Embedded - 86: Madeupical Word

Episode Date: January 28, 2015

86: MADEUPICAL WORD Erin McKean (@emckean) is a lexicographer, programmer, and start-up founder.  We spoke to her about Wordnik (the online uber dictionary), Reverb (smarter recommendations), and her many books. Wordnik: Adopt-a-word Developer Erin's favorite list Reverb Erin has written many books, some about words, one about dresses (The Hundred Dresses), and one fiction novel about The Secret Lives of Dresses. She has also given two TED talks. Watson on Jeopardy Brian Garner talks about skunked words in his book  Modern American Usage Five Intriguing Things via Tiny Letter [Feb 2, 2015: This link is broken today but it is the right link, google "Five Intriguing Things" to see if they've fixed it.] Elecia's Wordy project if fully documented over on Hackaday Reaction Housing is hiring!

Welcome to Embedded, the show for people who love gadgets. I'm Alicia White, here with Christopher White, and our guest, Aaron McKean, writer, programmer, and lexicographer. Before we get started, I want to remind you about Reaction Housing. They're looking for one electrical and one embedded software engineer. They build rapid development houses to and one embedded software engineer. They build rapid development houses to be used in disaster areas. You could be using PIX, sensors, and mesh networking to help people when acts of God turn their lives and their homes upside down. The job is in Austin, Texas with relocation possible, and I will add a link to Reaction Housing in the show

notes so you can apply. But now, let's get wordy. Hello, Aaron. Thanks for being on the show with us. I had some pun I was going to come up with, but I failed to do so. That's all right. Thanks for having me. You think the bar is really high that we have to have good words on the show? Yes. I wanted to impress with... Linguistics? I can't even come up with a word. So, there we go. Oh, the intimidation. I like all the words.

Could you tell us a bit about yourself? Well, I never know really where to start with that question. I like to make dictionaries. I've been doing it pretty much my whole adult work and life. First on children's dictionaries for Thorndike Barnhart, and then at Oxford University Press, and now at Wordnik. And I like to sew. I like to make stuff. So you were an editor of a dictionary at one point, is that right? Yes.

Well, editors don't really, I mean, dictionaries don't really have writers. You have people who write definitions, but almost everybody who works on a dictionary is kind of in an accumulation mode. And not so much like what you would think of as sitting down in writing mode. Right. Yeah. And you've done that too. You didn't list the fact that you have written several books. Oh, yeah. Some of the books were more agglomeration type books.

So like Weird and Wonderful Words, I did three of those. And then another book called That's Amore, which is also about words, like words for love. And then The Secret Lives of Dresses, which is a novel, which is about dresses. And then a book that came out in, when did it come out? Last year. Last year. Called The Hundred Dresses, which is like a field guide to dresses.

So what does it mean if you wear a flapper dress? What does it mean if you wear a June Cleaver dress? It was fashion and history combined. Yeah, and a little bit of, well, I hope humor, but just trying to point out dresses that are funny, like the Bjork Swan dress and dresses that are iconic. And I liked it for that, but this is a lot. You also have two companies.

I mean, maybe not right now, two right now, but you've had, there's Wordnik, which is the online dictionary. And then that kind of worked itself into Reverb, and now you're separating again? Yes. Tell me about those. So, um, so Wordnik really started, uh, the idea of it started back in 2007 because I was lucky enough to give a talk at TED about how, um, paper books were the, the wrong container to hold the English language. And in the audience, there was Roger McNamee who who came up to me immediately after and said, wow, that's a really interesting idea.

Let's talk about this. I think this could be a company. Because when I was, at the time I gave the talk, I was working on traditional dictionaries, and traditional print lexicographers don't ever really wake up one morning and go, wow, I really want to found a startup. But Roger is like a confidence generating machine. And if he thinks something is a pretty good idea, it's a pretty good idea.

So we talked for about a year. And then on Leap Day in 2008, we incorporated Wordnik. Good reminder day. Yeah, it's great because we used to joke that the company was less than a year old for most of its life, like less than one birthday. But when we started with Wordnik, we realized that the power of Wordnik was not to be necessarily... Wordnik's the biggest dictionary in the world by number of lexical items, by inclusiveness. But even being the biggest dictionary in the world is not necessarily a business that's going to generate a venture-backed return.

For instance, when dictionary.com sold around the time that we were founding Wordnik, it sold for about $100 million. And that's a lot of money to normal people. But in startup terms, it's not like enormous so the idea was that the technology behind word neck the thing that actually let you make a kind of machine augmented dictionary that could figure out what things meant that that technology is actually really valuable so is this natural language processing yes it's like n NLP and it's machine learning and it's a whole bunch of really cool stuff. But the underlying core idea was that if you know more about more words than anybody else, then you know what pieces of text are about.

If you think of every word as a data point and then you kind of graph the relationships of all those data points together, you can say, oh, this conglomeration of words in document A looks a whole lot like this conglomeration of words in document B, all these data points across the surface of the English globe. And then we used it with Reverb to give directions to people. So, for instance, if you're really interested in reading about, say, Node, you know, Node.js, then it will also show you articles about NPM and a whole lot of other node-related topics without having to know anything about you personally. Because actually, I'm a middle-aged mom. I am not the demographic that people advertising JavaScript to advertise to. But I'm interested in it.

And so the idea with Reverb is that kind of like what you are shouldn't determine what you think your interest should be, right? So if you're a middle-aged mom, you shouldn't only get, you know, advertisements for cleaning products and, you know, articles about how to make the world's best Halloween costume. Those things are interesting, but they're kind of broadly demographic. But if you have an idea of what topics people are interested in, you could show them more about that topic, and you don't care whether they're a mom or a senior citizen

or somebody who makes $300,000 a year or whatever. Sounds a little bit like Watson almost, except at the word level, right? Whereas Watson was linking concepts together. Are you, is there, there's an obviously an AI component to this that is sort of being glossed over, but it sounds pretty powerful. Right. I think AI is a really strong term because, you know,

you imagine, you know, reverb going on television and winning at Jeopardy, but it's, it's true. I kind of think that it's all about getting to the most precise signal. And a lot of times you use data just because it's there. And so, for example, using cookie data to understand what people are interested in, well, it was a lot easier to give, you know, put a cookie in somebody's browser than it was to actually look at the text that they were reading and to make inferences about it. And it's a lot easier

to say, oh, all your friends on Facebook like this thing, you must like it too, than it is to actually figure out what people like. And so the idea is that the harder the information is to process, the more valuable it ought to be. And we're kind of getting to the point where natural language is tractable, that it's possible to understand a piece of text, not in a human way, but at least in a rough machine way like Watson does, to say, oh, I have a pretty good confidence level about what this is about. Like at Reverb, we talked a lot about aboutness.

Like what's the aboutness of this piece of text? Is it about tennis? Is it about tennis stars liking particular brands of watch? Is it about, I don't know, Federer? watch is it about i don't know federer who's it about what's its aboutness and then trying to connect that to other kind of little islands of aboutness that people were interested in and so every word becomes a key word because it's partnered with all of its other words. And so together they make a theme?

Yeah, they have a connection. They have a relationship. And in fact, one of the things that linguists and lexicographers sometimes talk about is that words don't actually mean anything out of context. If I say the word toast right now, do you know whether you're going to be handed a piece of bread with jam on it or a glass of champagne? Could go either way. Or whether you're doomed. Or whether you're doomed, you could be toast.

And so it's the relationships of words to each other that give them meaning. Now, obviously, there are scientific terms that are always themselves wherever they are, but even they can be used in a metaphorical manner, right? You would say, okay, well, gold. Gold is an element in the periodic table, but something can be pure gold that is in fact not made out of gold. So every word has its kind of potentiality for

meaning that it only gets activated when it's in the right slot. Is that slightly unique to certain kinds of languages? I mean, is English worse than say Russian or Mandarin as far as specificity goes? Or is there a lot of context that's just because our language is kind of a mess? We do lack a one-to-one and onto sort of precision that I miss sometimes. I think all natural languages are messy and lots of constructed languages are messy too. Oh yeah, even programming languages are not. You can definitely rephrase things and get the same result, but totally different code. Yeah, I'm actually kind of interested in this topic and I'm starting to look into it a little bit.

I'm going to give a talk about the linguistics of JavaScript at the Fluent Conference this year. Oh, cool. JavaScript is really funny. I don't know enough about other languages to really say whether they have it better. I think languages that have more direct representation of syntax might be easier to pin down. Because if you have to give an ending that says something is, you know, a direct object, well, then that's what it is. But even then, there's so much idiomatic power in human language. It's hard to say for certain. You edited dictionaries, you, you did the words

and now you're a programmer. How, how is that journey? It's kind of funny. I was thinking about this the other day that it's taken me like 25 years to call myself a programmer. And I took a Pascal class in high school. I took AP computer science. I didn't take the test. And then when I got to college, I took a computer science class there too because it got you out of quote, real math. I took two corners of something called computer science as a liberal art and wrote stuff in HyperCard. That was a long time ago. It was not a long time ago. I still remember HyperCard. HyperCard was lovely. I enjoyed HyperCard a great deal.

And it's pretty funny because about every 18 months, I see another website or app that is doing the very same thing that my project in HyperCard was, which was enter in everything that's in your pantry and then tell you what the heck you could make for dinner. Yeah. So I think that's a perennial project. Yeah. I mean, it's got a little bit of database

and a little bit of going out and searching other stuff and matching. Yeah. Well, I mean, to some extent, Wordnik has to have a database. Yes. And then it has to, I mean, it's actually a pretty reasonable problem. I'm thinking about that pantry thing. Yeah, it's a pretty reasonable problem. And now, of course, it's all integrated with UPC scanners, so you can scan amazing. And now, of course, it's all integrated with UPC scanners, so you can scan things. Eventually, with the Internet of Things,

every can of beans you have will just ping some central server and say, eat me. I, A, hope not, and B, doubt it for financial reasons. Yeah. I don't know if there's a Moore's Law for cans of beans. Maybe Whole Foods beans. Yeah, the high-end beans. You'll pay extra for it to tell you to eat it.

And then the first real job I had out of college, I worked for an educational publisher. And most people who work for educational publishers are ex-teachers. And I was not. But they were all like, oh, Erin, you took a computer science class in college. You can do this computery thing. And so I ended up being given two O'Reilly Pearl books and six weeks of being told, hey, can you translate these files that are in typesetting code to SGML? I was like, okay.

I didn't know that I, you know, I didn't know what I didn't know. So I was just like. Sometimes that makes it possible. Yeah, I just spent all day hammering things into arrays and then pulling them back out again like over and over, and it worked. And then we used SGML-based dictionary editing software that was made by a lovely company in Copenhagen, which meant I got to go to Copenhagen several times to oversee data translation, and it ran on OS2.

So I had to do all the OS2 networking. And so then they paid for me to go get a Unix systems administrator certificate and to take a C++ class. And they were really hoping that I would be able to port this editing software from OS2 to the Mac. And this was in Mac Classic era. So they sent me back to Copenhagen and they had this agreement to let me look at the source code. And, you know, very nice Danes named Jens and Holger sat down with me and we opened up the code for the first time. And then the thing that nobody up until this point, after like nine months of discussion, had thought of was that all the code comments were in Danish. Which I did not read.

And then at that point, the company I was working for got bought by a giant other publishing conglomerate that had their own STML editing system. That was a bullet dot. Yes, exactly. Everything quietly went away. And then even at Oxford, I did a lot of data munging. We'd sell electronic rights to a database and somebody would have to put it into the format that made the tags just as small as they could possibly be because it was going to go on a device the size of a pocket calculator. Yeah. And once I left out the entire letter J of one data transform, and it took four months for somebody to miss it.

Not that important. Yeah, you know, J is a very small letter comparatively in the English language. Too bad it wasn't G, and then you really wouldn't have gullible in the dictionary. Let's just say that was a good thing. Jumble was missing. Okay, so you've been doing all these things, but you only recently call yourself a programmer. Yeah, because when we started Wordnik, I realized that like,

I just didn't know enough to know what needed to be done. So I started, we started with the idea of, okay, we're going to put an alpha together. And in four months, we had a working alpha that was four times the size of the OED in terms of number of words. And I was working with this really great guy who was the VP of engineering. He was the guy who was actually going to build the thing. We had a couple of good coders. And every day I found out some other thing that I didn't even know was going to be a problem. I was like, okay. The guy that I was working with,

Stan, he was like, you know what, this problem is really interesting, but it's going to be a much bigger database problem, and I thought it was more of an interface problem. Because we were talking about reinventing the dictionary, and you thought, oh, well, that's more like a UI. That's more of, you know, an interaction issue. But then he was looking at the scale of the data, and he was like, I'm not the right person for you. And then Roger introduced me to Tony Tam, who's the CEO of ReverbNet at this point. And he was like, oh, okay, I understand databases. I know what's going on.

This is how it works. But at that point, Reverb didn't exist. That was still WordNIC. Yeah, it was still WordNIC. So we started with the dictionary, and then the next step was we built a big API that we still run. There's more than 12,000 developers who have a WordNIC API key. People make everything from GRE study apps to Word of the Day and ways to cheat at Word games. And a lot of people build some beautiful Twitter bots off the WordNick API, like Darius Kazemi's done some great stuff with the WordNick API, like random Amazon shopper,

that's him, metaphor a minute. And then we got really into the idea of text recommendation and we would go we thought oh wow we can take a piece of text and we can find other related content and we built a related content system for blogs and my blog was one of the first blogs to use it like put it on dress a day and and it really worked and we'd go to publishers and we'd say, look what we can do for you. And they said, you're a giant dictionary company. What the heck do you know? And so we said, okay, well, we need a new name. We need to be called something different because Wordnik is too...

Word specific? It's too word specific. And even though all the technology was based on words, it didn't make the jump to publishers' minds. So, in fact, Mike Maples, who's one of Wordnik's investors, he said, you have a branding problem. And he set us up to work with Mike Cronin and Karen Hibma, who are the people who named the TiVo and the Kindle.

And they said, oh, okay. They were so great to work with, like really fun, just intuitive, thoughtful people. And they said, Reverb. And we're like, oh, okay, that's the name. Because you're looking for echoes. We're looking for echoes. We're looking for connections. And if looking for echoes. We're looking for connections.

And if you think of the initial piece of text as like the rock dropped into the water, the further out the ripples go, the more tenuous the connection. And then it had verb in the name. And there are just so many reasons. And it was a real word. It wasn't made up. It wasn't made up. I like made-ical words, but it was nice to have a word that was there.

And so we ended up being Reverb Technologies because there's a reverb design studio somewhere else in California. And so that went well and things happened. Yeah. And then you said, well, this is way too successful. I'm going back to the dictionary. Bye, guys. Well, Reverb had the problem that everybody wants to have, which is Reverb found product market fit.

The technology that was developed to do content recommendation, text-based content recommendation, works. I know. Whoops. I know. I know. I always sound so surprised. But I'm like, it really works. Like, stunningly works. Like if you think about when you see those really crappy recommendations at the bottom of a webpage, that's like 15 plastic surgery disasters of your favorite celebrities.

I just assume those aren't actually real recommendations of any kind. Well, yes, they look super fake. And the click-through rate on that, as you can imagine, is like on the most bored day on the internet, like maybe a percent. That seems high. Yeah, less than a percent of people will click on that. And when Reverb shows recommendations, it's two orders of magnitude over that.

So you're seeing like 8%. Inside the Reverb app, it was like 15% in terms of people going, oh, I want to read that. I want you to combine Reverb with Goodreads so you can give me new books to read. I know. It's harder for fiction because you like, just because one book is about, I don't know, ice fishing in the Arctic. Well, it's a mystery and the other ones may be a romance. And for me, it's more about how the words go together.

Yeah. I'm all about the story and the language. And I don't, you know, I love Dick Francis. Sure, horse racing, that's awesome. Oh, I love Dick Francis. But, you know, I also like space opera. Dick Francis is my go-to for when I'm really sick. Like, I'll just reread all of Dick Francis.

It's so comforting because his heroes are always these, like, ordinary guys who all of a sudden, like, have to step up and be heroic. And they're not morally ambiguous. No, they're nice guys. And not, like, nice guy, capital N, capital G, nice guys. They're like truly good people. I suspect we could do a whole show about that. But once you actually have technology that works and that people want, so people are coming to Reverb and saying, we want to do a proof concept, we want to work with you, this is really interesting. You have to double down and you have to focus.

And when we looked at Wordnik, we're like, okay, Wordnik is great. People love Wordnik. It does what it says on the tin. You know, you can look up just about any word. People, there's a thriving community of people who leave comments and make lists and add tags. And it's a fun thing. But it needed to be self-sustaining. Because otherwise it would be a distraction from the main thing that Reverb was focused on doing, which is the text-based content recommendation. So I talked to the board and I talked to my coworkers and we're like, well, what would make the most sense for Wordnik? I was like, well, in order for it to be self-sustaining, it should really be a standalone

nonprofit because then it can be a mission-driven organization with a mission being to share as many words as possible with as many people as possible of the English language. And they were like, huh. And they're like, okay, let's do that. Was Wiktionary, the Wikipedia's dictionary sub-area of their website, did it exist? Did you feel like you were competing with it? We kind of feel like we're augmenting Wiktionary. So, you know, Wiktionary is open content.

And so we incorporate Wiktionary into Wordnik. And lots of people devote, I think, good effort into making dictionary definitions for Wiktionary. so that at WordNic we really think that the way people learn words best is through context. And dictionary definitions are actually very lossy in terms of context. They're trying to be as general as possible description of the word, but that kind of leaves out the nuance and the color that lets you actually use a word correctly yourself. Most people know this as thesaurus disease. So you go to the thesaurus and you look up a word and you're like, oh, combustible, that's an interesting synonym for burning. And then you'd end up using it in the wrong context, like, oh, he had a combustible desire for his landlady.

That's not how that works and so but if you saw 10 sentences that use the word combustible correctly you would internalize it and most of the words you've learned in your life you didn't learn by looking them up in the dictionary you learn them through context but I do love the fact that now when I read on a Kindle or iPad, I can push a word and learn what the canonical definition is while I'm learning the context. Yes, but canonical is a hard thing. So most dictionaries are updated in kind of a rolling manner. So for example, the latest OED update started in the letter M. So they haven't really touched most of what is before the letter M. They're starting to work out of alphabetical order, but not a lot has

been updated. So if you think of the OED as what most people think of as the canonical record of the English language, if it's before M, you may have a wait. And words change all the time. And also there was a study that was done that was published in Science. I think it came out in 2010. It might have been 2011, where they looked at the Google Books Corpus. It looked at about 4 million books. And they found that 52% of the unique lexical items that were in those books were not in any dictionary.

So 52% of the terms or, in the books that were published. Yeah. Didn't exist in any dictionary. Yes. So that's word types, not word tokens. So if you think about it, if I have 70 cents and I have two quarters, a dime and two nickels, you know, I have five coins, but I only have three types of coin. So the word the shows up in the corpus like a bazillion, gazillion times. But a word like slentham, which is a kind of antique musical instrument,

which showed up in the corpus, it didn't show up very many times, but it also wasn't in any dictionary. So there are lots of unique words out there. And dictionary makers, traditional dictionary makers, just don't have the time to create the canonical definition. How do people learn these words to use them if they're not in the dictionary? They see them in context. They see them in context.

And the thing is that I keep saying over and over again that most journalists are better lexicographers than most lexicographers because they have to deal with new words with very little information to go on. And they end up writing these great sentences that explain in context and in passing what these words mean. So think of something really new like blockchain, right? Cryptocurrencies, Bitcoin. Every time you read a newspaper article about cryptocurrency, they will explain more or less what the blockchain is because you can't really go look it up anywhere. And so what we're starting to do, that was actually the inflection point where we realized that the thing that we wanted to do, which was record every word possible of English.

We were like, where are we going to get all the editors? How are we going to write all these definitions? And we're like, we don't need to write the definitions. All we have to do is go identify where these free range definitions are out in things that people have already written. To collect the 10 sentences with the context. And call that the definition? We show them as example sentences, but we try to rank them by how explanatory they are. And you do this through natural language processing, not through human interactions.

Not through human interactions. So early on with Wordnik, a couple of us were lexicographers, and we had kind of internalized the patterns that free range definitions use. So things like called by scientist X, or things as simple as being set off by em dashes are kind of hallmarks of a sentence where something's getting explained to you. And usually it's a word. And so if we can find all these free range definitions that are just out there waiting to be discovered and then compress them and put them all on one page at Wordnik, you can spend two minutes with that page, read 10 sentences and walk away with a better, fuller understanding of that word. even if you never see a definition, even if a definition never ever gets written. That's so strange. I'm sorry, my brain is...

It's hard. Yeah, I can see that because it is how I learn words. It's definitely the sentences. A lot of people feel like they get thrown in at the deep end because the definition is a life preserver. It's like the water wings. And you oh if i read the definition i'll know what this word means but oftentimes the definition gets compressed so small and tight because it's got this heritage of having to fit into a print book alongside 250 000 other, that it's kind of been freeze-dried, right? All the juice has been pressed out of it.

And so you might have a flat technical understanding, but all the color's gone. Yeah, and I mean, I only buy dictionaries that at least have some pretense at doing a small etymology so that I can see what the history of the word is. Yeah. Although history can be a snare and a delusion. It can be definitely misleading. Yeah. Because lots of times, um, um, lots of times we use things in a way, uh, that isn't true to the object's history. And, um, that's And that's why we have so many Pinterest pins of people using mason jars for decorative objects, right?

Like these mason jars, they're supposed to be full of canned vegetables and preserved fruit. And we're drinking lemonade out of them and we're filling them with bath salts. And the intent and the origin of the thing doesn't constrain our use of it now we can do anything with a mason jar that we want to that that we can and the same thing happens with words people get really upset about words like decimate where they're like well the original meaning was that you would kill one

in ten it's a precise word. It is a great word because it is very precise. So what you said about journalists in this whole discussion has set gears going. And what you're saying, and forgive me if I summarize this terribly, but it seems like what you're saying is the consensual contextual definition matters more than the original coiners intent for like the word blockchain. That has a very specific definition in a technology. Journalists may not perfectly understand what it means when they start using it,

even when they attempt to define it themselves in their articles. But if they all kind of define it in the slightly wrong way, the same way, that becomes the consensus definition. English is really just a shared delusion. We all agree together what something means. And if the consensus moves far enough away from the origin, then sometimes the term can become skunked, like Brian Garner, who's this great writer on English usage. So dictionary makers are supposed to record what's actually going on. And Brian Garner and people like Brian are the people who tell you what's good, right? So he's got this great dictionary of American English usage. And a skunked term is one where there are kind of two camps, right?

There's the camp that uses it in its original form, and there's the camp that uses it in the new form, and they dislike each other and they fight. And so you're encouraged as a writer to avoid skunk terms because you're just going to piss off half the people yeah like um people who are reconcilable yeah literally and like people uh lots of people now use enormity to mean really big instead of really bad. And so if you say, oh, she was surprised by the enormity of something, well, it's hard to tell which one it is. I didn't know that the former definition was the right one.

And there's stuff like forte, which I guess was originally pronounced fort, and now there are some people who steadfastly say fort, but then enough people have thought, well, you know, it's got that E on the end. It looks like it should be foreign. Let's say forte. You have to have an accent before you do that. And I just avoid using effect and effect in written communications. That is the safest course.

Because that's one where I know that it was one way, and everybody said it was absolutely the only way, and yet some people have said, oh no, we can adjust it. And I'm like, no, there's only one way. You can't just mess up language like this. But you're saying, yeah, because it is a shared delusion, as long as we communicate our delusion, it's okay? It's about what you care about and what you need to have happen.

So in just kind of general chatting with people, if you confuse affect and effect, it probably won't make that much of a difference. I'm Californian. They sound the same. That's true. It's kind of like language is a tool. So what do you need this tool to do?

And sometimes you need the very precise tool for the job because otherwise you're going to screw things up. And sometimes it's okay to use the butt of the screwdriver to drive in the nail. You don't have anything else around. You just have to get it done. And I really enjoy seeing kind of all the different kinds of Englishes. Oftentimes when people meet me and they find out that I work on dictionaries, all of a sudden they clam up and they're like, oh, I don't want to make a mistake around you.

I was going to ask, do people become critical of your language? Oh, yes. I used to write for the Boston Globe. I did like an every other week language column. that I kind of called more in sorrow than in anger, you know, where people would find a grammatical mistake that I had made or something that they thought was a grammatical error. And, you know, all I did was thank them because they paid me like the ultimate compliment, which is they actually read what I wrote.

They read it and thought about it and took action on it. Yes. That is a pretty high compliment, yes. Yeah, I'm firmly in the school that says any criticism, any feedback is a gift, right? And you just say thank you when someone gives you a gift. And it doesn't matter whether it's a gift that you particularly wanted or not. You thank them, and then you figure out what to do with it. You say things like, I can't believe the enormity of your feedback.

And let them wonder exactly what you mean. So I would say that I'm pretty lucky in the, in like the type and the amount of feedback that I get. Most people are cheerful. Even when they find a mistake, it's definitely of the more in sorrow than an anger type. They're like, oh, I couldn't believe that I found you making this error and i was like oh well yeah that's what i did and then like sometimes i learned something like oh i didn't know that that was considered 75 years ago to be like to be an error i was like oh all right but i mean stuff like this um i love it when people dig up old usage manuals. Like, for example, Five Interesting Things, Alex Madrigal. He has this great tiny letter called Five Interesting Things. And it might not be a tiny letter.

It's an email newsletter. And at the bottom, he gives excerpts from this 1957 usage manual. And sometimes you look at the advice that they were giving and it kind of feels like that kind of advice where they tell you that ladies never light their own cigarettes in public. And you're like, wow. First of all, who smokes?

And secondly, it was actually considered outré for women to light their own cigarettes. It's like so many levels removed of what you think of as appropriate behavior at this point. Yeah. Yeah. Very interesting. Yes. All right.

I get that. Do people also like rush to tell you their favorite word? I wish they did more often. Often they tell me the word they hate the most. So, you know, people say, oh, I hate the word irregardless. That was the word, not irreconcilable, irregardless. That's the one everybody hates. Everybody hates irregardless.

Not a word. Luckily, there's a very simple solution. If don't like it don't use it but i don't want anybody else to use it what about all those other people who are wrong it's it you know occasionally one will be put in a position of power say grading high school english essays in which you can make your desires known. And oh, there was, I think Jeff Bezos, right? Doesn't he have a memo format? I bet if he sent out a memo and said,

well, here are the words you can't use in the memo, he could control a whole lot of people doing that. Well, it works the other way, too. Didn't one of the U.S. presidents say normalcy, which wasn't a word? Oh, right, and then it't one of the U.S. presidents say normalcy, which wasn't a word? Oh, right. And then it was. And then suddenly it was. I think we've re-spelled nuclear at some point, too.

It's spelled the same, just pronounced differently. Right. Nuclear. I think it was Jimmy Carter that said nuclear, and he should know because he worked on a nuclear stop. He was a nuclear engineer. And Wordnik has a definition for irregardless. Yeah, because people use it. But you can fix that right now, right?

The good and the bad of Wernick's automaticity is that there's so much more data there than any number of lexicographers could accumulate and edit in one lifetime and at the same time it's like there's not much i can do right the data is there we promise to show you the data even if there's no data we'll try to show you how many times that word's been looked up because then you can say oh wow i'm the first person to ever look this up this has got to be wrong or you say oh wow person to ever look this up. This has got to be wrong. Or you say, oh, wow, 13,000 people look this up, but there's no data? What's up with that?

But this means people can just make up words. Yeah, I want them to. More English. More English is better. But less precision. It depends, because sometimes people make words that are very precise. Somebody sent me a word today um actually it might be on the community page of wordnik um we encourage people if they

want to invent a word right that the way to add it to wordnik is you look it up and then you leave a comment because it'll automatically generate a page. And what was this one? Oh, God, the temptation. Yeah, but why not? Because if people like it, they'll use it and then it'll be real. And if people don't like it, it will just fade into obscurity like, you know, hundreds of thousands of archaic and obsolete words before it. Thinking of sewing confusion with words that are just slightly different from existing words.

All of the misspellings. Right. Misspelling is pretty conservative. Like in English now, because we have such a text-driven culture, there's a lot less of the kind of reanalysis and misspelling that went on earlier in English. So my son was showing me something on iFunny the other day,

and it was a decorative planter that had a picture of a bird on it, and the word bird, only the word bird was spelled B-R-I-D. And then it had a big sign on it that said, SAIL, because here are all these planters, obviously manufactured in China, that have a misspelled word on them. And I looked at it and I said, well, actually, Henry, bird used to be spelled bread. And unfortunately, I have too much of a habit of pulling my son's leg

and he was like, uh-uh, ma. But no, bird used to be spelled bread. And there is no gullible in the dictionary. I swear this is true. This kind of like switching of letters happened a lot before people were really literate because, you know, sounds change. Brid is a little bit harder to say than bird. I don't know. But things like apron.

Apron used to be a napron, beginning with an N. And if you say a napron fast, it turns into an apron. And it got reanalyzed. There's a whole set of words in which this happened in English because spelling before dictionary editors was pretty loose and free. In fact, there's some joke, I think, about it being a poor sort of man who can't think of many ways to spell a word. I can believe that. Elegant variation.

Even now, US versus British spelling, there's many, many words that are very different. And then there are the ones that mean different things. Yes, those will get you in trouble. Pants. Pants is one of my favorites. And also, it's not so much the things that mean different things that have different denotations. It's the ones that have different cultural meanings. I once got in trouble for saying bollocks on BBC radio because they were like,

oh, are there words in British English that are offensive, that aren't offensive in American English? And I was like, la, la, la, la, la. Yeah, bollocks is one. Americans don't find that word offensive at all. And they were like, you can't say that on the radio. You just got bleeped. There's a bunch of those, right yeah all right i don't have no cultural baggage associated with that doesn't sound like anything but in

britain it's oh stop swearing yeah in fact most people who teach esl now they like there's really good curricular material about that for people learning English, because that's the kind of stuff that you don't get in a dictionary very much. It might say offensive or informal or colloquial, but it won't tell you, am I going to get punched in the face in a bar if I use this word? Oh, in learning Spanish, there are some words that... But so adding words, this is like the opposite of what the French did for so long, where they had a static, you may not change our language. Have. Still have. Still have.

Still have. The only words that are loud are the ones we say. Yeah. Well, the funny thing is, is that I think that, I mean, Americans in general seem to be all about choice, right? And especially out here, right, in California, it's all about let's be innovative, let's disrupt, let's make new stuff, and then... Is it evolution or revolution?

And then when it comes to language, it's like, oh, whoa, whoa, whoa, no. You know, I will drink Soylylent but i won't do this and and i think that it's it's interesting to me that people don't have the same attitude towards the language that they have towards kind of other things that they think oh yes try something new try something different see how it works and And especially when people say, oh, well, I hate cliches, but then they don't want to invent new words because a cliche is just a phrase in English that's outworn.

It's welcome. It's no longer exciting. It's no longer evocative. It's boring. So I think that the cure for that is to make up new stuff because a new word usually gets people's attention whether they love it or they hate it the French academy though, the Académie Française first of all, they get swords

and they have a special uniform with epaulettes and they get chairs, uniform with epaulettes, and they get chairs. Like, all the chairs are special. Like, the first member of the Academy who sat in that chair, it is their chair, so, like, there's Voltaire's chair, and people will often decline nominating. Not often, but there have been cases where people have declined being nominated to the Academy

because they're waiting for the person in Voltaire's chair to die so they can get that chair. So I think as pageantry, I would love it if English had an Academy. But in terms of the actual outcome, first of all, it's like King Canute trying to keep the waves back to try and keep words out of a language. And second of all, I think that the language really belongs to the people who speak it. And if they want to add a word, then why shouldn't they? That makes sense to me. Well, one of the things we talked about was your books. And I want to come back to that. Because we have had some episodes, episode, I guess there

were two where we really kind of described writing a book as a difficult enterprise. You've written both fiction and nonfiction. Do you have advice for people who are thinking about writing a book? So far, our advice has been along the lines of give up your nights and weekends and kiss your wife goodbye. Yes. It takes a lot of time. You're going to stick with that one too. Yeah.

I mean, I've really enjoyed, I think there are a lot of jokes about writers where they enjoy having written more than writing. Oh, yes. Oh, yes. Yes, I enjoy having written. And I really love it. Like, every once in a while, I'll get an email from somebody who really, really, really liked the novel. It was a really good novel. Oh, that's really nice of you.

I mean, I liked writing it. It was fun for me. I like writing kind of, I like that I was able to write kind of a soft genre book because I'm not really a literary fiction person. And so I was like, okay, cool, more of what I like. But I think fiction is hard and there's so many other easier things to do and publishing is hard. And I was lucky because I'd already been working in publishing for a long time.

I mean, it was an academic publishing, but I was also an acquiring editor. I like looked at book proposals and made recommendations as to what the press should buy and got things published and worked with authors and did the, you know. So you knew the industry, not just the nonfiction, but the fiction industry. Well, my sister is a literary agent. She's not my literary agent because the same way that she wouldn't be my dentist if she were a dentist. But she also is a huge help because she does young adults. She does mystery.

She does genre fiction. She was the agent for the I Can Has Cheeseburger books. So, yeah, she's a really good agent if you need one. But really, I think all of the traditional advice is traditional for a reason. Sit down and write every day. Practice. Practice. And don't think that every word that comes out is the last word. It's not. Definitely kill your darlings. Yeah, all of that advice is advice that

gets repeated for a reason. And I think people, as with so many other things, people want there to be a magic shortcut. I think people get confused too, because it's one of those few things that everybody does every day. We all write, we write emails, we write letters, we, you know, all put words to paper or to computer screens. But we don't do that in a professional capacity, all of us, or in a specific genre, writing a story, writing fiction. That's not just putting words to paper. It's the difference between driving to work

and being a Formula One race car driver. Yes, you're both drivers. Yeah, it's kind of like trying to figure out, okay, well, what's going to happen next? Like, what needs to happen on this page in order to get somebody to turn it? And that can be really difficult. And so you added author, both fiction, or I guess novelist and author, if we're using the proper terms. And you have mother,

and you definitely sew quite a bit. Yes. You sew almost all your own clothing. Yeah, I pretty much wear exclusively dresses and skirts. And if it's a dress or a skirt, I'm 99% likely to have made it. And programmer, we've checked that one off. Yeah, not a very good one, but still. You get to own it.

I enjoy it so much, like a really ridiculous amount. So it's just fun. It's like magic. It's like Harry Potter. You say something to the computer and something happens in the real world. Also, I really love writing tests. So that just makes me really happy. It's like the crossword puzzle of programming.

Oh, writing computer. Yeah, writing software tests. SAT tests. Oh, no, I hate writing those. Sort of. But I like writing tests, software tests. SAT tests. Oh, no, I hate writing those. Sort of. But I like writing tests, software tests. And you've been an entrepreneur. Yep.

And now I suppose I am a social entrepreneur because it's a nonprofit. What other identities do you have? I wish that I could say crime fighting superhero, but I'm a little too live and let live to be a really good crime fighter because somebody would be jaywalking just in front of me and I'd be like as long as they look both ways. When I worked at ShotSpotter, we made a gunshot location system and we would joke about putting crime fighting on our resumes. Oh, yeah.

You can do it. Not bother with the day walkers. That's petty crime fighting. Petty crime fighting. Captain Meter Maid. Yeah, exactly. How is writing books similar or different to writing code? I think they both go better when you have an outline or an architecture in your head.

I think it doesn't need to be in your head, but okay. Or, you know, somewhere. Because I think, so this is a funny story. When I was writing The Secret Life's Addresses, I didn't set out to write a novel. I had written all these little like vignettes of secret lives addresses on my blog. And I started getting agents sending me emails going, do you want this to be a book? And because I'd worked in publishing, I knew that books or short stories don't sell and don't make any money. So I was like, well, I should probably turn this into a novel because novels actually sell. And so I started working on that.

And I, like, sat down. And I sat down to work on the novel like I would sit down as a reader. And I sat down and started at the beginning of the story and thought I was going to, like, read straight through to the end and then stop. And, like, after about a month of, like, sitting down every day to write, I was like, I don't really know what happens next. I'm just kind of like writing off into the void. So I started, like I stopped and I hadn't started reading about plot and like, oh, okay,

well, this is the kind of stuff that needs to happen. And I had a really good editor at Grand Central who helped a lot. And it's like, oh. And the same thing kind of happens when I'm trying to write code. You know, like, okay, well, what do I need to have happen at the end? What are the inputs and what are the outputs? One of my favorite formulas for writing a novel was said by this woman named Kathleen Norris, who was a best-selling, like Stephen King

level best-selling novelist in the 20s and 30s who lived in San Francisco. Most of her books are set in San Francisco and it's this wonderful insight into the time. And she was a pacifist and a pretty devout catholic and her formula for a novel was get a girl in trouble and get her out of it and i was like sometimes i think about code i'm like okay get the data in trouble and then get it out of it like you know what what's the data going in what's the data going out and what has to happen in the middle and um her books, like, after I moved to California, I reread a lot of her books. And, you know, they're very typical of the time.

And they're very soft. I think we would think of them as being very soft. And there are a lot of convenient deaths because she didn't believe in divorce. But, like, reading about rich people in Burlingame, well, there's still rich people in Burlingame today. The world has not changed all that much. There's still people eating oysters in San Francisco.

It's still foggy. There's still cable cars. They're really kind of beautiful books. And the descriptions of the clothes are amazing. Who was the author? Kathleen Norris. I'll put it in the show notes. There's also a Kathleen Norris who's a spiritual writer.

And I've never read any of her stuff. We'll link to the right stuff. But if you Google Kathleen Norris, often what comes up are books entitled, like, you know, Walking Meditations, but that's not what you want. You want books called something like The Rich Mrs. Burgoyne. I think that's an actual title. Do you ever put stories in your code? I sometimes think of my code as it should have at least a plot

and occasionally a joke in the comments. Not forced, but sometimes you can keep people reading your code and I think that that's important. I hadn't thought of that. There's a Twitter account I really like that's called something like Developer Swearing, which is like get commit messages. That's just like my Twitter account. It's like get commit messages that include profanity. It's a bot, right? And I often I think a lot about being kinder to my future self

when I'm writing code, except I think my future self is kind of a dumbass. So I leave comments like, remember that this is here because there's an error in the API and this is what you have to do to make this work. Dumbass. Why don't you remember this yes i don't leave stories so much as i leave kind of like post-it notes to whoever's gonna have to read it next it says this may look stupid but it's stupid for a reason so don't go think you're smarter than me right now because you're just going to revert it back to the thing that didn't work.

Well, I have to relate this to embedded systems somehow. Oh, well, language is like the key embedded system. It's embedded in all of us. Oh, yes, indeed. But you mentioned that WordNet has an API. Yeah. And I have my little ring that when I tap it, it gives me another word and then a definition. Oh.

Yeah, so it just decided vacuous and then it gave me the definition. Oh, that's awesome. Yeah, so that is incredibly amusing at parties when I sit off in the corner and amuse myself. I think you should use that for like, aleatory of like, determination, right? I do sometimes. Yes. What am I going to do next? Tap the ring? Use whatever word is like a signal.

Because I do tend to be a little socially awkward. I will look at the ring and play with it for a minute or two and then I'll like, oh, you know Colo Cuter came up. I should go talk to someone. Oh, I like that word. Although I'm not, about half the words came from your favorites list because I was looking for strange, interesting words and the other half are a GRE set that I pulled from somewhere. Yes, it's good to do it off the lists for the API because one of the improvements that we're hoping to make, not too far off now, is to

improve the random word generator because it turns out true randomness is kind of boring. Oh yeah, true randomness on Wikipedia leads you to a lot of manga articles. Oh, yeah. True randomness on Wikipedia leads you to a lot of manga articles. Oh, I can imagine. And in fact, the same goes with Wordnik because we imported a lot of dictionary content. They had a bot go through and put in all the inflected forms for all the verbs. So you'll get things like- Like walk and walking, walked.

Yes. Okay. Yes. And that is not very interesting. I did build an interestingness generator a while back. Like when I was working on the first like weird and wonderful words book, I procrastinated an enormous amount. And then, and I didn't get any time off. That's not right for writing.

Right. I didn't, it was part of my job because the like, my boss said, hey, we want you to do a weird words book for the next season's list. I was like, sure, fine, no problem. But I didn't actually get any extra time or lose any other things that I had to do. And I also kind of had a new baby. And long and short of it, I had about a week to write 1,000 entries. And I was like, I can't just sit here reading through the oed

without a plan because i'm not going to find weird words they're all going to be start with a right and i set up this regular expression search that looked for words that were over over a certain length that had greek and latin in their etymologies that included like the less common word letters in english you know like, like K. Oh, yes. And then just got 10,000 words. And I was like, 10,000 words, that's manageable. And like read through that and then just kept tweaking the search to try and come up with weirder and weirder things.

So there's like in the same way there's that Neil Simon quote about words are only funny, like words that start with K are funny. Words that have a K in them are interesting. And I use the reverse search on Wordnik a lot. Like if you go to the related section, then there's a list of definitions that have the word that you just looked up in their definition. So looking up something like crime and seeing all the words that have crime in their definition or looking up something like pallor and seeing all the words that have that in that definition,

then that can get pretty interesting. That sounds like fun. I need to go try that. It's a lot of how some of the words of the day are selected because we're like, oh, well, let's find something unusual. So Wordnik is not-for-profit. We have incorporated it as a not-for-profit company, and we're working on getting our official IRS 501c3 ruling. Okay, so a dictionary that's not-for-profit,

that really goes against all known things out there, other than Wiktionary, which it sounds like you're incorporating that. How are you going to make any money? Are you sending your son to the mines? I would like to send him to the vowel mines and he could just dig up words for me all day. So we want to be a community-supported site. And to that end, we launched on January 1st an Adopt-A-Word program.

So just like you could adopt a highway, you can adopt a word. And so you get your name on the word and your name can link to your Twitter handle. And adopted words will get first priority for the data refresh that we're working on. So your world will be fancy and shiny, just like a highway that's had the trash picked up off of it. And then we're hoping to launch a new version of the API soon, which will actually have premium data in it. So we've got a brand new dictionary coming that will be available through the API, a new commercial in copyright modern dictionary. So people who want to use that for GRE apps or to embed in their e-readers, that'll be available as a paid API service. We'll always have a free API service. And if I want to use the free API service to update my ring or to

build something with Wi-Fi that has a word of the day, how do I do that? You go to developer.wordnik.com and you sign up for a key and then you email us to get your key. That is one of the parts that did not transfer well in the move over to the new servers. The coordination server got a little uncoordinated. Let's put it that way.

That should be fixed soon. But we generally turn the keys around pretty fast. And then you read our terms and conditions to make sure that you are following them. The really important one is that we want people to be sure to you are following them. The really important one is that we want people to be sure to credit the dictionary text correctly because we license these works, whether they're from an in-copyright source

or whether they're a Creative Commons licensed source. And in order for Creative Commons licensing to work and for us to be able to provide the data, we need people to. If that dictionary definition is going to be seen by human eyes, other than just your own, you should include the attribution information. That's the one thing that trips people up when they write to us about,

hey, is my app conforming to your terms? We say, oh, please list the publisher. And so my little ring here would have to say that the definition came from the American Heritage Dictionary or Wixnary or the GNU version of Collaborative International Dictionary of English. Yeah, that one's pretty much just called G-Side. It sounds like something you're killing. That's the open-sourced updated version of, I think, the Webster's 1913. But these, so you are actually paying

to get these definitions from various people? For American Heritage, we have a licensing arrangement with them. And the Century Dictionary? The Century Dictionary is fantastic. So the Century Dictionary is the best dictionary that nobody knows about. It was the only American dictionary to ever try to compete with the OED. It was sold in like 12 volumes in the early 1900ictionary kind of overlooked. Technical terms, Americanism, scientific terms. It was fantastic. And they kind of overreached themselves and then there was a depression and then they just kind of went out of business. But the definitions are great and the etymologies are actually really good. We don't show as many of them as I would

like, unfortunately. But the early editor of the century was one of the first Sanskritologists in the United States. So the etymologies in the century are just as good today as they were 100 years ago because he was ahead of the curve. Everybody else had to play Sanskrit catch-up, but not Whitney. He knew what he was about. It's a beautiful dictionary. I have two copies of it at this point, which is kind of a lot to have 12 volumes in paper. And I don't look at them because, of course, it's in Wordnik. But I would probably rescue them from a burning building because they're just that great. That sounds like a lot of fun.

And I think we are about out of time, although I am not out of questions. Sorry, I could talk about Dick Shares pretty much until the heat death of the universe. Well, I feel that way about embedded systems, so we're good. But how can people try Reverb? Because I want to go back to that because it was pretty neat and it's a neat company. You have an app? Yes, there's an app for iOS available in the App Store. It's just called Reverb App. It has a big upside down R.

And then I would like to say that with any luck, it will soon show up to be powering, you know, the parts of the webpage that you're probably ignoring right now. Get better recommendations. Yeah, I mean, why would you waste a reader? If you've got somebody on your site, interested in something, enough to read an article, why would you send them off to

celebrity plastic surgery disasters when instead you can send them to content that's actually related to what they just read? To me, it makes perfect sense, but a lot of things seem to make sense to me that don't make sense to other people. So anyway, yeah. And you can always go to helloreverb.com and take a look at what's going on there. All right. I can try that. Chris, do you have any questions?

I have one more question, which is the question you probably hate getting asked a lot. What is your favorite word? So I try not to play favorites, because I have to cheat the whole English language equally. It's just like kids. When pressed, I do say that my favorite word is erinaceous. You know, my name is Erin, and erinaceous means of or like a hedgehog. So I just think that is awesome.

Hedgehogs have their own adjective. So it has resulted in people sending me hedgehog stuff. Like my mom now sends me hedgehog stuff. And I, I think hedgehogs are adorable and thank goodness for all the hedgehog gifts out there now. But I do like the word slightly better than the animal. If that's not the best answer, it's indistinguishable from the best answer.

Exactly. It's isomorphic for the best answer. So Erin, do you have any last thoughts you'd like to leave us with? Oh, just thank you so much for having me on. It's so much fun to talk about this stuff. I hope you had fun too. I did. And I suspect you and I will chat some more after. I wanted to ask a little

bit more about adopting a word before we sign off. Oh yes, thank you. If you just go to wordnik.com slash adopt a word then you should be able to adopt just about any word. If you type in your word into the box it will tell you if the word has already been adopted. And petrichor and embedded have already been adopted. I can tell you that. Those are two really good words.

Petrichor is a favorite. Yeah, I like that one. It's a great thing too. Like sometimes great words don't have really great things but petrichor, that like lovely chemical after rainfall, that's a really great thing. It's nice that there's a word for that. Yes, that's why I think people should make up more words so there can be a word for everything. It was a little expensive. It cost $50 to adopt a word.

Yes, so we're still experimenting with that, so probably the early adopters are going to get a lot of schwag. I've just started designing more schwag. Excellent. Oh, I was supposed to remind you to bring me a t-shirt. I have it. I have it in the bag. and Wordnik and current head of Wordnik. If you'd like to adopt a word, as long as it isn't Petrichor or Embedded, you can do so at wordnik.com slash adopt a word. I'll have a link in the show notes too. As always, thank you to Christopher White for co-hosting and for producing. And if you'd like to say hello to us, hit the contact link on embedded.fm or email us show at embedded.fm. I want to thank all of you for listening,

and I do have a final thought for you. This one is from Stephen Wright. He's a comedian I've never really heard of, but I liked this quote. I was reading the dictionary. I thought it was a poem about everything.