Learning machine learning for SEO w/ Britney Muller
Birtney Muller, Senior SEO Scientist at Moz, talks about Machine Learning, learning in general, elephants, and crazy chess stories!
Britney Muller on Machine Learning, Chess, and Elephants
Timestamps
0:00 Intro
0:54 2 Chess stories and why it’s not like Poker
9:00 What chess and Machine Learning have in common
11:28 How Britney fell in love with machine learning
16:06 Applying machine learning to SEO
18:47 How bad is TF-IDF really?
21:09 Google’s NLP API
25:00 How we can save elephants with machine learning
31:43 How machine learning will take Google to the next level
43:44 What is Moz planning to do with ML?
45:30 The Snapchat thief
50:55 The importance of learning
Shownotes
- https://twitter.com/BritneyMuller
- moz.com
- chess.com
- www.bitly.com/help-the-elephants
- Book: Annie Duke - Thinking in Bets
- AlphaGo documentary: https://www.netflix.com/title/80190844
- Andrew NGs course: https://www.coursera.org/courses?query=andrew%20ng
- The Snapchat Thief: https://gimletmedia.com/shows/reply-all/v4he6k
Related posts
Transcript
Kevin: 00:00 Britney, welcome to the show. Show.
Britney: 00:03 Thank you.
Kevin: 00:04 It's so good to have you. Thank you for making the time. For all of those who don't know you and if people don't know you, then they're fools. But for those of you, of the people out there who don't know you, who are you?
Britney: 00:19 Who am I? I mean, I don't know. I guess I am Britney Muller. I am Moz's senior SEO scientist. And before I was at Moz, I started a boutique medical marketing agency in Denver, Colorado, and had so much fun building that out over the course of like around five years. Before that, it was just kind of getting into the weeds with the SEO stuff and becoming addicted to this digital game that we play every day. Trying to get things ranked on search. It's so much fun.
Kevin: 00:55 Yeah, and then you became a rockstar, which is awesome. But that's an awesome Segway with the digital game that we play, because you told me that you have a chess story. What was that about?
Britney: 01:09 This is actually so embarrassing. So I have loved chess for a very long time. I think it's the most beautiful game, it relaxes me. I play all the time on Chess.com. But I have sort of hit this classic plateau, and in order to get better, I realized I would love to have a teacher or an instructor once a week, maybe once every other week. And so I found this company that does just that and I had my first chess lesson yesterday. Kevin, you're going to die. He is so sweet. I don't think he's old enough to drink.
Britney: 01:48 We were video chatting and he's in his college dorm room and there's four Frisbee's on the wall. I was just like, "Oh my god. What have done?" It was so sweet. It was really fun.
Kevin: 02:01 That's so cute.
Britney: 02:02 Yeah, so that's my new chess instructor. So he's showing me the ropes. It's really cool.
Kevin: 02:07 Is he like a brainiac, like is he totally killing it?
Britney: 02:11 Yeah, the way he's already starting teaching me about shifting the way that I think about piece play into positional play and seeing the board better. I mean it's already so awesome. I love it.
Kevin: 02:23 How does that make you feel, honestly?
Britney: 02:25 I know. It makes me excited. I need stuff like that.
Kevin: 02:29 Yeah.
Britney: 02:29 To continue to sort of learn and dabble in. Otherwise I just get bored. I'm sort of an all or nothing person. If I'm going to master something, I really want to master it. So that's kind of a curse and a blessing at the same time, I guess.
Kevin: 02:49 Oh, I totally understand you. I can so relate to that. But yes, I also love learning and I also believe that if you want to be successful in SEO, you kind of have to have that mentality. Because is changing so much and there's so much going on. But I also love chess and I'm definitely not good. So you could easily beat me, I think.
Britney: 03:12 Well, you should totally play. Do you have a Chess.com account?
Kevin: 03:16 I have one for another site. What's it called? Chess-IO or something like that. But I'm totally happy to set one up on Chess.com.
Britney: 03:24 You should. It's free, it's super fun. You totally should check it out.
Kevin: 03:30 I'm down. Yeah, and I love … So I learned there's a deeper philosophy behind the game of chess.
Britney: 03:34 Yeah.
Kevin: 03:35 In comparison to poker as well. Yeah, it's-
Britney: 03:39 Yeah.
Kevin: 03:40 It's that mind-blowing. You know, it's similar to the difference between math and statistics and how those are two different approaches to see the world.
Britney: 03:48 Totally.
Kevin: 03:49 With chess and poker, I learned this by the way from a book that's called Thinking in Bets. I'm blanking on the author, but I'm reading it right now. Her name is Annie Duke. Right, Annie Duke. It's an amazing book. She's this poker player, she has a crazy story about accidentally discovering poker or having to play it to make a living and then becoming this world star. She kind of writes about all the principles that she learns about poker.
Kevin: 04:17 Long story short, she talks about how chess is basically more like math. Right? Because you have a limited amount of outcomes and you could technically calculate the quality of every position or of every move. Like you have a defined, limited number of outcomes because you know all the options is basically where this is going. And poker is the opposite. Poker is probabilistic, it's more statistics. Because there's a chance that somebody has certain cards, but you don't know because you can't see their cards.
Britney: 04:50 Right.
Kevin: 04:50 I freaking love chess to train that kind of mental skill of learning how to move parts within predefined limits or constraints. But then I also want to play more poker because that is a fascinating view of the world as well.
Britney: 05:04 Wow. That's really interesting.
Kevin: 05:08 Yeah. Yeah, it's a great book idea. I wish I came up with that.
Britney: 05:10 Yeah, yeah, yeah.
Kevin: 05:12 So how often do you play chess? So I mean, like daily?
Britney: 05:15 So I went through a phase where I was just obsessed. I mean it becomes a problem, honestly. So I enrolled in the United States Chess Association. I had my card that, I remember dropping it a bar in Denver and this guy picked it up for me and was like, "Here you go. Like who are you?"
Britney: 05:34 I would attend tournaments but I wouldn't play in them. I got really serious about learning end-games, openings, reading the Bobby Fisher Teaches Chess, all of that stuff. I just became obsessed, and then I kind of didn't have anyone to really play with. And so I sort of, yeah, got burnt out on that for a while, but want to get back into it.
Kevin: 05:58 Oh, for sure.
Britney: 05:59 Actually, there's a machine learning old-school program here at the … It's called The Museum of Computers or something in Seattle. It's one of the original ones that was built to play a person in chess. There's this book of photos, and if you beat the computer, you get your picture in it. And so I was sitting up there, this was a couple months ago, and I was … Sorry the lights go off here.
Kevin: 06:25 No worries, it's all good.
Britney: 06:27 It happens every time. Like, I know this is weird.
Kevin: 06:29 No problem.
Britney: 06:30 They might come back on. I can brighten this too.
Kevin: 06:33 It's no problem at all.
Britney: 06:35 Okay. I was up there playing and the employees of the museum, they came up to me and they were like, "Do you play?" I was explaining that I play a little bit and I really want to get better. They go, "You know, we have young girls come in here all the time and they look at this, and they look at this album and they are always asking, why isn't there a girl in here? Why isn't there a girl?" And my heart, seriously just like exploded.
Britney: 07:03 I was like I have to come back. I will be back. I'm going be in there. I'm going to bring other women with me. Screw that. So that's kind of what motivated this new inception of chess 2.0 for me.
Kevin: 07:15 Oh, that is such a cool background story, and it is kind of sad that there is so few women and girls in that book. But what a cool idea. Man, that's awesome, that's worth so much. You're actually setting me up for success here, because I love how you draw that bridge to machine learning because that would have been my next question. How is chess and machine learning related?
Britney: 07:41 Oh, that's a great question. So they say that there are more possible games of chess than there are atoms in the universe. It's incredible how many games of chess could potentially be played with a number of moves. I think in that same head space, machine learning, you could create any sort of algorithm, a million different ways to solve the same thing. Right? To get the same end result.
Britney: 08:09 The difference being, A, the data scientist behind a machine learning model and the person playing obviously. But it's also kind of like the training data. So a training data, as a person playing chess, is my previous experience. It's my knowledge of the game. And a machine learning model, it's the literally training data that's going into this. So I think that's where they're sort of similar and could potentially look a little bit alike. Yeah.
Kevin: 08:39 Totally. Totally. I really love the documentary around AlphaGo. It was so good.
Britney: 08:41 Good.
Kevin: 08:42 It is, oh my god, why was it only two hours?
Britney: 08:49 It was so good.
Kevin: 08:49 Honestly, I could have … There's so much fascinating stuff around this. Anyway, the next question I wanted to ask is how does it … Like you're absolutely passionate about machine learning and about chess as well. So what is it … There has to be some sort of a mental connection for you. I'm curious about how both of these things fascinate you so much? Is that because of the grand scale of possible outcomes and training data or is there something else?
Britney: 09:22 I think I have just always been super competitive, almost to a fault. I hate bowling because I'm so bad at it and I know there's a very low probability of me winning. And so I choose not to play. I freak out about the gross shoes and you have to put your fingers in these holes in the balls and it's gross. Really, I'm just a coward and I don't what to lose and I want to win at everything. So I think … I don't know where that really came from. Maybe it's being the oldest sibling and having a younger brother and we were just constantly in competition with each other.
Britney: 10:00 Yeah, I don't know. Competition motivates me so, so much and I've always been just ridiculously curious to a fault. Where I'm doing stupid stuff or asking the really weird questions because I'm just … I don't know why my head went there. You know?
Kevin: 10:17 I can so relate to that.
Britney: 10:18 Yeah. I think those two things sort of make a really solid perspective for anything like search engine optimization, machine learning, chess, you name it.
Kevin: 10:33 Yeah, and I think, so first of all, yes. I'm the exact same way. That's kind of the second important trait as an SEO, you have to have that sort of kind of competitiveness. And there are many different ways to be competitive. Some are just competitive with themselves, which drives them to learn, and then others, they just want to measure up against others.
Kevin: 10:53 And then I think you and I were more the people that if we don't see a way to win, we're just not going to play. If that makes sense. I'm the exact same way. Because I know people like you can challenge them on anything and they will chime in because they just want to beat you. Right?
Britney: 11:07 Yeah.
Kevin: 11:08 But if you challenge me, and I have never done this before, I don't care. Right? But even something where you see a chance, ah, then I care very much and then I just want to win.
Britney: 11:17 Totally. Totally.
Kevin: 11:18 So I can so relate to that. I mean, you are so deep in the topic of machine learning. What was one of the things that surprised you along what you learned so far?
Britney: 11:29 Oh my god, there's been so, I mean, every time I re- kind of get into it, something always surprises me. I think the initial … So I think I had found out, like people were talking about machine learning and how becoming a data scientist was going to be the sexiest new position in the future and the most powerful and badass. And I was like what is that? I want to learn about that. I want to be that.
Britney: 11:56 And so I remember going to GitHub because I thought that's where all the technical things were. And I did a search for data science, and I did tons of digging within repositories and different accounts. I discovered that Harvard CS109 Data Science Course was on GitHub. I don't think it was supposed to public. They would release lectures every Tuesday and Thursday, this was like six years ago. And so I followed along, and it just blew my mind and it gave me that new kind of excitement that SEO gave me like six years before that.
Britney: 12:39 You know, where I felt like I was mastering it, what else is there to learn? Like there's experiments I could do, but beyond that I wasn't able to feed on anything else. And so this gave me that fuel of, holy shit, there's this whole other world out there that is just taking this stuff to the next level. And so I just became obsessed with it. Obsessed. I mean I would do the homework assignments, I didn't have a Harvard email address, but the TA's would answer my questions when I would send them questions from my Gmail.
Kevin: 13:09 Nice.
Britney: 13:10 Yeah, and so I felt like I was a part of this thing and it was so much fun. I just obsessed over it. I couldn't wait for those classes to come out.
Kevin: 13:20 Wow, my heart is shining, because I love to learn. I love all of that stuff. But what I'm wondering about is how do you deal with all the math and statistics?
Britney: 13:28 Oh, it's so hard. I'm not naturally gifted in math or statistics. That's why I think the Harvard course was really, really good for me. Because it really broke it down in layman's terms, but it also provided insane, easy to understand examples. That unfortunately I don't feel like I got from things like Andrew Ning's Course, which is an amazing course and you should definitely check it out. But the intro that Harvard and I'm sure other universities and places have, it just was such a great, easy introduction into the theories of it. Into base theorem and where that came from and what that means. Here's an example. It just really allows you to sort of wrap your head around it.
Kevin: 14:19 I love that. Yeah, I'll certainly link to all of these courses in the show notes. Because I think with math and statistics, it's such a fascinating topic. And at the same time, I don't want to give away the responsibility, but [inaudible 00:14:36] like the school system in Germany kind or really failed me on that one. Because it's actually fascinating but I hated it in school.
Britney: 14:42 Yeah.
Kevin: 14:43 I kind of got better in university and college. But then in college, we had like 15 hours where we went from the absolute numbers to exponential functions in math. Right? So there's noway to really relate to that, just pound your head and try to survive on the test. But that's when I got curious about it. So I will check that course out, it sounds amazing.
Britney: 15:02 You definitely should. And I would say that you don't need to know as much of the probability, statistics and math behind it as you would think. So the platforms that you build upon, like Tensor Flow and Caras and Anaconda and all these things that you build upon, they have all of that baked into them. So it's really a matter of having just a basic understanding and being able to evaluate things like the loss function. And to see how well it's doing. But you don't need to be proficient in the equations in and of themselves at all. At all.
Kevin: 15:38 Gotcha. Gotcha.
Britney: 15:40 I would argue that most people doing machine learning, they couldn't write out the function if you asked them to. Like it is so confusing. And then if you did, they wouldn't be able to do it without programs.
Kevin: 15:53 Sure. Sure.
Britney: 15:53 Yeah.
Kevin: 15:53 I mean you don't need to understand on Canva works to take a picture.
Britney: 15:56 Right.
Kevin: 15:57 But that's sounds amazing. And so how do you apply … You're a senior data scientist at Moz right? How do you apply machine learning to SEO in your work?
Britney: 16:08 Yeah. So lots of different ways, and most of the time it fails to be honest. But I mean I've tried lots and lots of ways, starting with automatically generating meta descriptions. That was super fun and successful, and you can plug that into a really large website to get you most of the way there. And then it can be everything from recommendation systems, kind of sending off flags of potential fraud, we've looked into that.
Britney: 16:47 Most recently I have been exploring various types of machine learning to solve for categorizing keywords intent. We've been doing that by looking at the SERPs themselves. Because quite frankly, we all know Google houses the world's information. They know what someone is looking for when they do a particular search. And so our idea was okay, if we can evaluate the SERPs that we see, could we categorize and be able to paint a picture for our customers of what that acquisition funnel looks like? Right? Where are there opportunities and whatnot? So that has been my most recent baby and I'm so, so proud of it. It's been really fun.
Kevin: 17:32 That is amazing. I love that. Yeah, I kind of feel like that sounds like a smart idea. So as senior data scientist at Moz, the way I understand it and correct me if I'm wrong, is that you do a lot of research to inform the creation of new tools and features. But you're also very out there at conferences and speaking about it. So does that mean that if you see an opportunity for applying machine learning to identify user intent at scale from the SERPs or SERP features, so that then becomes a Moz feature? Is that how I can understand it?
Britney: 18:07 Exactly. Exactly.
Kevin: 18:09 Very cool.
Britney: 18:10 Yeah, and I take the time either when I'm at conferences or even just online with our Twitter community to sort of keep a pulse on what people are interested in the current space of SEO. But also what problems could we potentially solve for people? And so that's what I try to take back to the product.
Kevin: 18:30 I love that job. That sounds amazing. That's all learning and all … You must be like a product evangelist and you're a product kind of … You do a lot of research or kind of what do you call it? Like a customer market research and that kind of stuff. So ah, that's amazing. And so you also recently tweeted about TFIDF and how-
Britney: 18:53 Yeah.
Kevin: 18:53 Like I'm on your side. I know, I know, I know.
Britney: 18:56 Yeah Kevin. Good boy.
Kevin: 19:02 No actually, see it's always easy to bash some of these concepts in SEO. But I think it's a bit of a quick shot to just say, "Oh, well this does not apply anymore, it doesn't make sense." Like I love how you basically said in that tweet that, "If we understand the principle behind it, that we can just learn from it, of what it might be today." And I whole-heartedly agree to that.
Kevin: 19:33 So Jon Bueller recently also tweeted that SEO's should thinking about just trying to build a search engine themselves to understand the principles better.
Britney: 19:40 Yeah.
Kevin: 19:41 I also think that's a smart idea, but just looking at some of these older concepts like TFIDF, absolutely smart. So can you explain TFIDF and how you think it's still valuable, at least for SEO's to know nowadays?
Britney: 19:57 Yeah. I think it's incredibly valuable. Like I said in that tweet, it is honestly one of very first things you will learn about when you begin exploring machine learning, data science, any sort of natural language processing or text understanding. Like you have to know the basic principle of TFIDF.
Britney: 20:17 So what it is in a nutshell is, it's Term Frequency Inverse Document Frequency. And so it's this pretty basic formula that evaluates a particular terms frequency within one particular document compared to the corpus. For example, the word 'the' would have lots of instances, sort of signaling that it's less important or less unique to the particular document in question.
Britney: 20:48 And so what it allows you to do is it allows you to pull the more unique topics, the more unique words and information from a particular document. And it allows you to do large scale text analysis, and it's incredibly important to play around with and just to have an awareness because you can use it for so many things in SEO.
Britney: 21:10 I think it's good to just be able to talk, be able to speak about it. It parlays into all these other really beautiful things in terms of natural language processing, text generation, you name it. So yeah, I'm a big fan.
Kevin: 21:29 Yeah, I totally agree. And to be fair, whenever I've used TFIDF tools in the past, I saw a good outcome with optimized content. I say that. I do acknowledge that Google probably doesn't use it anymore or maybe uses separation of like a progression of it. But there is some underlining principle that some of these tools hit on the head. So I wouldn't … Again, I wouldn't take some an absolute stance on some of these things. Especially when you see results. Right? Like you always want to verify your assumptions in those regards. I think it kind of also is a beginning of an understanding of entities and how they might relate to each other, which is the hot shit at the moment in SEO.
Britney: 22:12 Yeah.
Kevin: 22:14 Have you played around a little bit with Google's National Language Processing API by chance?
Britney: 22:19 I love it. I think it's so fascinating and more people need to know about it. Yeah, I'm a huge fan. I'm a really big fan of the categorization and that they give you a confidence score. So my biggest thing is okay, if you're trying really hard to rank for this term, and you're sort of neck in neck with another website. Put your content into their natural language API and then put the other content in and compare how Google's categorizing you and your competitor.
Britney: 22:47 Often times I see that the person trying to rank higher sees that their categories are diluted across maybe like three or four. Where the competitor has one solid confidence score for a very specific industry. So I think yeah, it's an incredibly powerful tool and not enough people know about it.
Kevin: 23:08 Yeah. I totally agree. I really like it as well. I think it can be, it's probably not yet ready to just sort of tell you what's going on. I think you still need to apply a lot context and critical thinking to what the data tells you, especially like what means "Salience" and how do the different types of entities compare. Because, sometimes you'll see the same entities popping up in different categories. But still like just looking at that and getting an understanding of how you can think about entities as a grand scheme, I think is so very helpful.
Kevin: 23:45 I think it's a good explanation for lots of things. My personal theory is that, that's why Author Rank or rather Author kind of went away, because they didn't need it anymore. Because now they know that people are entities and they can understand the relationship. So why would they … So if you plug your stuff into that API, you can find a lot of these things.
Britney: 24:07 Totally. That's such a good point. Yeah.
Kevin: 24:09 Yeah, that's a conspiracy theory.
Britney: 24:11 I love it.
Kevin: 24:11 You've got to be careful with that.
Britney: 24:15 It's dangerous to say that one particular algorithm or thing of a space isn't applicable anymore. I mean we live in this tiny, tiny bubble. There is a huge, huge world out there using all of these incredible tools, algorithms, technology that we haven't even begun to explore.
Kevin: 24:39 Right.
Britney: 24:39 So I think it's just dangerous to put labels on different things. I mean, I try to second guess myself as well as frequently as possible about this stuff because you have to be testing, you have to be exploring. There's just so much more out there then we make assumptions about.
Kevin: 24:57 Oh, preach. 100%.
Britney: 24:58 Yeah.
Kevin: 24:59 Has the journey of machine learning taken you also into areas outside of SEO? Was there something that you found along the way that you just got stuck in and went down the rabbit hole and now you're also passionate about? I don't know, machine learning in logistics? I don't know. Probably not.
Britney: 25:17 Yeah. I mean not logistics. Yeah. But the machine learning world, I think that's exactly what it does. Right? Like I want to hear what you would fall into if you started getting deeper into the world of machine learning? Because I think for every individual person, what you tend to do is you tend to think, well what do I have domain knowledge in? What background or just kind of fundamental awareness could I leverage to solve for in a particular space?
Britney: 25:46 So once you start to learn the foundations of machine learning, you start to understand what's possible and what's not possible. That's where your interests and your passions come into play. Something that I've been thinking about for years is applying machine learning to illegal poaching of animals. Elephants in particular, because they are the best and I love them so much.
Britney: 26:14 I just think there's so much potential in this technology to do really beautiful things. There was a relatively secretive project, which I'm pretty confident they're speaking about now at Google. Where they would take old cell phones and put them in these weatherproof boxes, way up high in the rainforest to detect illegal deforestation, using machine learning to identify the sounds. Right? And so taking that same idea and that same principle, I thought, well why isn't anyone doing this for poaching or for animals or crime for that matter.
Britney: 26:54 You know, it could apply to so many things. It was funny because I became obsessed with the project that Intellectual Ventures were working on, and they're just across the water in Bellevue here. I reached out to Pablos Holman, who works on all these really incredible projects. He informed me that Paul Allen, before he passed, started getting really into elephants. And he developed this same theory and he started creating drones that were powered from the ground, so they could stay up there for a super long time. And the drones I believe they're using high definition cameras to identify poachers.
Kevin: 27:36 God. Wow.
Britney: 27:36 His sister is taking over that project. So there's these incredible things that are going on in our world that are as an SEO space. Like we have blinders, and I would encourage everyone to explore this stuff. It is so exciting and you can apply different aspects of all of these spaces to what we do every day. You know?
Kevin: 27:56 That is so cool.
Britney: 27:59 It's so fun. It's so fun.
Kevin: 28:00 That is such a cool project. I just love the way that it's thought through. That is amazing. What got you so passionate about elephants? Where does that come from? It's clear that it's your spirit animal, but what's the story?
Britney: 28:12 I know. I honestly, I have no idea. I've always just sort of loved them. I know they're super smart and they just seem so wise and adorable and human-like. I don't know, I just want to live among them so bad, it's crazy.
Kevin: 28:29 Have you seen that movie Dumbo?
Britney: 28:31 Yeah.
Kevin: 28:32 Was it according to your expectations?
Britney: 28:35 I haven't seen … I've saw the old Dumbo, like ages ago. Is there a new one?
Kevin: 28:39 Oh. I think there's a new one. It just recently came out with a pretty good celebrity line up. I think it might be worth checking out. And then in Germany, where I grew up, there is this kid's show called [inaudible 00:28:54] It means Benjamin Flowers or something like that. I don't think they have this in the U.S., but it's this big elephant that goes on adventures. So I was wondering is there something in the U.S. like that, that you got hooked on as a kid?
Britney: 29:10 I think I always just thought they were cool, and then I started reading lots of books about them. And so they have this sixth sense. There's a book called The Elephant Whisperer and the way that they interact with humans and bond with humans and know when people have died that they know. I mean they have an awareness on a whole different level than humans. It is fascinating. You should totally check that book out.
Kevin: 29:34 I should. It sounds like a super-
Britney: 29:37 I mean I cried the whole time. They're so in tune with humans. They knew that this very close human who took care of them for a decade had passed and they migrated across this huge sanctuary. It had to have taken them like a day or two. But they got to the house of his and they stood out there for 48 hours, sort of mourning him.
Kevin: 29:58 Wow.
Britney: 29:59 I know.
Kevin: 30:00 Damn. Elephants, so underestimated.
Britney: 30:05 I know.
Kevin: 30:06 The better humans.
Britney: 30:06 I know. They're the best. We don't deserve them.
Kevin: 30:09 Yeah, definitely. We just don't deserve elephants. Wow. That is so fascinating. I realize I know nothing about elephants, legitimately nothing. I think my level of knowledge is from the Jungle Book. That's what I know about elephants. Jesus. Okay. That's yeah, that's a big one.
Kevin: 30:30 So now I'm trying to cut a Segway into another topic. I kind of don't want to leave there necessarily. But elephants are also, I think there's a link to a donation project in your Twitter account. Can you tell me what it's called real quick again, so I can include it in the show notes?
Britney: 30:54 Yeah. I know I made a Bitly link for it, didn't I? Let's see.
Kevin: 30:57 Yes. Yes. I think. Something elephants.
Britney: 31:00 Yeah. Bitly. It's just help-the-elephants.
Kevin: 31:06 Awesome. Yeah, I'll add that to the show notes so people can donate.
Britney: 31:10 Yeah, and that's the guy. That's the animal whisperer guy.
Kevin: 31:12 Oh, crazy.
Britney: 31:14 Yeah.
Kevin: 31:14 And so, okay so we talked about the missionary project for elephants or a couple of those. Excuse me. And yeah, let's hope they're successful. So how do you think machine learning will impact how Google's search will change over time? Like with the knowledge that you have, it's probably a very broad question, but I think you have a deep understanding of what's possible with machine learning. So what do you think people can expect in the near future, maybe the next three years?
Britney: 31:48 Do you want to get really crazy?
Kevin: 31:52 Please.
Britney: 31:52 Okay. I haven't talked about this to anyone really yet. Okay.
Kevin: 31:56 Okay, you heard it here first.
Britney: 31:57 Yeah, this is crazy, and I can't stop thinking about it. It's been on my mind for at least a week and a half now. And I feel like we could talk this out. So where I see some of the stuff going, A, obviously search will just continue to get better. They're getting a rich understanding of the real world, they're going to rely less on digital signals. But where I've started thinking a lot, bear with me, is the Google Vision API, have you played with that?
Kevin: 32:26 I have not.
Britney: 32:27 Oh my god. Dude, so you upload a picture. I uploaded this really old picture of me and like a roger that I had gotten from Moz Con, and the Moz logo is so small it's not even fully in the frame. And within the different sections of that API, there's a logo section, and it identified Moz.
Kevin: 32:52 Wow.
Britney: 32:53 It also identifies where else that photo has been, who it thinks is in the photo. What things are in the photo. It picks up emotion. And when you look at the source code of the JSAN it's looking at the corner of my eyes and my pupils.
Kevin: 33:11 Jesus Christ.
Britney: 33:13 Yeah, so they're measuring all of that within a photo.
Kevin: 33:17 Wow.
Britney: 33:18 Oh, and also take into consideration Google Photos has free, storage-wise, space-wise, for our photos ever since it came out. Correct? And now, they're just going to start charging I believe like with the new phone, the new Pixel. My theory is like, oh they've gotten enough data. They have enough of our data to evaluate some of these things. So they don't need to give it to us for free anymore. What has been on my mind so intently, is I had purchased and I have the new Google Home with the camera and the screen. It has gesture detection. So I will ask it to play something and I'll be in the kitchen and I can go like this, and it stops. And then I go like this, and it plays again.
Britney: 34:08 I'm standing there in front of the camera and I'm thinking, holy shit, I'm wearing like a LuLu Lemon zip up. They know that. I'm reaching in my cupboard for Khasi Cereal, they see that. They see that logo. I mean we will become the knowledge graph in real time. In our own, we will own our own knowledge graph of preferences, products. They know when I wake up. They know when I go to bed. Do you know what I mean?
Kevin: 34:40 God.
Britney: 34:40 When I started thinking about it, and I've had a really hard time adopting the product since then. Because I'm just constantly aware of the things they could be recording and using.
Kevin: 34:56 Oh my god. This is so crazy.
Britney: 34:58 Kevin, Kevin, they can see what scares me. Like if I go, "Hey Google, good morning." And it plays the news for me, they can watch my eyes and detect, oh, that made her upset. Oh, that really made her excited. They can start to just tweak and understand you better than you probably understand yourself.
Kevin: 35:17 My god. This is insane. And it's so crazy that you mentioned that because one of the four tabs that I have open is for Google Lens, and they have that stuff in there as well. So first of all, they're forging a competitor to Pinterest. Where you can take a photo of something and it will tell you what items are in it, so you can shop for them. This is something Yeah, this is something where I think Google and Amazon will just fight to the death. Because, traditionally Google has been pretty bad at this bottom funnel type of stuff. Right?
Kevin: 35:49 They're good at setting traffic, but bad at the conversion. I think that might be a game-changer. Thanks to all the people filling out the captures or clicking on the pictures to identify stuff and trying Google Smalls. It is such a crazy idea that all we're doing is feeding Google. We're basically all hooked on drugs as SEOs. We're all hooked.
Britney: 36:11 Completely. Completely.
Kevin: 36:13 We can't get off of it. We can't. So just this Play, just the way that Google has set it up is insane. I totally get what you're saying. Right? Like we'll just keep training this thing. It will get feedback from us, it works at scale, so it'll learn really, really quickly. And the depth of what they can already do with the data is mind-blowing.
Britney: 36:33 It's wild. It is wild. I was thinking too, like they know what kind of wine I like to drink and how often probably, if they're looking at that camera. I mean, they can start to predict and surface all sorts of stuff for you very specifically. It's crazy.
Kevin: 36:52 The big thing about it also is the ecosystem that they created. Because they will know when you write about this in Gmail to another person and how you think about it. They can even finish the sentences for you. God, they'll probably write the emails for you pretty soon. Do you know what I mean? I cannot even be in there.
Britney: 37:07 Yeah.
Kevin: 37:07 So it's all going to happen automatically. And then with all the other avenues where they have products and integrations. I mean, look at Maps, they know where you are all the time. So when they add a vision to it, an actual eye to the system, then the brain is pretty much complete. Right? I mean, okay, I'm talking like a crazy person right now. I hope everybody understands that I'm not expecting Cybernet to launch tomorrow. But if you spin that thought of it further, it can get pretty crazy, pretty fast.
Britney: 37:39 No. No. Absolutely. So I'm just curious, where do you see SEO fitting into some of that in the future? Like with them having richer understanding of stuff?
Kevin: 37:49 Yeah, I don't think it fits in at all, to be honest. I have a really hard time to see an SEO play in that. And to bring it a bit closer to reality. Okay, like this is, like we're talking about what's possible.
Britney: 38:04 Mm-hmm (affirmative)
Kevin: 38:06 But I think what's real and how fast it's going to progress, I think that's going to be much, much slower. Right? And in much, much smaller increments and much more tangible.
Britney: 38:14 Yeah. Yeah.
Kevin: 38:15 So SEO's not going die tomorrow. It's probably obviously not going to die in the next couple of years.
Britney: 38:17 No.
Kevin: 38:18 But in that future that we're painting right now, I don't think SEO has a very strong place. I think there will be less waste to optimize that system because it's not an answer. It's not a search system, it's not a pull idea, it's a push idea. Do you know what I mean?
Britney: 38:34 Yeah.
Kevin: 38:34 It will tell us stuff and suggest. And that's what we see, I think already in the search results. It's less and less of search and it's more of predictive answers that Google is trying to solve. And so I think that Google is trying to get to a point where you don't need to search at all anymore. They're trying to get rid of search, and so with machine learning it's perfectly possible. Because you know as soon as a nervous system has a thought or a need, it will be detectable from some sort of a camera or from sort of an action or an indicator or behavior.
Kevin: 39:06 And as you said, it will probably know us better than we know ourselves. So why would we even search for something? And again, I'm going back and forth between some scenarios that are a bit more surrealistic and not given yet and some things that are more applicable nowadays. But I think emails are a great start. We already see Google finishing your sentences, which is one thing. Right? But I think very soon they'll be able to just take that data to the next level and just know how you're going to probably respond to that.
Kevin: 39:34 Again, I think there's so many different avenues that are now coming together. You know, like Gmail, Maps, and I'm probably forgetting five other things. There's Google Home, there's Google Search, there's all that kind of stuff. I think they're capitalizing on that because machine learning allows them to.
Britney: 39:47 Absolutely. Absolutely. I have a hard time visualizing search going away completely. I feel like there would still have to be aspects of you as an individual having really unique thoughts or questions or that would throw them off a bit. But I also feel like Google is as close as we probably are to a general purpose machine that has general understanding of things as we know it today. My fear too is that we will come up on another AI winter, and this stuff is going to kind of, not only not come to fruition, but be rolled back a bit until further funding supports it.
Kevin: 40:32 I actually hope so. I think it would be healthy for the natural progression because some of these things can develop so quickly, they are not comprehensible for humans. I think it can honestly be dangerous because I think humans need some time to adapt to technology. We already see, excuse me, how things like social networks have a certain impact. Or smart phones, we're just not, I mean this stuff is brand new. Right? I mean the first smart phone, it depends on how you want to define smart phone, but the iPhone came out about 10 years ago, a little more than that.
Kevin: 41:03 And so look at the trajectory and the journey of like the hype cycle and then people now are thinking, ugh, like maybe we should step back and be careful with that. I think machine learning is one of these things that could progress so quickly that we're not able to catch up from a mental perspective, cognitively. So it would be great to see another AI winter and get some regulations in place, that at least most countries in the world or the Western world agrees upon doing certain things. Like prioritizing human needs or humans in whatever decision is being made.
Kevin: 41:33 Again, I understand I'm venturing into some very dystopian ideas here, but I think it's at least a good conversation to be had without painting the apocalypse or predicting the apocalypse. But I just think we should all think about that, because we're at a point where we can still put things in control.
Britney: 41:56 For sure. My only concern about an AI winter is that, that won't necessarily be the case for these large companies that have tons of hardware and tons of software. and they can continue. But the everyday person can't to some extent.
Kevin: 42:12 Yeah.
Britney: 42:12 So I think that, that would be pretty dangerous in some aspects. But yeah, who knows? It will be so interesting to see where this stuff goes and shakes out.
Kevin: 42:23 Oh totally. Totally. I'm actually not that pessimistic as I might sound or come across. But how do you think SEO's should think about this whole development? What is something that people should start looking at, and obviously the courses that we mentioned, but how should SEO's start to think about the impact of machine learning in SEO?
Britney: 42:44 So I think first and foremost, we should be thinking about it in terms of this really powerful tool that can take over everyday boring tasks that we do. It will allow us to level up as an industry and evolve to focusing more of our time on higher level strategy, higher level thinking. And allowing machines to do some of the silly, traditional, low level SEO tasks that we know and do today.
Kevin: 43:19 I would love that. I would love to see that happening.
Britney: 43:22 Yeah, it'd be so nice.
Kevin: 43:24 Because there's still so many points, as you mentioned, where you just do something really dumb. Or maybe it's just me.
Britney: 43:32 Me too.
Kevin: 43:32 Okay, where you just do a lot of manual stuff. But yeah, I think there's a lot to be gained and I can't wait for more tools to provide some really kick-ass machine learning features. Is there anything that you can already say about Moz adding some of that? Besides maybe the user intent thing?
Britney: 43:50 Yeah, Moz is working hard to innovate in that space big-time. With our whole mission really being to just provide instant insights. So instead of telling you, give us your data, do this, do this, do this, do this. We want to be like, just give us the data or allow us to pull data, and we will just boom, provide you insights. So we're trying to shorten that path from collecting the data to getting the insights and making it way more accurate, way more efficient. And then making sure that it gets executed so that you improve.
Kevin: 44:27 For sure. Oh, I can't wait for that to happen.
Britney: 44:30 Yeah.
Kevin: 44:30 Is there anything else, outside of machine learning, that currently fascinates you, that really has your interest?
Britney: 44:35 Oh, I love this question. Danny Dover, do you know who Danny Dover is?
Kevin: 44:43 Oh yeah. Yeah, yeah.
Britney: 44:43 Legend. We were at a birthday party or something where I got to meet him. And it was this big, long table and everyone was kind of all over the place in their conversations. Then he goes, "Let's go around the table. I want everyone to say what they're interested in and how someone else could get started."
Kevin: 45:01 Love it.
Britney: 45:01 It was the most fascinating night of conversations ever.
Kevin: 45:06 That is so much better than a typical, What do you do for a living?
Britney: 45:10 Totally. Totally. It was really, really interesting stuff. And so now I feel like I'm just repeating some of those. But-
Kevin: 45:17 That's cool.
Britney: 45:18 I asked my friends at a book club recently that question. And my friend, Tanya, recommended the Reply All Podcast episode on the Snapchat thief. I haven't been sleeping since I listened to that. I mean this stuff is crazy. That has really peaked my interest in online security. And then-
Kevin: 45:43 What happened? Sorry to cut you off. What happened? I need to know.
Britney: 45:46 Oh my god, it talks about these hackers who hack like OG handles. So they wanted this handle that was just Lizard on Snapchat.
Kevin: 45:58 Okay.
Britney: 45:59 And so they basically are sim thieves. So they will call your cell phone provider and they will tell them, "Oh, I got a new phone." They will get the cell phone provider to switch access to the phone so they can get into your Snapchat that way, through two-factor authentication. Yeah. And then, it gets into like doxing and swatting and like really scary shit. The journalist investigating this was so scared that he was going to start to uncover who these hackers were and they were going to swat his house or his family.
Britney: 46:36 And so he met with this old CIA Director who consulted for Mr. Robot, and he told him all the steps to take to protect his online security. It will get your mind spinning.
Kevin: 46:51 Wow, that is fascinating.
Britney: 46:53 Yeah.
Kevin: 46:54 I've recently been to my first SEO Oktoberfest. There was also a guy, whose name I'm not going to mention, but certainly one of the top hackers in the world. The stuff that he told me in just a short conversation, was enough to get you thinking. Honestly, like there's so much stuff out there that can happen to you, so much bad stuff. You really don't want to know.
Britney: 47:21 No. Yeah.
Kevin: 47:21 It feels like that's one of these things.
Britney: 47:23 I know. It really is. Yeah.
Kevin: 47:25 So are going to write a machine learning algorithm to protect people's Snapchat
Britney: 47:29 How cool would that be? Yeah. I mean that would be amazing. Yeah, I don't know, it just got me thinking about that stuff a lot.
Kevin: 47:37 Oh. Yeah. There's so room for potential. See, the problem is also that that stuff gets us thinking. But I would say we're pretty much on the cutting edge of, well not super deep in cyber security, but we're techie. Right?
Britney: 47:51 Yeah.
Kevin: 47:52 We work in the tech sector. Like what the hell's my mom going to do when she's on Instagram? Do you know what I mean?
Britney: 47:57 I know. I know. Yeah, that's not fair.
Kevin: 48:00 Yeah, you need to save the world, Britney. You need to write some security algorithms because that's really bad.
Britney: 48:06 I know.
Kevin: 48:08 Another really interesting question that I like to ask besides what are you currently curious about? Is who do you follow or who do you look up to currently? Who are people that intrigue you?
Britney: 48:18 Yeah. Oh, that's really good. I just recently got pretty hooked on Mel Robinson, I think her name is. She's just incredible and theories about how bullshit motivation is, and you're never going to feel ready to do the things that you have to do. So just count down from five and do it. She's awesome. I really like the School Of Greatness Podcast with … I'm blanking on his name. He's an incredible host.
Kevin: 48:48 Athlete. Yeah, head ball player.
Britney: 48:50 Athlete. Yeah. He's so just, I feel like you would love him.
Kevin: 48:55 Yeah, yeah. I've been listening to his stuff. I know him, I know him. I mean, I just am blanking on the name as well.
Britney: 48:58 Louis Howe. Louis Howe.
Kevin: 48:59 Oh, Louis Howe. Yeah, yeah. Louis Howe.
Britney: 49:00 Or Louis Howes?
Kevin: 49:02 Yeah.
Britney: 49:02 One of the two. He's just an incredible interviewer and has really cool people on. The episode with Robert Greene is amazing. The episode with Ryan Holiday is so good. Two of like my all time favorite authors for sure. Yeah, I've just been trying to read a lot this year as well. So I've just been doing lots of Audible, Kindle, but I love the books in person. I'm a big note taker, and I like to highlight and underline and-
Kevin: 49:34 Preach. I'm doing exactly the same thing.
Britney: 49:36 Yeah.
Kevin: 49:36 It's like my treasure. It's one of the three things I would save if my house was burning, is my book collection over there with all the notes in it. Right? And all the marginalia. I love that.
Britney: 49:46 I love that.
Kevin: 49:46 Oh for sure. Easily, honestly.
Britney: 49:47 Yeah.
Kevin: 49:48 Oh yes. Me too. I recently just covered the Kindle syncing with Audible, so that you can listen to a book and read it, and it's going to sync that. I'm like, oh yes, I needed that. Because you can also take notes in the app so-
Britney: 50:00 Oh wait, you can like start listening to a section and it'll bring you up to speed? Noway?
Kevin: 50:05 Yes. It is insane. It doesn't have that for all the books. But it will basically show you the words that are currently being, that you can listen to, and often narrated by the author. So you can follow along and you can take notes by marking and highlighting. You can export them all.
Britney: 50:24 Wow.
Kevin: 50:25 Yeah. It's next level stuff. I love that. It's like ah.
Britney: 50:28 That's cool.
Kevin: 50:29 I can just listen all the time. I feel like the certain topic of this conversation is learning.
Britney: 50:35 Yes.
Kevin: 50:35 To keep learning. It's just-
Britney: 50:36 How fun is it? My mental and emotional wellbeing is always at its best when I'm learning.
Kevin: 50:44 Yeah.
Britney: 50:44 It really is. I feel so much better, I have to be learning. Otherwise, like what are we here to do? You know?
Kevin: 50:51 I'm the same way. All the personality tests that I did show that learning is one of my core kind of motivations. Curiosity, like it's all maxed up to the end. So I mean, do you know why you're so much into learning? Where that comes from?
Britney: 51:04 That's a good question. I think my parents put a huge emphasis on it as a kid. Both of my parents are in education and that was a super, super big thing in our house. That was how you got praise. You know?
Kevin: 51:18 Yeah.
Britney: 51:19 And I just loved it. My mom was a stay-at-home mom and taught us how to read from a super young age. She always made it fun. And I think this was the smartest thing she ever did as a parent, was we could either do our chores for 30 minutes or go to our room and read. So of course we would go read. Both my brother and I just fell in love books and reading and yeah.
Kevin: 51:45 That's the dream.
Britney: 51:46 So yeah.
Kevin: 51:47 I feel like it's a completely rounded story now. You know?
Britney: 51:49 Yeah. Right.
Kevin: 51:51 Like there's nothing to add. I don't know what else to add. That's a perfect conversation we just had here. It's a perfect show.
Britney: 51:57 Oh, that's so fun.
Kevin: 51:58 Britney, thank you so much for your time. This is awesome. Before we wrap it up, where can people find you?
Britney: 52:05 Yeah. People can find me on Twitter at the handle just Britney Muller. One T, N-E-Y. I'm not super active on LinkedIn but you can connect with me there and send me a message. I'll make sure to try to go through those. britneymuller@gmail is probably the best email to get access. Yeah.
Kevin: 52:25 Awesome. I appreciate the German name by the way.
Britney: 52:27 Yeah.
Kevin: 52:29 Go Germany. Yay. Oh, there are cooler languages. Anyway, let's not get wrapped up in that as well. Thank you so much for the time, Britney. You are an absolute rockstar and I can't wait to meet you in person next. But that was super insightful. Thank you so much for sharing.
Britney: 52:46 Oh my gosh, thanks for having me. It was fun.
Kevin: 52:47 You're so very welcome.
Britney: 52:49 It was so fun. I'm so excited for your next newsletter. I learn tons of stuff from you.
Kevin: 52:55 Oh god, you read that?
Britney: 52:56 Yes. I love it.
Kevin: 52:57 Oh, thank you.
Britney: 52:58 I love it, Kevin.
Kevin: 53:00 oh my god.
Britney: 53:01 It's so good. We have very similar taste in stuff. It's awesome.
Kevin: 53:04 Oh yes, I feel like we're 100% brain-synced.
Britney: 53:06 Yeah.
Kevin: 53:07 So much.
Britney: 53:08 Yeah.
Kevin: 53:08 But thank you very much. Now the pressure's on. Yeah, let's try to forget like who reads that stuff, because if I think about it, honestly I'll just stop sending stuff.
Britney: 53:19 Right.
Kevin: 53:19 That's awesome though. Thank you so much, it's super kind.
Britney: 53:22 Yeah, of course.