ahead of its time. 

PF Yeah. 

EM Yeah! Except, you know, I had known her for ten years at that point, and, yeah, and so I was, like, commuting from the Upper West Side to Dumbo. 

PF For the non-New York listeners that’s not in anyway convenient. 

EM No. And then it kind of turned into a start-up in a weird way in that, like, somebody actually asked on Quora, like, what Wordnik’s business model, and the post is still up and I basically like pasted in the underpants gnome meme from— from South Park. 

PF Oh right! So like something, something, something, question mark, question mark, question mark, profit. 

EM Yeah, exactly! Like it was do something; collect words; profit. 

PF How do your funders feel about that at the time? 

[11:14]

EM So the . . . the idea was that if we had the best map of the English language then we could give directions between any two pieces of text. So, basically, we were backing our way into a recommendation engine. And we built one and it was good but we had the wrong business model. We made better recommendations, and we would go into publishers, and we would say, “Hey!! . . . People come to . . . your . . . property.” We had to call things like magazines properties because that’s what they called them. It just felt really weird. “People come to your property. And, you want them to stick around, and we could show them five other things that you’ve published that we think that they’re gonna be interested in, and then you can make them stay on your site longer and get more ad money!!” And they were like, “That’s very nice but Outbrain pays us five figures a month to run, like, this one weird trick ads, in that particular space.” 

PF Right, so you’re like, “Hey, I’m gonna make this more meaningful and semantically connected and wonderful—” 

RZ Useful, yeah. 

PF I’ve built the whole platform and they’ve said, “We prefer this huge pile of Arby’s roast beef sandwiches.” 

RZ Yeah. 

PF Rather than thanksgiving dinner. 

EM Yeah that’s because the people we were talking to always reported to the person whose budget got expanded by the money from the other content recommendation services. 

PF Yeah the world’s not really optimized for quality. That’s not the focus. 

RZ [Laughs] Or good— good intentions [laughs]. 

PF Velocity. So, now wait, tell the people what you built those. What I remember with Wordnik was like, “Hey, we’re gonna build a dictionary,” and everybody’s like, “That’s cool!” And then you came back and you said, “It’s a large scalable API.” And then, all of a sudden, you were doing lots of like API-ey things. So like what led to that? 

EM So, I really have to give Tony Tam, my cofounder with Wordnik. He was the technical cofounder, he joined— he joined right at the part that the creaky PHP thing started to creak alarmingly. I mean, you know, no dis to the people who built it cuz they like— we had no idea what the scale was gonna be, we were like, “Let’s just throw somethin’ together!” And I knew just enough about coding to know that I was qualified. Like I was gonna like admin the domain names, that was basically where I was. So Tony comes in and he, like, hires more folks, and he like sets us up with Mongo, like this is so early days of Mongo like— 

[13:34]

PF Back when it was unreliable, before it became reliable, and now it’s known as unreliable again. 

EM Uh. [Hesitatingly] Yeah. 

PF [Chuckles] Ok. Like it’s a journey. We’ve all been on a journey [Rich chuckles]. Ok, so Tony comes in and he’s like, “What have you done?” 

EM Tony never says, “What have you done?” His typing just gets louder. 

PF Right, every engin— but every engineer thinks it. 

RZ Oh yeah. 

EM [Chuckling] Yes. Every engineer thinks this. And— and so we were building this frontend and then we were thinking about like what else we were gonna do and he’s like, “This should all be an API.” And so all of a sudden everything’s in Java, and the frontend’s Ruby, and Rails. 

PF That was a very like serious— I mean people should know, like, fron— this is the frontend to the backend. Like this is like the back end that talks to even further backend.

EM Yes! Yes, we had like this— this very nicely architected app, then we had this big API, and Tony actually invented the API specification framework formally known as Swagger. 

PF Which is the one of the standards for sharing an API and documenting it. 

EM Right, now it’s called Open API and it has all the like that your heart desires, that he came up with it, at Wordnik. And I like to joke that it was because he was tired of telling me what API did what [mm hmm]. He’s like, “If I just solve this problem once, she will never ask me again.” [Rich laughs] I did like the horrible like . . . CEO that wants to be actually doing real work thing where I would work late at night and send emails to people asking them for things. 

PF You know what? That’s not horrible. I— I’m tired of CEOs being the villains for sending emails after 5:30 pm [Rich laughs]. I don’t think that that’s a terrible crime. If people ignore the email til the morning— for the morning, that’s ok. 

[15:20]

EM I think people are better about ignoring now. 

PF Yeah. That’s true. 

EM We’re very good about ignoring things in 2009, 2010. 

PF That’s good because— yeah, no, as far as I can tell, they’re experts at ignoring these days [Rich laughs]. It’s going great. Ok, so you were at home, smashing the keyboard, saying, “More words!!!” 

EM Yeah, so we had this API and Tony’s like, “Hey, why don’t we just open it up?” And I was like, “Yeah! Let’s do that. Let’s have a developer API.” And this was in late 2010. And then like, you know, ten years later 26,000 people have a key. 

PF Ok, so that’s— So for people to understand, too, like, you think about Instagram, it has 32 trillion users and you’re like, “What’s 26,000?” 26,000 is a lot of developer keys. 

RZ Keys to a lot of usage beneath each one of those keys, right? 

PF That’s right cuz each one of those— Some of those probably represent one person, but some of them probably represent a couple million people. 

RZ Right. So, walk us through what— why I would want the Wordnik API? 

EM Our biggest use cases are people who are adding dictionary definitions into other things, like you have something and you need two cups of dictionary. A lot of those are games. A lot of those things are apps that help you cheat at word games. 

PF Oh word games! 

RZ Word games. Got it. 

EM And a lot of— a lot of our usage is people learning how to code, like there are lot of glitch apps, there are a lot of tutorials out there that use Wordnik because it’s pretty easy to get a hold of a key. Like, you know, you just ask. It’s something that doesn’t really require a ton of specialized knowledge because the knowledge you need to work with a dictionary API, you probably learned in fourth grade. 

PF Right. Right, right. Here’s some definitions, and you can just— so you can just hit that, you get your JSON back and even if you’re early days, you can— you can build something with that. 

[17:10]

EM Yeah! And people build pretty cool stuff like a lot of people for a while were building ACD style password generators. One of our biggest use cases right now is something called a Magic Mirror. 

PF Yeah, no, you’re gonna have to walk— walk us through [Rich chuckles nasally] what that is. 

EM You know when you— when you see in a movie the guy shaving and his mirror tells him all sorts of stuff like, “Here’s the weather!” And it kind of J.A.R.V.I.S.-es it up and it’s all supposed to be very AI enhanced, well you can build a reasonable facsimile of that with things that you can buy at whatever replaced Radio Shack, and then you can power it with a Raspberry Pi or something and one of the things that people like to have on their Magic Mirrors is the Word of the Day. And we have an API that gives you one! 

PF What were the surprises? You’ve got— now you’ve got a surface that has the entire English language; and you’ve got nerds accessing it; and I’m assuming like when people come to Wordnik the first— they look up profanity immediately, that those are some hot spots. 

EM Oh! You know, I think Urban Dictionary has cornered that market.  

PF Oh! Right, right. 

RZ Mmm. 

PF Let them have it. Let them have all the swear words. 

EM I love Urban Dictionary. It makes me happy that it exists and it’s this tremendous release valve because people generally if they want to— to, [chuckles] like, if that’s the kind of experience they wanna have, they go and have it over there. 

PF Right, you don’t need to— you— you can shut the vile sex acts vertical at Wordnik [Rich laughs]. You don’t have to do that. What are the other linguistic hot spots? Like where do people— what do they wanna know? 

EM A lot of people come to Wordnik because they wanna make their own word list to help them study for the GRE. 

PF Ok. That’s utterly sensible. 

RZ Interesting. 

EM That’s not that surprising but a probably significant percentage of the first word list created by most Wordniks has something like “GRE Study” in the title. But, ok, so here’s my favorite, favorite user interaction on Wordnik: every day, for going on the last four years or so, one Wordnik has written a limerick about our Word of the Day. 

[19:11]

PF Oh. Ok, wait, wait, so we refer to Wordnik users as Wordniks? 

RZ Well, I was just about to ask this. Are people who use Wordnik Wordniks? 

EM People who use Wordnik are Wordniks, unless they are people who came through the site Wordy that Wordnik acquired in like 2010? A large chunk of them have hung around and now call themselves Wordyniks. 

PF Wow. Wow!

RZ Oh boy. 

PF That is both great but I’m also exhausted.

RZ [Laughs] Quick nap! 

EM Yes, because, you know, Wordnik’s not a venture-backed startup anymore. I mean it was for a long time but then it didn’t actually make a venture-style return. The recommendation kind of engine chunk of it got sold off . . . in an acquihire to a company that made pre roll video advertising— 

PF Oh yeah! 

EM— recommendation. 

PF Yeah, that’s the good stuff right there. 

EM But the investors who are all stand up people, said, “Hey, we love Wordnik, and we don’t want it to die, so how about we give you all of the intellectual property and a chunk of money and you re-incorporate it as a non-profit?” 

PF You know— 

RZ And a name? 

PF— that doesn’t happen very often. That is great. 

RZ What?!? What. 

PF Who are these— let’s— let’s actually say it: can you name these investors? Cuz like that is quite unlikely. 

[20:22] 

EM Yes! So Roger McNamee, I think he really orchestrated all of it and he— he was the person who personally donated the chunk of money to get it started as a— as a non-profit. It came from him personally, not from any funds he may have had. And so, yeah, but like Steve Anderson, and Baseline, and Floodgate, Mike Maples, and Mohr Davidow. Like, they were like, “Yeah, you know, this thing is obviously never gonna make any money. We have been trying to make money, you know in the meantime we built something pretty cool, so let’s just see if we can’t let this crazy woman run it herself?” 

PF Well, I like to see things where value isn’t destroyed. 

RZ [Clearing throat] It’s not just that, from what I can hear from Erin is that you didn’t read that as failure by any means. Most entrepreneurs are completely keeping score by money. 

EM [Giggles] Well, you know, when you go into lexicography as a profession [yeah]. 

PF Yeah, that’s fair. That’s fair. 

RZ Yeah. Yeah, yeah, yeah. 

PF Well, also, you did the thing! Like you were— I remember as— as Wordnik was growing like you did the things you set out to do and tried to prove out how this could work. 

RZ And it’s real and it’s known and it’s out there and it’s— it’s a thing that’s part of the fibre. You know, it reminds of Readability that way, in that it just kind of is out there and is used— 

PF It’s infrastructure. 

RZ It’s infrastructure, right? And that’s— 

PF I think that this is the gre— I’m very glad to hear that people are funding that infrastructure. 


RZ Yeah. Is that still the case today? 

EM So, we— Wordnik makes the money that keeps the servers running, and you heard me mention a day job, so that’s what keeps, you know, my kid in college. People donate to Wordnik by adopting a word. So, if you want to have your name in a link to your Twitter or website or whatever, next to that word, you pay us 25 dollars and we send you a certificate, [laughing] we send you some stickers in the mail. We did finally launch paid API use last year. So now people actually pay us to use the API. We also have a downloadable, offline, licensed dataset for game developers who don’t wanna connect to an API in their game. 

[22:27]

PF Oh I see so it’s in the download. 

RZ Yeah. 

PF Ok. 

EM Yeah! It’s like 190,000 words and definitions that are mostly like all the Scrabble legal words that you could just imbed, a big old JSON file. And that— that has been a really nice thing like because people were often asking us, “Hey! I just wanna download all your words.” And we’d have to say, “Please read the terms of service.” 

PF What was— what was the moment when you knew, like, “Ok, this is isn’t gonna explode. We need to find another path”? 

EM So about halfway through Wordnik’s startup life, the investors were like, “We think a new CEO is gonna, like, turn the tables. Change the tide. You know, do whatever it is.” And I was like, “Sure!” Right? Because like, you know, I was not waking up every morning with a magical path to a venture return. And I was like, “Yeah, I’ll just keep pluggin’ away on Wordnik over here, and, you know, the new CEO could do all the fancy things.” And, um— 

PF Right, but, “I’ll be your lexicographer.” 

EM Well, basically, like I was then, you know, founder. Which is just the worst, like the most amorphous title. And I was just like, all I— my main goal was like don’t be a jerk. And don’t let Wordnik die. And it was like if I can just do those two things then I am happy whatever happens. So then I realize like, “Oh my goodness, we’re not gonna get anymore engineering resources to work on Wordnik because Wordnik is no longer the thing.” Like we renamed the company, it was then called Reverb. 

PF This is— I mean this is a very, very common story. You know, you were gonna support ad tech, and you were gonna, you know, of course, of course. 

EM Yeah! All the things. So then I was like, “I’ve got to learn how to do all of the things that the engineers were doing because they’re not gonna be doing them.” So I started like writing tests for the Javascript of the website, and then I like just kept moving my way back through the stack, and then in the fall of 2014, I was, all of a sudden, a full stack engineer by default. 

[24:33]

PF This is— I mean this is great. It’s wonderful you learned it but it’s probably a suboptimal application of your skills and talents. 

EM So the thing is is that I started, when I realized like, “Oh goodness, my son is about to go to university and I need to have a day job that actually pays me money.” Being a mediocre and like community-taught developer is a really, really good qualification for working in developer relations because you have so much empathy for the majority of the people who use, you know, developer products. And so I started working as a— a Developer Advocate at IBM. I did that for a couple of years and I switched over to Google last fall. 

RZ In the same role? 

EM Uh, actually, now I work as a Program Manager, rather than a Developer Advocate but the job is pretty like help developers. 

PF How does one help developers? 

EM [Giggles] Work on open source strategy. And I’m mostly working on how to make documentation of open source tools and projects better. 

PF Oh my God, that sounds wonderful [Rich laughs]. 

EM It’s really nice. I get to work with so many really great technical writers and all of them want my Semicolon Appreciation Society stickers which, you know, not everyone always wants. Even when I force them on them. 

PF Oh, so you’re— oh, my God, so you’re— you’re like organizing documents; figuring out hierarchies, tables of contents; and then you have the programming skills to do all the parsing and figure out new ways to communicate. I’m gonna cry. 

EM It’s— it’s so nice. So there’s some quote about lexicography that I really like, that’s something about how every day you feel like you’ve advanced in some small way the great work towards completion, and every day is full of absorbing small problems, but nothing’s ever ruined. Right? Like you— it’s not like being a real scientist where like your hypothesis is completely disproved and then you have to like start all over. I really enjoy things where it would be incredibly difficult for me to make things . . . very much worse. But very easy for me to just continuing to make things a tiny, little bit better. 

PF What— what I hear throughout this narrative is . . . this literal can-do attitude balanced with a like, “Well, I probably suck at this but let’s see!” 

[26:44]

RZ Yeah, that’s what I’m about to say is like most people in our world and the tech world it’s just . . . you start with optimism but you get your ass kicked constantly, right? I mean there’s alway— it’s just so— it’s competitive, and it’s— it’s aggressive, and you’re probably one of maybe seven well-adjusted people on the other side of what you’ve experienced [laughs boisterously]. 

PF I think that’s totally real. You know what you hear though? As we psychoanalyze Erin [Rich laughs] is— no! It’s— they say, “Hey! Do you wanna do this startup around dictionaries? You don’t really know anything about it but I think you’re really smart.” And you clearly went like, “i’m pretty smart, sure, let’s see what happens.” 

RZ “Let’s see what happens.” 

PF As opposed to— 

RZ It wasn’t like, “I’m going to be wealthy and—” 

PF Or, “I must prepare myself to be found worthy.”

RZ Yeah. 

PF It’s just like, “Well! Let’s see what happens. And, you know, I have friends.” 

RZ Paul, when you come up to a founder and you tell them, “Look: it’s time to bring the turnaround CEO in.” That is a moment [Paul laughs] for those founders. It is quite a moment. 

PF They rare— I don’t think they rare— They rarely go like, “Well, I better start programming.” 

RZ [Laughs] Like, “You know what? That’s like a pretty good idea. Let’s give that a go.” It doesn’t land like that. Rarely does it land like that. 

PF It lands as like, “I have taken a life of failure and delivered only more failure.” 

RZ [Laughs] Or they start rationalizing it as in like, “I don’t really care much abo— I’m not passionate about this anymore anyway. Someone should take it.” They start to rationalize it. 

PF Obviously, I mean, Erin, you’re— you’re cheerful and positive, you’re not pollyannaish. I mean it’s like what— you know, you’re aware that things are going poorly sometimes and well the others. Like, what was your balancing mechanism? Was it alcohol? Or did you just say [Rich laughs], “To hell with it! I’ll just come into work—” Like is there any like, you know, what everyone listening to Track Changes wants to hear is just like a nice little aphorism that they can repeat over and over again, so can you distill— 

RZ Yeah let’s close this with a quote. A nice quote. 

[28:38]

PF Can you distill all of your life’s wisdom into maybe ten words right now? 

EM I have to say that at the time when I was really upset about things that I had no control over I often found myself thinking, “The only way out is through.” Which seems very dark, actually, in retrospect. I also ended up doing a lot of running then which my knees do not thank me for now. 

PF So get the serotonin up. I mean I think this is important like, you know, you’re a positive person but those were not easy days. And so like— 

EM I also found myself sewing like a new dress every weekend. 

RZ [Laughs] While running [laughs]. 

PF Sure.

EM Yeah well like I still have like all these moments of like existential terror— would you like to hear a horror story? 

PF Absolutely. 

RZ It’s gonna be one of the most optimistic, cheerful horror stories I’ve ever heard. 

PF Oh, let’s see. Let’s see. Erin can get in there. 

RZ Ok, [laughing] go, Erin.

EM So, you know . . . 2014, I’ve like taken over the whole thing and everything’s always broken on a website, right? There’s always something that has to be fixed; something that gets deprecated and you have to update it. And like the API because it was written by really good engineers was just kind of this solid like lump of Java that just did the thing. And it was always kind of like, “Yeah, I gotta update the API. I gotta update the API.” Cuz the— the lump of Java was literally a lump, like, one of the underlying packages had been taken apart to quote/unquote “fix it” and never put back together again so it could never be rebuilt. So I, you know, I was just chugging along, runnin’ off Mongo, and Mongo just gets older, and older, and older. And one day I wake up and it’s like, “Oh we’re on Mongo 212.” And which has now been like end-of-lifed for like— I think it’s older than your children, Paul. 

PF No, of course, of cour— I mean that’s the thing we actually— we think we’re modern cuz we’re on the web but it’s a legacy software stack, like any others at 25 years of age. 

[30:33]

EM Yeah, so M-Lab, who are awesome, by the way, said, “You know, we really have to turn this thing off.” And then they gave me two extensions and I got up this morning and I wrote the last endpoint, I got up this morning at 4:30 because the examples endpoint hadn’t been transferred over from the lump of Java to the new happy Javascript backend. I did actually pick an API framework called HAPI, H-A-P-I, and it’s very nice. And it makes me happy. And, yeah, so that’s what I was doing at 4:30 this morning so that I could switch it over because it turns off tomorrow


PF Oh that’s a good feeling, so the whole thing with the 26,000 developer keys could just break. 

EM Oh yeah any second. And if I get hit by a bus, like Wordnik’s bust factor is like one. 

PF Right. It’s you. 

EM It’s me. And I used to be like, “Oh my goodness, I can’t let anybody know this or why would anyone use this service?” And then I realized like HAPI? Maintained by one guy. Like every— now that I work more in open source it is terrifying how many things, it’s just one person. 

PF Well, do you remember the heartbea— heartbleed bug where all the security on the internet was down to a bunch of really tired dudes with no money. 

EM Yes. And so now I’m gonna be like, “You know what? Wordnik is my hobby. It makes enough to keep the lights on. My goal for it is to make enough money that when I retire, somebody can actually take it over as a job. I just want it to be in a good enough shape when I’m ready to like hang up my lexicographer hat, that someone else can take it over and have it not have to be their hobby. 


PF It’s a mission. 

EM Yeah our mission is to collect every single word of English and share it with everyone. 

PF What’s the best word? Just . . . the best word. 

EM See that’s the joy of working on a dictionary is that every day there’s a new best word and [chuckling] yesterday the word was um [giggles] improworsement . . . And so improworsement is an improvement that basically makes things worse. So I was apologizing to the Wordniks because like on the site some of the features are gonna be turned off while I migrate some data and I was like, “Hey, I’m making things better but in the meantime things are gonna suck!” Oh and [chuckles] so the Mongo turning off, I thought I had managed to trick the API into thinking that Mongo was still there so that all the user data which was in my SQL would still have to flow through the pipes. And it turns out . . . I didn’t trick it as well as I thought I did, and so instead of attending the Dictionary Society of North America meeting this last week, which I had taken off time to do, I spent four days feverishly rewriting all of the user access to use a managed user identity service instead of the old API. 

[33:!6]

PF You sound pretty good— sound like you, I mean— 

RZ You’re still laughin’ about it. 

PF Yeah this is what it sounds like, you know [chuckles], I never hear better stories than this. I never hear like, “And then it all worked great.” [Rich chuckles]

EM Yeah, there’s still some people who can’t login because, you know, they signed up in [laughing] 2009 and God knows how many email addresses they’ve had since cuz they don’t run their own domain names like an idiot. 

PF It’s not like they can’t use language in the meantime. 

EM [Chuckles] Right. And, you know, they can just email me and I’ll reset their email addresses for them. 

PF It’s all gonna work out fine. 

EM [Laughing] I hope so. But there’s a lot of of improworsement going on. 

PF Well, look, if people wanna get in touch with you, and if someone out there wants to take on the mantle of protector of Wordnik, what— how— what do they do? 

EM People can basically just go to Wordnik and hit any of the buttons that let you email anyone at Wordnik and they all just go to me. And they can always hit us up on Twitter, we’re Wordnik um Wordnik API [@wordnik].  

PF Alright, programmers, lexicographers, dev ops. Right now that’s you. 

RZ It’s a call to arms. 

PF Yeah, so other people should jump in, it’s time to help. There’s a global dictionary of English ready and— and willing for, you know, it’ll take you a minute to get things so you can take some help [music fades in] but come on everybody, it’s time! Step up! 

[34:34]

EM Thank y’all for lettin’ me rant about Wordnik for so long. 

RZ Thanks, Erin. 

PF Bye, Erin. 

RZ Take care. 

PF Take care . . . Hey, Rich. 

RZ Yes, Paul. 

PF Wordnik is a cool API. 

RZ It is. 

PF We like big, open platforms like Wordnik. That’s cool. That’s good stuff. 

RZ Open, big platforms that other things can rely in are— it’s just cool. It feels good, even though— even though they don’t have business models often. 

PF Hey [laughs]. You can’t have everything. 

RZ They’re still cool. 

PF But if you wanna build something big. 

RZ Yes, Paul. 

PF And you don’t have a team of people like Erin like immediately to your right, helpin’ you do it— 

RZ Mm hmm. 

PF You could conceivably go to an agency, a product studio that would help you get something like that done. 

RZ A digital products studio? 

PF That would let you build an API and products on top of it. 

RZ Where would I go? 

PF I would suggest you hit up postlight.com. I would absolutely [Rich whistles a long note, in amazement] say that’s a good place to go. 

RZ Ok. 

PF If you want a little personal connection: [email protected] 

RZ Yup. 

PF That’s an email address. 

RZ Paul and I look at every one of those emails. 

PF Every single one comes in, I respond to almost every single one. 

RZ Yeah. 

PF And unless it’s not really for me . . .So, get in touch [music ramps up, plays alone for six seconds, fades out to end].