Anthony Shore’s naming partner is a neural network
Anthony Shore is one of the most experienced namers out there. He has over 25 years of experience in naming and has introduced more than 200 product and company names to the world. Some of the names he’s created include Lytro, Yum! Brands, Fitbit Ionic, Qualcomm Snapdragon, and Photoshop Lightroom. In 2015, he was featured in a New York Times Magazine article titled “The Weird Science of Naming New Products,” which tells the story of Jaunt, a VR company he named. And a BBC News article called him “one of the world’s most sought after people when it comes to naming new businesses and products.” Anthony has led naming at Landor Associates. He worked at the naming firm, Lexicon, and now he runs his own agency, Operative Words.
I had a great time talking to Anthony. He shares a bunch of knowledge, some great tips and examples, and we even got to nerd out a bit talking about recurrent neural networks. Anthony’s using artificial intelligence to supplement his own name generation; it’s fascinating to think about how tools like these might be used in the future.
Anthony also gave a great overview of his naming process and provided a list of tools and resources he uses when generating names. Some namers I’ve talked to seem to prefer analog resources (i.e., books). In contrast, Anthony almost exclusively uses software and online tools*, including the following:
- Wordnik (“a great resource for lists of words”)
- Sketch Engine (a corpus linguistics database)
- TextWrangler (a plain ASCII text editor)
- Microsoft Excel
Anthony and I rounded out the conversation talking some of his least favorite naming trends, as well as what he likes most about being a namer.
I highly recommend you check out Anthony’s website and blog at operativewords.com, where he has a bunch of amazing content that goes into way more detail on some of the topics we discussed. Below, you’ll find the full transcript of the episode (may contain typos and/or transcription errors). Click above to listen to the episode, and subscribe on Apple Podcasts to hear every episode of How Brands Are Built.
* To see a complete list of online resources listed by namers in episodes of How Brands Are Built, see our Useful List: Online/software resources used by professional namers.
ROB: Anthony, thank you for joining me.
ANTHONY: Thanks so much for having me, Rob.
R: One of the first things I wanted to ask you about is something I don’t talk to namers about that much. It’s artificial intelligence. So, I saw that you’ve written and talked about the potential for using neural networks and brand naming. Can you tell me a little bit about what made you start down that path and then maybe how it works today?
A: Sure. I love talking about this. Artificial intelligence, and really using computers in general as an adjunct to what I do, has always been near and dear to my heart. Way back in college, I created a self-defined AI major. And so, when recurrent neural networks started becoming available and accessible over the last few years, I took an interest. And a woman named Janelle Shane, who is a nanoscientist and a neural network hobbyist, started publishing name generation by neural network. And this really caught my interest. And she was doing it just as a hobby and for fun, but I could see that neural networks offered a great deal of promise. And so, I engaged with her and asked her to teach me what she knew, so that I could also use neural networks to help me create brand names, in addition to using the other tools that I use, like my brain and other bits of software and resources.
R: And is there…how technical is it now in your use of it? Is it something that anyone could do or does it really require a lot of programming knowledge?
A: Well, right now I’d say it’s not for the faint of heart. The only interface that really helpful is through command line, really using a terminal. So it’s all ASCII. It’s done in Linux and there’s various and sundry languages that have to be brought into play like Python and Lua and Torche.
R: So you’ve got to know what you’re doing a little bit.
A: Yeah yeah. It’s not something that’s just a web interface that you plug ideas in and it’s going to work like a charm. Now, that is right now and it’s changing constantly. I mean, even in just the few months, six months that I’ve been doing this, I’ve been seeing more and more neural networks front ends on the web pop up. But their results aren’t very good at all. But it’s clear that that’s going to change.
R: And I saw that Janelle has named a beer I think using her neural network it’s called The Fine Stranger which is a cool name for an indie beer. Have you had any success using it yet for some of your naming projects?
A: I’ll say this: that neural networks have, in my use of them, have illustrated to me some really interesting words and ideas, and clients are interested in AI and neural networks as part of the creative process. But there haven’t been any names yet that a neural network I’ve trained has generated and the client said, “Yes, that’s going to be our name.” But it’s only a matter of time before that happens. But I’m bullish on AI and neural networks.
R: Well, it’s funny because, I know this isn’t the same thing, but every now and then, I’m sure you see this too, there are these doomsday proclamations of naming…the human aspect of naming dying out because computers will be able to do it themselves. What are your thoughts in terms of how people and computers will interact in the future to do this job?
A: Oh, without a doubt, accessible AI tools for name generation will increase everyone’s access to interesting names. But just because you are shown a word or a list of words doesn’t mean that you’re going to know, as someone in the company for instance, is this really going to be the right word? Does this have the potential to become a brand? And there’s other aspects of naming such as understanding and ascertaining what the right naming strategy should be. What should the right inputs that an AI should be trained on? You know, what kinds of words should the AI be trained on? Helping a client see how each word in a list of words could become their future could become their brand, and helping them to see the the assets and potential of each of these names. That’s not something AI is going to do. So there’s still a place for professional name developers.
R: I want to back up a little bit and just talk more generally about about name generation. Can you just give me a 30,000-foot view of the entire naming process before we dive into some of the specific steps within it?
A: Yeah, sure, I’ll be happy to Rob. So, I’ll be briefed by the clients, and maybe they’ll provide me with an actual creative brief, or not, but from that, I’ll develop name objectives that succinctly capture what the name needs to accomplish; what it needs to support or connote. And once we agree on those marching orders, I’ll get into creative. Now the first wave of creative is a mile wide and an inch deep, where I explore many different perspectives of the brand, different tonalities, different styles of names, different executions. And that process takes about two weeks of creative development. At the end, there’s probably a thousand or several thousand words that have been developed. I’ll cull the best 150 names and run those through preliminary global trademark screening with my trademark partner, Steve Price. And from that, there’ll be 50 to 70 names, and I’ll present those names to the client. And I present them in a real-world context so they look less like hypothetical candidates and more like de facto, existing brands. And I present each name in the exact same visual context to really keep the focus on the name and not confound variables by changing up the color or the font. I present each name individually, talk about their implications and what they bring. And at the end the client gets feedback—what they like, what they don’t like, what they’re neutral about—and that informs the second round of creative work, which is an inch wide and a mile deep, where I delve into what was really working for them. And, it’s important to have a couple of rounds of creative because it’s one thing to agree in an abstract brief, but what clients really react to are real words, and that’s where you can really find out what’s going on, because it’s difficult for people to really understand what they like and don’t like in a name until they see them. And so that second round of work focuses on what’s working for them. And that process again is about two weeks, thousands of names developed, 100, 150 go into screening for trademark and domains, and then 50 names plus are presented to the client. And the client chooses from all of the names that’ve been presented across both rounds—typically over 100 names. They bring a handful of names into their full legal screening. Maybe there is cultural and linguistic checks that have to happen, and their full legal checks and then they choose one final name to run with.
R: What steps do you take when you just start generating names?
A: All right, so once we all agree on what the marching orders are. The process looks like this: I’ll first bring up my go-to set of software and applications and resources that I use pretty much in parallel, and I bounce between them as I go through development. So, I’ll bring up I’ll bring up Wordnik, which is an important piece of software online, a great resource for lists of words. I use OneLook, Rhymezone, an engine called Sketch Engine, and various other applications. And I will use those to identify words, word parts, that are interesting to me. And so over the course of that development I will use different techniques in order to unearth every possible idea I can find. I will also go through prior projects that I’ve done through Operative Words, and if I find a good word for this project, I’ll search on my computer for all files that I’ve worked on that also contain that word, and so I’ll be able to mine from my prior work. And so, that creative process happens for about two weeks. At the end of two weeks I will have amassed thousands of ideas, and if I bring neural networks and software-based combinations and permutations there are literally tens-of-thousands of ideas in the picture.
R: You mentioned Sketch Engine awhile ago as one of the online resources that you use. I’ve seen that you’ve written quite a bit about it and how you use it. But can you just briefly explain what it is and why you recommend it so highly?
A: Yes, Sketch Engine is a corpus linguistics database. So, let me explain that. Corpus linguistics is using a very large body of real-world language. That’s a corpus, and it’s plural is corpora. And using computers to sort of analyze and tag and organize what’s in there. So a corpus might be, for instance, the one I use is all of the news articles that have been published between 2014 and 2017. All of that real-world text—that’s 28 billion words—all of which have been tagged by part of speech, and it’s recorded all of the words that live next to all of the other words. In other words, it records what are called “colocations.” Now, colocations are useful because you can learn a lot about a word by the company it keeps. So if there’s an attribute that a client is interested in, let’s say ‘fast’ or ‘smart,’ I can look up a word like “fast” or “smart” or any other related word, and discover all of the words that have been modified by it. So, therefore I can find an exhaustive list of things that are fast, things that are smart, or verbs related to things that are fast and things that are smart. And so, the benefit is, one, is exhaustiveness, two, is also linguistic naturalness. That is, you’re finding how words are used in a real-world context, and I believe that linguistic naturalness in names is very important for names being credible, for names being relatable, and for names feeling very adaptable. You’re not foisting ideas on people that make no sense.
R: It rolls off the tongue, to use kind of the layman’s term.
A: Yes, that’s right.
R: You’ve mentioned so many online tools, I’m just curious, is there anything offline that you frequent?
A: I’m typically watching some kind of movie or TV show or some other sort of visual stimulus while I’m doing my creative development.
A: And those things provide visual stimulation and there is dialogue and other ideas that come up that provide an extra input to my creative process.
R: Do you choose what you’re watching based on the project, or is it just whatever you happen to be watching anyway?
A: No, no, I do. Absolutely. So, with projects that are very technologically driven or scientifically driven, I’ll watch something that’s sort of technological or scientific.
R: That’s fun. Do you ever just, you know, there’s been a movie that you’ve been wanting to see anyway, and you feel like, “Oh, that fits this project,” and you put that on?
A: Yeah, absolutely.
R: Another technique that I saw that you wrote about, it’s called an “excursion.” Can you can explain what that is? Is that related to the idea of watching a movie while you’re doing naming?
A: In an excursion, you identify a completely unrelated product category. Sometimes the less related the better. And you look for examples of a desired attribute or quality from that category. For instance, if you’re naming a new intelligent form of AI, let’s go ahead and consider examples of intelligence from the world of kitchens. Let’s look for ideas of intelligence in the world of sports. By thinking through an attribute as it appears somewhere else, you are able to find ideas that are differentiated but relevant, because when you take a word from a different category and drop it into a relevant category, it immediately becomes relevant to that new category. People are very comfortable with this technique.
R: I have a couple of tactical, logistics questions that I’m curious how you would respond to. What about the actual medium that you use when you’re writing down or documenting your name ideas? Do you do this in Excel or do you have a pad of paper with you while you’re doing all these other exercises, and you’re just furiously jotting down ideas?
A: I’m using Microsoft Word, by and large, for this. I also use another text application called TextWrangler. I use Excel when I’m charged with developing a generic descriptor for a new product.
R: And what is TextWrangler? Is there an important difference between that and Word, or just, you happen to use both?
A: TextWrangler is a text editor. So, there’s no formatting whatsoever. It’s plain ASCII text. It has another sister application called BBEdit, and these applications are very useful when you’re working with pure text, and it has some terrific tools like the ability to eliminate duplicates, the ability to use pattern recognition, something called Grep, in order to find words that include certain patterns. So, very useful tool and an adjunct to the toolset that I use.
R: And then the other logistical question is just about timing. You mentioned usually a two-week period of time for your first run at name generation, but I’ve heard other namers say they like to have a four-hour window to really immerse themselves in a project anytime they sit down to do name generation. Do you have any rules of thumb that you adhere to in terms of timing?
A: Over the course of two weeks, the process is, I will immerse myself completely in a project maybe for four hours, maybe for a day, maybe for two days, or three days even. And then I put it away. And then I forget about it, and I work on something else for a day or two, and then I come back to it. And so, I have this repeated process of immersion and then incubation and I repeat that in order to do creative work. That’s a process that’s been demonstrated and proven to help maximize creative output. Those “aha” moments—those Eureka moments you have in the shower—happen because you’ve been thinking about something and then stop thinking about it, consciously anyway. But meanwhile there’s something bubbling up under the surface that comes out when you least expect it.
R: You’ve mentioned a lot of things that you could use if you get stuck on a project. Do you ever get writer’s block so to speak, and if so, is there anything that you haven’t already mentioned that you would use to kick yourself back into naming gear?
A: Sure. You know, the writer’s block happens when a client is looking for something that isn’t different. If their if their product or their brand doesn’t really have something new to offer, that’s a more difficult nut to crack. And so, in those cases, I will look at projects that are utterly unrelated in any way, or other kinds of lists. And in this way, I expose myself to words that have nothing to do with the project whatsoever. But, because of how I see words and how I think, I can look at a list and look at a word and go, “Oh, wait a minute. There’s a story there.” I can see what would be related or that would be interesting. So, really, it’s a process of compelling me to look at words just in order to see what happens. It’s a little bit stochastic. It’s a little bit random, but it’s actually very useful and interesting and new ideas can come out of it, even for projects where there isn’t something wildly different under the surface.
R: I like to ask whether there are any names or naming tropes that you see that you’re getting sick of. You know, like any other creative process, there are trends in the industry—startups ending with with “-ly,” for example. Are there any specific name ideas or trends like that that you want to call out or that you wish would discontinue?
A: Well, Rob, there’re always trends that I wish would go away. In fact, any trends, by and large, I wish would go away, because they’re unoriginal and they don’t serve the brands that they represent. They look derivative. They look unoriginal. And what does that say about their company or their products? So, yes, I’m not crazy about the “-ly” trend that’s been going on, just as I wasn’t crazy about the “oo” trend that was happening after Google and Yahoo found success, just like I wasn’t crazy about the “i-” or “e-” prefix trend back when that was happening. You know, I’m just fundamentally opposed to these ideas because they don’t they don’t serve their clients and they, I think, reflect a company that isn’t truly original. I’m also not crazy about the trend to randomly drop consonants or vowels, or double them, because it’s clear that it was done just in order to secure a dotcom domain, and it feels like domain desperation.
R: Right, it feels forced.
A: Exactly. And linguistic unnaturalness, where you do these things in order to shoehorn words in order to get a free dotcom, I don’t think serves a brand well either, because they’re immediately off-putting, they look unnatural, and they’re difficult to relate to.
R: The last question I like to ask namers is just what your favorite thing is about being a namer or coming up with name ideas.
A: Well, I really love the process of identifying, exhaustively, every possible perspective of a new brand. If I’m looking at a list of a thousand potential names, those are a thousand different perspectives, a thousand different ways of framing you looking at this company. And those are a thousand potential futures. And then seeing when a company finally adopts a name that I’ve helped them with—to see how they adopted the name, breathe life into it, and then run with it, and do their own, get their own inspiration from the name. So, as an example, a while ago I worked with an architectural and design firm called Pollack Architecture, who needed a new name. And eventually, I worked with them and developed the name “Rapt Studio” for them, R-A-P-T, “Rapt Studio” for them. And they do brilliant interior and architecture work and branding work as well. Really brilliant and wonderful people. And so once I gave them “Rapt Studio,” they ran with it and they called their employees “Raptors.” I didn’t give them that idea. They have meetings once a week, which are called “Monday Rapture” meetings. All right. So, I love when a name can inspire a client with great ideas. That makes me very happy.
R: That’s great. Well let’s leave it there. And I just want to say thank you again for your willingness to share some of your thinking and how you do what you do.
A: Well, thank you so much, Rob. You know, I really do this for selfish reasons because I hate ugly words, and names are an unavoidable part of our environment and our habitat, and wouldn’t you much rather be surrounded by beauty and gardens than blight? I feel that way about names and so I give away what I know, because I want other namers, even my direct competitors, to come up with with great names so that they can also populate the world with words that are interesting and creative imaginative, and words we like to have around.
R: Well, you call it selfish but it seems selfless to me. I really appreciate it and thanks again. Let’s go make some more beautiful words out there.
A: Yeah, let’s do that. Thanks, Rob.
R: Thank you.