#117: What's in a Job Title? Maybe the Data Shows!

[music]

00:04 Announcer: Welcome to The Digital Analytics Power Hour. Tim, Michael, Moe, and the occasional guest discussing digital analytics issues of the day. Find them on Facebook at facebook.com/analyticshour, and their website analyticshour.io. And now, The Digital Analytics Power Hour.

[music]

00:27 Michael Helbling: Hi, everyone. Welcome to The Digital Analytics Power Hour. This is episode 117. At the heart of any analyst is this desire to understand systems, and optimise their outcomes. A great analyst can move whole businesses and industries, and there’s never been a time where the need for analysts, data engineers, data scientists has been higher.

00:54 MH: By now, we’ve probably all seen the McKinsey Global Institute report suggesting as much as a 250,000 person shortage in data scientists that’s coming up. Because of that scarcity, we’re enjoying unprecedented growth in our incomes, and all those things as well. But as analysts, we’re drawn to seek optimal outcomes even in this area. Talent acquisition is truly the modern arms race. Hey Moe, does it feel good to be so in demand?

01:23 Moe Kiss: Oh, geez. [chuckle] Yeah.

01:27 MH: Are you a data scientist, or a senior data scientist?

01:30 MK: I don’t know.

01:31 Tim Wilson: Or a principal data scientist?

01:33 MK: Data analyst? Pick on the day, I guess.

01:36 MH: Alright, Moe does not suffer from the Dunning-Kruger effect. And yet, Tim, is this why you refuse to be a people manager now because it’s so hard to attract talent?

01:48 TW: Yeah, let’s just go with that. I think it’s harder to retain talent once they worked for me.

01:54 MH: Okay. I’m Michael. I’ve had some modest success leveraging heuristic methods in the “begging people to come work with me so we can rise above my flaws” area. Okay. How do we put the tools of analytics and data science to work in this environment? The answer is, we don’t know, but that’s why we have our guest. Maryam Jahanshahi is a research scientist at TapRecruit. Her research uses computational linguistics, machine learning and behavioural science to reduce bias in decision-making. Prior to that, she did cancer research and obtained her PhD from the Mount Sinai School of Medicine, and today she’s our guest. Maryam, welcome to the show.

02:36 Maryam Jahanshahi: Hi, guys. Thank you so much for having me.

02:39 MH: I said a mouthful of things I did not understand at all in that intro, so maybe as a…

02:47 MJ: Is that the Dunning-Kruger effect?

02:47 MK: Right. [laughter]

02:49 MH: That’s… No, that one, I know. [chuckle] Maybe as a start for me and our listeners, maybe just give us a brief overview of TapRecruit and what you do there in a little more detail.

03:04 MJ: Sure. TapRecruit is a company that helps other companies write better job descriptions. By better, what we try and do is help you calibrate the job description for the talent market, things that are very traditional economics questions, as well as to get a broader audience, so a more representative population of job seekers to apply for your jobs. We do all of that, we change the way people write about their job description. On one level, it’s about language; but on the other level, it’s about the content of the job description.

03:40 MJ: One of the things that I didn’t really think about until I started this job is, what is in a… The core of a job description is really about the requirements and responsibilities in jobs, and if you don’t calibrate those right, you really miss out a big opportunity in terms of attracting really good candidates, and attracting female candidates, and attracting candidates from underrepresented backgrounds, and so that’s what we try and do. We do that both on the level of the job description, as well as the title of the job, and make all these sorts of suggestions based on the market dynamics. It’s a combination of lots of different things that I think is really fun.

04:17 MK: I’m so super excited about this topic, and while I wanna delve straight into it, there’s probably a point before that, which is that, continually, I hear from very senior staff that they don’t actually put enough time into job descriptions. Often, they’ll just recycle what they used last time. How do you get people to even first care about this problem?

04:40 MJ: What we find is a lot of companies… This is human decision-making at its best and worst. On the one hand, when we talk to companies, they’re like, “We get thousands of candidates for every job.” When you look at their data, they might have had thousands of candidates for their internship positions, but a lot of their other jobs are doing really poorly, they’re not attracting even 20 candidates and things like that.

05:07 MJ: You… Humans tend to remember the more extreme versions of things, the one in the 99 percentile, but they don’t remember the fact that the vast majority of what they’re experiencing is really in the middle of it. We hit them over the head with data a lot, and show them that when they do write their jobs in a way that is really calibrated to the talent market, it’s much more successful, and it’s much more… And they get a lot more candidates, and they get really good candidates as well. This happens on all levels.

05:34 MJ: What I’ve found really interesting though about that is we typically see… Job descriptions can either be the… What the person before that filled that role had, like the qualifications at the end of their time. Someone’s trying to, for example, hire a junior data analyst, but they’ve had someone who’d been with them for six years, and so now they want the spark cluster, and they want people who have data engineering skills, and they have all of these sorts of things that that person left with, but didn’t come in with. Then they also want them for that role, but at the $80,000 level, which is not something you’ll get out of someone who’s been in that role for that long, have had tenure. That’s one of the things that we frequently see.

06:13 MJ: The other is, sometimes, it’s a wish list of things that they would love to have, requirements and responsibilities that would be awesome. Everyone wants the unicorn data scientist, so it doesn’t stop people asking for data science skills, as well as data engineering skills, and things like that in quite junior roles. What employers… What we are trying to prove to them with the science is that when you ask for all of those things, what you end up with is nothing.

06:39 MJ: It’s not that you don’t get three-quarters of the way. What you get is everyone gets turned off from those roles because they’re like… Anyone who’s quite qualified in that role is like, “These guys don’t know what they want. Why the hell would I go work for someone who has no idea that this is totally outside of my area? That I… I’m a competent data scientist, and I can’t put out data… Complex data engineering pipelines, for example. That’s not my role.” You turn off qualified candidates in that process by asking for way too much.

07:09 MJ: Our goal is… And we see that you also turn off female candidates. From the confidence gap perspective, we know that this is a well-established method, even if you don’t totally believe the phenomena, the reasons behind it, but there’s that idea that… Sheryl Sandberg I think wrote in her book, on Lean In, where something like six… “Women won’t apply until they meet all of the requirements of a role, versus men will apply if they meet a smaller proportion of the requirements.” Those sorts of things, those dynamics sort of work in that level of the qualifications of a job, not just the language.

07:43 MH: You… Just to back up a little bit, you’ve got a pool of… I think it was from an earlier iteration of TapRecruit, you guys have an enormous volume of text data for job descriptions, and then you also have the outcomes. Is that… You’re essentially taking and mining the text data, and then mapping that to the outcome, be it qualified candidates, and then hires? Is that how it works?

08:09 MJ: We have all of that information. The ex… The huge… The previous iteration of our company was actually a search engine, and it still exists, and it’s very helpful from a user behaviour standpoint. We actually have a sense of what users search for, and then what they get out of that process, and what things they click on, and what things they don’t. Then actually, that’s been very helpful for the insights that we make.

08:31 MJ: One of the stories that I tell is how when you search for something, and you get a different set of results, what happens to user behaviour in that context? Then we have a smaller pool of information that is tied to outcomes, and we’ve been mapping all of these job descriptions in a lot of different… We’ve been trying to feature… Do a million types of feature extraction, and really trying to understand how the language and the content and the structure of these things interact to result in specific outcomes.

09:00 MJ: We also have on the other end of things… That’s one side of it, but then we also go… Those are usually helpful for creating hypotheses, and so my idea is that with this data science side of what we do, it generates a hypothesis, then we go and test it. We do tests with job seekers and things like that to figure out how significant is that impact? Because we might see something that has a huge impact in terms of confounding effects, but it might not really translate into real effects in terms of the job seekers, whether they’ll apply for a job or not. There could be other factors at play, especially jobs are as unique as human beings are, in some senses. We try and figure that out.

09:41 TW: You’ll actually go to… You’ll get panels, or…

09:44 MJ: Yeah.

09:45 TW: Job seekers who are… What’s kind of intriguing… Because I think it’s only been in the last year that I’ve started to really understand the cost of data. I think… Until you made that comment, I was thinking, “You just have all this historical data that use… Observer collected.” But you’re saying, “No, really to refine and hone in on this, you actually also need to at times go and say, ‘Let’s get really detailed data on job seekers’ attitudes and reactions to… ‘”

10:13 MJ: Certain things. Yeah, exactly.

10:15 TW: Certain things. Okay.

10:16 MJ: Exactly. We do a lot of experimental testing. I’m an experimental biologist by background, so that’s where I was sort of excited about taking a lot of our work, is to go beyond the data analysis, because I wanna be able to validate it. If I’m telling a company, “You should do X, Y and Z,” I need to know that it works because otherwise, you… We all know that things… I think the funniest thing was like, “Do cigarette smokers… ” As in, track with lung cancer. You could invert that whole idea and be like, “Do people who have lung cancer like to smoke cigarettes?”

10:46 TW: But it’s not…

10:48 MJ: But we [chuckle] know that’s probably not the case, but it could be due to a whole number of other factors. That’s part of what we’re trying to do.

10:57 MK: Maryam, I saw your slides from a recent presentation, and one of the things that I’m completely obsessed with is whether or not titles have the word “senior” in them. Can you chat to us a little bit about your research on whether or not it’s “Senior Analyst,” “Senior Data Scientist,” and how that impacts hiring, or even just the candidates that make it to interview?

11:19 MJ: Yeah, that’s a great question. We… Actually, we’re sort of addressing that question of seniority, and the title of a job from the standpoint of, “Does over-qualifying a job cause changes in the type of people who apply for it?” Again, this idea of the confidence gap, whether for women or other groups of people. What we did was we looked at a panel of both tech jobs, and data science jobs. These are jobs that only require one to three years’ worth of experience, from what I remember, and so calling them a “Senior Data Analyst” is a little much… Sorry, “Senior Data Scientist” is a little much. We didn’t do it for data analyst positions because they’re a little bit different.

12:03 MJ: What we saw was when you do that, the number of people who apply to those jobs dropped, the quality of the pool dropped. That was counteracting to the whole idea that you put “Senior” on a title, or you give it… You give some job a really fancy title, and then more people are gonna apply, and the good people are gonna apply. In fact, what we saw was the opposite, was fewer people applied, and less… Not as many good people applied. The result of that was that those job searches tended to be on batch a lot less successful than the job searches that did.

12:38 MJ: We tried to… In these pools, tried to drill down to the subsets. Most of these experiments were done in Tier One cities in the US, and so these tend to have a high velocity of hires for the most part, so we could track them in the same amount of time. Sometimes companies… It’s funny what turns on different types of people in the audiences because when I present this analysis, they’re like, “Yeah, but… Big deal, I wouldn’t… I’d believe it more if I saw two paired jobs.” When we show people paired jobs, they’d be like, “Yeah, but that’s an anecdote, but… ” Whatever else. Sometimes companies…

13:14 MH: What do you mean by “paired job”?

13:15 MJ: Some companies, sorry, would run a senior engineer and just an engineer both at the same time, and we would have the… I could show you the result from the ATS, where it was like one of them was successful, and the other one was terrible. They…

13:29 MH: They’ll actually do it with identical job descriptions just to see…

13:29 TW: Wow.

13:32 MJ: Yes.

13:33 MH: Oh.

13:33 MJ: They did. It’s not something I would recommend because they do cannibalise each other to some extent. It’s not an ideal type of AB test, but it was something that I was like… When I’ve shown people like, “You have to show the aggregate statistics.” When I show the aggregate statistics, they’re like, “I’d like to see the perfectly paired example.” [chuckle] You can never win sometimes with data scientists, [chuckle] I figured out, but it is a sort of interesting… It was… Our goal was to understand what causes…

14:02 MJ: We… What we wanna understand is when you over-qualify roles in all of the different ways, and we’ve looked at this more extreme, and with roles that have like asking for years of experience that people don’t have in jobs. There was… I think a couple of years ago, there was a job that was asking for crypto experience that… It was asking for 15 plus years of crypto experience.

14:23 MK: Really?

14:24 MJ: It was like, there’s no one that has that, you crazy people. We’ve seen stuff like that, you see the technologies that have only been existent for a certain amount of time, and you’re like you just can’t… As in, this is someone trying to be like… Work their way backwards, or I’m not really sure what they’re trying to do. I’d like to talk to those people. If you are one of those people, please contact me because I really… I’m very curious what goes on.

14:47 TW: That’s likely the Frankensteining of job descriptions gone horribly awry. Some… Probably a senior recruiter, who may not be a senior recruiter, said, “Let me take this, and let me just swap this out,” and it totally makes sense.

15:01 MJ: Yeah. It is one of those interesting sort of documents from a collaboration standpoint that can show you a lot about a company, and the way they work. Increasingly then, were realizing that especially super-long job descriptions, they do very poorly because you’re asking for way too many qualifications, you’re asking for way too many things, and that usually happens because no one’s standing there and taking ownership for that thing, and trying to be like, “Okay, but this doesn’t… This role doesn’t make sense. This is not something I’d hire for.”

15:30 MJ: You’ll see that that’s one of the ways we’re trying to just ascertain… It’s a very indirect way of getting to that point. But really well-written, thoughtfully put together job description is as good as a well-written, thoughtfully put together resume. We all appreciate those things. There’s no way that it’s different.

15:47 TW: Do you guys have a collection of finds of horrible, atrocious job descriptions? Is that TapRecruit water coolers conversation?

15:56 MJ: It’s like we have a Slack channel [chuckle] in our group because we have people who are constantly trying to figure out ways that our language model is foiled. They always post these sorts of things that they’re trying to be like, “What does this mean? What does this [16:12] __ say?”

16:13 MJ: We do have… It’s very easy to find terrible… I find it really easy to find terrible job descriptions when I give talks. Obviously, there are certain companies that I go to. It’s funny because we had one of them contact us the other week, and they… Every company that talks to us is like, “How are we doing?” I think the salesperson on our team was like, “Yours are some of the worst job descriptions I’ve ever seen.” They’re like, “Really?” We’re like, “Yeah, we use them as examples when we wanna give an example of a horribly over-qualified job description.”

16:47 MJ: They were like… I think at the end of the call, they’re like, “We love you guys.” We’re like, “Yeah, I know because we’re telling you how it is.” But really, honestly, you can… There are certain companies that have an MO essentially in that. There are gonna be odd ones at other companies as well, but I know which companies to go look for which have very, very corporatised, super-long job descriptions, and those are problematic for them as well.

17:12 MJ: Yeah, it’s becoming more… They’re becoming more and more aware of that sort of stuff, but as an industry, consulting job descriptions, if anyone was to look at them, like look at the big consulting companies, they have some of the worst job descriptions ever.

17:27 MH: They’re recruiting on reputation almost.

17:29 MJ: They are, but they also have problems. A lot of them have come to us, they have problems, those sorts of things. Yeah, they’re also like, “Yes, we’re this… We’re a Big Four consulting company,” but it’s like, you also have problems, so let’s try and work through it because you can use your reputation for good or for bad. This is something that it can be used effectively for.

17:50 TW: Do you have to coach what the job title is versus the internal role? Like I know a guy who, he was… I think he… I think where he wound up getting… His company got acquired, and then his job title changed, and it dropped way down, which is really weird given who he is and how much experience he has. I think of the… I think of Deloitte, that has manager and senior manager. They have very defined roles… Or their companies, if you look at banking, where you’re a VP after you’ve been there for six months…

18:23 MH: You start as an Assistant Vice President.

18:25 MJ: Yeah.

18:26 TW: Yeah, start as the AVP. Is that part of it calibrating and trying to say, “That may be the title that you’re gonna list, but then that’s… Either you need to recalibrate your HR and internal levels, or you can’t advertise necessarily for what you use internally”? Or does that not happen?

18:43 MJ: Job titles are the hardest thing that we’ve had in terms of getting people to change, but when they change it, they see huge results. That is one part of it. But in general, within an industry, we try not to tell the whole industry [chuckle] they have to change things to make it consistent with us. What we do try…

19:04 MJ: I know, for example, in investment banking, when you come out of an MBA program, typically, you’re an associate, which sounds like a very low-level role, but it is… Then you go on to become a VP, and then you become an ED, and then all the rest of that side of things. If you fall outside of the markers of what is typically searched for by candidates, that’s when we pull you up, and start pushing on the fact that this looks like a VP role. Typically, an investment banking candidate with this much years of experience is looking for VP roles at other banks. Let’s not call it something weird just because it’s fun.

19:42 MJ: We see all of these sorts of things. We see a lot of weird storyteller or artist… Like what, data artist and/or things like that, that people are trying to be creative with those titles, and we’re like, “That’s great. You can call it whatever you want, internally. You can also call it whatever you want in the job description.” But you have to admit that, as in the way that job search happens, the title is a big driver of search results, and so you have to optimise that title to get search results.

20:08 MJ: I think it’s really important to say in the job description, “Internally, this will be called this,” because people do anchor to a job title a little bit, and it does impact how they… What things they click on. That’s… You imagine what that role will be like. But the reality is, we all have job titles on LinkedIn, which may or may not be different to what our job titles are on our business cards or signatures, or what we use in talks and things like that, and it’s always context-driven. Part of that is trying to make sure that for this context, which is to be able to be found by candidates, you can actually… It actually works.

20:46 MK: Okay, I understand why taking “senior” off, I think, would expand a pool of candidates, but what I don’t understand is then wouldn’t that have… You have a job description that says “data scientist” or “senior data analyst,” how would you differentiate then as a candidate, or even as a company between someone who has one to three years experience in one programming language, and a little bit of experience with some machine learning versus someone that has five to seven years’ worth of experience, three programming that… Do you know what I mean? How do you then… Is that… Does that then… Do you not worry about the title, and you worry about the content of the description to differentiate between how senior the level is?

21:29 MJ: In our case, yes, it is very much the description that drives what… How senior we interpret the title. We also have all these other tools that we add on that I think are really interesting from a behavioural standpoint. To guard against people asking for way too many qualifications, we have a salary estimator tool. As you’re writing the job description, it’s giving you an idea of the salary. The whole point of that is if you’re asking for like five to seven years of ML experience, that’s gonna be a much more expensive candidate than someone who doesn’t have that.

22:02 MJ: The point of that is to push back a little bit before the job description goes out, to get it… Tone it back down to where you actually want it to be. It’s okay if you want five to seven years’ worth of ML experience, but that’s not gonna come at a… It’s not gonna be cheap. The worst thing that you could do is go through that whole process, and not hire someone, and then have to do the whole thing again, which is what everyone hates, from hiring managers to recruiters to everything else.

22:25 MJ: One part of that is definitely from the job… As in parsing out what are the requirements of the job, and what they mean. That’s a big effort that we have that… From a language model perspective, trying to map those skills and qualifications.

22:40 TW: Do you get pushback on if there are terms or phrases or characteristics that do look like they have a gender bias? Do you wind up having clients who don’t believe that, or push back on that?

22:56 MJ: That’s the easiest thing to get people to change, which is really interesting. They are most happy to delete those sorts of phrases. It’s much harder to get the harder things to get changed, so the titles, or reducing the number of qualifications, and things like that. Yeah, those are the hard things. It’s like we’re the dentists who’s gonna make you do the work that you really don’t wanna do.

23:21 MJ: But yeah, the gendered language is really probably one of the easiest things. The only pushback we’ve gotten is actually from non-profits that tend to be disproportionately attracting women, and they wanna go the other way. But what my experiments, and I’ve spoken to them, show is that that doesn’t work that way, and that’s not the way that…

23:42 MJ: If you put mixed language, it’s not like it’s gonna attract more men. It’s like you have the pool… That you have the pool, and you just got to attract the broadest pool, and then go through that process. But it is an interesting… It’s become a really interesting idea, is that there can be codes in these job descriptions that will activate these latent job seekers, that I think is fascinating, but you have to be like, “You won’t do that, so why do you think that there are these zombie job seekers that will suddenly rise from the dead and take on… ”

24:11 MK: Can you give us some examples, what… How do you know if you’re using gendered language in a job description?

24:18 MJ: We have things as extreme as military metaphors and sports metaphors, that happen in a lot of startup jobs. There’s a… Not necessarily gendered, but they’re not… A lot of people don’t understand what they’re talking about, and so it’s very strange. One part of it is not stripping the personality, but it’s like this doesn’t make sense to the vast majority of people, so we try and get rid of some of that.

24:42 MJ: Some of the things that people talk about… Traditionally, in social psychology, gendered language has been on the idea of… There’s one axis that they talk about a lot, which is about communalism versus competitive… More self-agentic language, I would call it. On the communalism, that’s typically associated with women. That’s things like collaboration and working as a team, and that side of things. On the other side of it is agentic language, which is much more self-starter, competitive, that sort of thing.

25:16 MJ: One part of what we’ve been trying to figure out is, does… Some of the stuff can describe a team, or it can describe an environment, it might not describe the candidate. We’ve been separating out those sorts of things as well, trying to understand does talking about a competitive environment turn people on and off versus talking about that you need to be competitive or a self-starter or like an independent self-starter in this role?

25:42 MJ: There’s a lot more that goes on with that sort of language that we really know. A lot of the classic experiments, I will say, were done with 19-year-old university students in psych classes, and haven’t really been replicated with real job seekers. They were also done with job descriptions that were like what an academic thinks is a job description, not what a real job description looks like, and so they…

26:07 MJ: Like I said, you read them, and you’re like, “I can tell what the intent of this experiment is.” Actual psych students were very aware of the intent, but it’s like we’re replicating some of those experiments just to see how extreme the effect is. Not surprisingly, actually, it hasn’t been anywhere near as extreme as the original study showed, so it’s not as dramatic in impact in terms of the candidate pools as we would imagine it would be. That’s part of what I do, is that… The behavioural science side of things, is like, “Does this really happen? Does this have dramatic impacts on the candidates?”

26:42 TW: One of the other ones that I think I’d seen you mentioned was… Which isn’t so much a gender thing, was the… But it’s like the senior, senior data scientist, and having a negative effect was I think the unlimited PTO, and that being a benefit, I guess, description.

26:57 MJ: Paid time off for the Australians?

27:00 MK: Yeah.

27:00 TW: Unlimited, sorry. Yes, unlimited paid… Unlimited vacation.

27:03 MJ: We just call it the annual leave.

27:04 MK: Yeah. Okay. Wait, sorry, that’s got a negative effect?

27:07 MJ: It depends on how you phrase it. One of the things that we noticed is… Actually, to backtrack a little bit, perks and benefits are not typically part of job descriptions, or haven’t historically been. You always assume that if like Coca-Cola is advertising for that position, there’s a bunch of… If it’s a full-time position, you’re gonna have like 401k, which in Australia is…

27:28 MK: Superannuation.

27:29 MJ: Superannuation. You’ll have health insurance because you pay for that. You don’t have to go through your company necessarily. Those things are not typically in job descriptions, but actually, a couple of years ago, and I started noticing this, a lot of startups, especially like… Started using that as a tool, so they would talk about their…

27:50 MJ: This was something that wasn’t happening very much, but they would talk about their perks and benefits in their job descriptions because the goal was to give someone a sense of the culture in this organisation that you’ve never ever heard of. Essentially, what we’re seeing now is actually it’s been taken up by Fortune 500s and big companies because they’re realising that they’re getting a lot of their candidates through search, and that’s where you get a sense of culture, and things like that.

28:13 MJ: We’ve been looking… I’ve been tracking actually perks because I like to see how… As a company gets bigger, what perks are they adding. I did a bunch of cluster analysis of the perks, and it was really interesting seeing like the unicorns getting very close to the Fortune 500s in terms of the perks that they offered. This was more general than job descriptions. There’s a maturity cycle that goes on with that.

28:35 MJ: But we also looked at the phrasing of the perks. One of the things that you would imagine is, in general, you will have some form of paid time off with most companies, but the phrasing of that matters. When we ask job seekers how they interpreted different types of perks, and what they thought the value of those perks were, what they saw was flexible vacation, or paid time off did much better than unlimited time off, which was sort of a negative in some senses.

29:02 MJ: People are assuming that they’re not actually gonna get time off because the Netflixes, and all of the Silicon Valley, like tech giants made that a cornerstone of the way that they can deal with benefits, particularly in California. It’s become something that’s quite widespread, but everyone knows now that that’s something that means that you don’t get as much time off as you would imagine, and so the language around that was actually very important.

29:26 MK: Which is funny because… There are some companies that have offices in the States and in Australia, and I won’t name which ones, and they give their American employees unlimited vacation, and they give their Australian employees four weeks. It’s because they know Australians will actually use unlimited vacation…

29:41 MJ: Exactly.

29:42 MK: That’s kind of the opposite of the intent, I think, behind the policy.

29:46 MJ: It is.

29:46 MH: Yeah, I’ve tried to be kind of Australian in that regard, so it’s taking the best of all cultures. [chuckle]

29:55 MK: But it’s funny that that’s… Yeah, now people are seeing through that.

29:58 MJ: It is, but I think it reflects a different type of culture. I… What I’m very curious about that doesn’t come through all of this is how we interpret company culture, and how we communicate company culture through these documents. That’s, I think, what these sorts of phrases do… Does, is it gives you an idea of how competitive that environment is, and something that was very clearly in the tech world that transported to other places, and how it gets re-used. It’s a cool area that is really interesting.

30:29 MJ: Unfortunately, with perks and benefits, there’s no easy AB test, although there are… One of the things that I would have to say that’s really nice is job descriptions being written by humans is great because they sometimes forget things, and so we have some isolated incidences of AB tests, but they’re just not big enough to be able to interpret anything out of because you have to match them in all sorts of way. We don’t have a big enough sample set, but it is one of those things that I think is, from that perspective, really fascinating about what attracts people to companies, and how those perks and benefits can communicate culture.

31:02 MK: Have you by any chance… It’s funny, actually, when I look at a job description, the first thing I do straight afterwards is Google the company’s values, and then Google to see if there’s any reviews on their culture. That’s like legit the first thing I do. But I’m actually wondering, is there… Have you done any research to date on whether inclusion of those perks in a job description impacts the diversity of the pool of candidates?

31:24 MJ: That’s the difficulty. We don’t have a good alternate test because, usually, it’s very… It’s easier to compare within a company because they typically act within a geography, and everything else. We tend to balance out our data sets. We don’t have a great way of dealing with that, unless asking for some of our customers to do some AB tests. But it’s definitely something that I think is on the mind of everyone. It’s one of those areas where people… You see these articles all the time in the popular media about how employees will be willing to take X percent less salary if you give them this much more leave, and things like that, cheap ways that companies can exploit you through perks and benefits, which is essentially [chuckle] what people talk about a lot.

32:09 MH: We’re gonna call that total compensation. [chuckle]

32:12 MJ: Yeah. One of my friends did this very silly thing a couple of years ago when her company offered her two options. They weren’t doing super well as a company, and so they gave her the option of extra annual leave in Australia, or extra paid time off, or… And a reduced salary versus just holding what she has and holding it up. She actually went with the pro rata-d kind of option, which has had dramatic impacts for her for the last couple of years because she came off on a lower salary, and has been trying to work her way back up to where she originally was, in part because of that leave scenario.

32:56 MJ: Smart on the side of the company, but it wasn’t something that I had anticipated at the time when she was telling me she was gonna do this. Otherwise, I would have been like, “What are you doing? You’re pegging yourself so much lower than you actually really need to be.” I would be like, “I will take the annual leave out of… ” As in, “I will pay for it myself in some ways because it leaves me at a higher salary, rather than going down and pro rata-ing myself to that.”

33:21 MJ: It has dramatic impacts when you start thinking about leave, and how companies go through distress. It’s fine, you can through distress; but as a job seeker or as a person who’s on the other side of that, you’ve gotta make sure that you don’t let your salary drop too much because women do that a lot more than men, and so that has dramatic impacts on the rest of her career because she’s now trying to get back to where her male colleagues were a couple of years ago, which is…

33:43 MK: Wow.

33:44 MJ: It’s really silly. This happens in some countries, it happens in some environments. It’s one of those things that when you’re counselling friends to think about is that value is probably more important, wherever you ended up with, because that’s an anchoring value for your bosses and everyone else.

34:00 MK: Are there any specific… Is there any specific around language and salaries in job descriptions that people should be aware of?

34:07 MJ: Yeah. That’s another really interesting thing. There was a lot of startups that talked about competitive salary in their… As part of their perks and benefits. Because in the US, for your Australian and British listeners, they don’t mention salary, typically, in job descriptions. There are rare cases where they do, but it’s not very common. Part of why perks and benefits are actually really interesting and important is that’s a way to suss out, are they gonna pay on market, or are they gonna be a little cheap about that?

34:40 MJ: A lot of companies were talking about competitive salary. We saw this been taken up by a lot of companies, but what we saw was that people just assumed that they’re not gonna get paid very well at those companies.

34:48 MK: By saying “competitive salary,” the assumption by people was, “I’m gonna get paid less.”

34:53 MJ: Yes.

34:53 MH: “We’re gonna compete with you, the employee, to… ”

34:56 MJ: Yeah, exactly.

34:56 MK: Wow.

[chuckle]

34:56 TW: push it down as far as possible.

34:56 MJ: It’s like, “Bet against yourself,” is kind of the thing that they do.

35:00 TW: Wow.

35:00 MJ: It’s like… I know it’s a virtue signal. They’re trying to signal that they’re gonna be generous and stuff, but there’s other ways to be generous and to signal that. This is one of those things that came out of the whole startup boom, at least here, is that it’s like competitive salary and equity package. It’s like, “Yeah, you’re gonna give me lots of equity, but so is a lot… ” As in, if you don’t say that, it’s still not… As in, I’m not gonna interpret that otherwise, but sometimes it’s like saying too much is less than not saying anything at all.

35:28 MK: Yeah, wow. You recommend not saying anything?

35:30 MH: I’m literally following along on these certain companies’ job descriptions right now, and there’s literally one line, “competitive compensation.” I’m like, “Oh?” Yeah. [chuckle]

35:42 TW: I bet.

35:45 MJ: I think it’s important to ask yourself if you are the job seeker in that environment, “How would you interpret it?” There is a lot to be said about your gut instinct as a job seeker, to giving you a clue about how other job seekers intend. We try and put ourselves in the shoes of the job seeker as much as possible when we can’t actually directly ask them.

36:04 MH: That’s awesome.

36:06 TW: I feel like I have to ask. We’ve talked about the senior, so what would your take be on a junior level analytics analyst? One clunky language, plus the “junior.” Does “junior” signal like, “We’re literally looking for somebody who is straight out of school?” That that may be… Because if it then has a job description that expects one to three years of experience, or…

36:30 MJ: I actually love that because that is the exact experiment we did. We looked at senior data scientists that required one to three years’ worth of experience, data scientists, which required one to three years of experience, and junior data scientists. There’s not that many junior data scientist roles that require… Just in general, it’s not a very common term. What we found was, when people search for data scientists, both senior and junior data scientists don’t come up very high on search results. They get… Those jobs get much fewer views, they get very few click-throughs and things like that. That’s very common. That happens on every platform, whether it’s Indeed, LinkedIn. I’m sure it happens on SEEK as well.

37:08 MJ: But actually, junior data scientists, despite of getting fewer clicks, got a lot of applications. For the number of clicks, the application number was up probably… I think it was like a 20% change. When you put “senior” in it, you got 20% fewer, when you normalise for all of the other variables. When you had “junior” in the title, it got 20% more. That’s removing a barrier, I think, of entry to a lot of people who are very interested in those roles.

37:36 MJ: Unfortunately, there’s just so few jobs, there’s so few of those roles, that we can’t really understand, “Are they attracting a broader candidate base?” and things like that. That’s what I’d love to be able to study. But we know that with “senior,” it definitely attracts fewer women at every stage of the process, and fewer qualified women at every stage of the process. It’s a just much more less successful process.

37:58 MK: I’m really struggling with this a lot, and I think the reason is that I do search for the term “senior.”

38:04 MJ: That’s probably because you are a senior.

38:06 MK: Yeah, but I just wonder…

38:09 MH: Yeah, that’s true.

38:09 TW: But she’s a Xennial, so I don’t know.

38:10 MK: Oh, geez.

38:10 MH: We should have an episode of the podcast where we don’t mention that.

38:15 MK: But then I’m just…

[laughter]

38:15 MH: I’m sorry, but…

38:17 MK: I’m just wondering, how do you then determine whether the job description… Yeah, I’m really struggling with this concept because then does it come down to your ability to negotiate a little bit, or… Because I just think about, how do you know what the salary’s gonna be? Or often people look for more senior roles, or more junior roles because it’s reflective of their salary range, and so that is in your mind when you’re looking. But yeah, I’m… I feel like I’ve gotta do… You’ve gotta do a 101 course on job hunting for people.

[chuckle]

38:54 MJ: It is something that I’ve thought about a lot, but one of the things we say is like when you pass a three years’ worth of experience, you should be looking for senior data scientist, or senior data analyst positions. That’s normal. My problem is when you’re asking for one year or straight-out-of-school, and you’re suddenly asking people to be a senior data scientist, and no one…

39:14 MJ: I’ve had… I was actually gonna include screen chats from conversations where I’ve had with people where I’m just coaching friends and being like, “You should apply to this job.” They’re like, “It has ‘senior’ in the title.” I’m like, “So what? It requires two years’ worth of experience, you have three years’ worth of experience. You should be applying to this.” But there’s this barrier.

39:32 TW: But you know what’s funny though is that a lot of the people coming straight out of school sometimes feel they should be in those senior roles very quickly anyway, and so there’s a lot of title inflation as well.

39:45 MJ: Exactly.

39:46 MH: To what extent is… Or do you know… I’m just gonna ask, and I don’t know. We took that senior thing, and we’ve been talking about it. To what extent are the people not applying to that job because they don’t feel qualified versus they look at that job, and realise they’re not asking for the right things, and I could tell, as a senior person, this is not the right job?

40:07 MJ: Right. I can’t tell you that for sure. What I can tell you is what we did was match reasonably similar jobs with a similar sort of as in requirements in terms of Python and things like that, as in typical requirements for either Python or R in statistical analysis in these sorts of roles. We tried to match it as much as possible in big cities, and those tend to be quite templatised.

40:34 MJ: I can’t tell you what’s going on on the end of it, but I can tell you that in terms of confounding factors, we’ve tried to reduce them as much as possible. That’s part of why we’re looking at a reasonably early level role is because those tend to be kind of similar across the board, and so someone… You could be turning yourself off for a variety of other reasons, for sure. This is one of those things that… I think if I didn’t see the junior thing, I’d be a bit more [40:57] __; but it was funny to me, when I showed that graph, it’s like senior drops you by 20%, and junior increases you by another 20%.

41:04 MK: Do you think that could be a reflection of the industry though, where people that called themselves data scientists existed, there were less of them, I guess in the past, and more people want to call themselves that now,2 so therefore, because the industry is less mature, there’s less people that are of a higher qualified standard? Does that make sense, or have you seen this in other industries?

41:25 MJ: These are… One of the things that I would say is this is a relatively junior position, that we’ve been looking at, to sort of exclude that that’s an issue. We’ve also seen it with other sorts of tech roles as well. These are a little bit more established, they’re software engineer roles, and things like that. We’ve also studied it in terms of most subtle ways of over-qualification, so not just in terms of title, but asking for things like agile experience.

41:52 MJ: People who do software engineering would be really familiar with the fact that you typically work in agile… As in, you might have scrum master, and be used to agile workflows in large tech companies. What we found was asking for agile experience reduced the quality and the diversity of candidate pools. That’s just like dollar for dollar matched on very junior tech roles. When you suddenly start asking for that, you’re like, “I don’t want someone from a boot camp.” Essentially, you’re signalling that I don’t want you unless you come from Google, Amazon, some of those big tech companies, or Microsoft somewhere where that is part of the thing.

42:25 MJ: Even if you’ve worked in an agile system, you wouldn’t have been sort of inculcated into that environment. Maybe it’s important, maybe it’s not in your context; but what we’re seeing is there’s been a huge growth of that. Sometimes they’re like, “Okay, but do you really need that because you don’t seem to have an agile workflow going on in so far as everything else?” But it’s one of those throwaway things that comes up. We’ve seen that, those more subtle things, or asking for ETL experience in data science roles.

42:55 MJ: The ETL is typically something that a data engineer would do. I know that some data scientists do it. They don’t do it very well, and so that’s another barrier that pops up from time to time. There are just subtle things that actually… If someone’s going through the list, and it doesn’t have to be a woman or a man, but if someone’s going through the list and being like, “That’s something that I don’t have, and if it’s gonna be between me and someone who has it, I’m not gonna apply.” Job seekers are savvy enough to know where they have a chance, and where they don’t. That goes on as part of the process.

43:26 MH: No, but it sounds like there are definitely situations where people should be trying to shoot just a little higher, and especially women and minorities, because they tend to be the ones who don’t think of themselves naturally that way, and so they should just be like, “Okay, you know what, I don’t see myself on every single line item, but you know what, I’m gonna apply anyway, and then just sort of see where it goes.”

43:49 MJ: Yeah, I agree. But that’s hard, it’s hard to put it at the feet of the people who have…

[chuckle]

43:53 MH: No, I completely agree. I completely agree. I’m not saying that’s the solution, is everybody just pick yourself up by your bootstraps and go apply for jobs you’re not qualified for.

44:03 TW: I think that’s more of a direct call to our audience, that if they are in the market and looking…

44:09 MH: Yeah, if you’re looking…

44:11 TW: That in a very targeted way working around a broken system.

44:12 MH: Push yourself.

44:12 MJ: I totally agree, that’s the advice I give to people who I do… Friends of mine who I career counsel, and other people who… Anyone who’s looking for a job, I sort of pitch them roles that are a little bit more senior than they expect to be because one part of what you’re doing is also levelling up. I know that it’s always a negotiation anyway. Like a company will take someone who’s a little more junior sometimes because they have a bunch of growth to sort of develop.

44:41 MJ: As in, we hire on growth in our company as well. It’s one of those things that we think is really important, and of interest as well. Absolutely, I think as people who career… Who counsel other people as well, don’t just find a job that fits that person’s current qualifications, like a little bit further out because they’re always wish lists as far as I can tell.

45:01 MK: One area that I haven’t seen from any of your work to date that you’ve investigated or analysed, but I’m really curious on is technical tests. Is that something that… Have you given advice to companies on that previously? I personally have a bit of a rant where I think really time-intensive technical tests could really… From my, I guess, perspective, I say that it could potentially impact the pool of candidates because if you have a family and a full-time job, where the hell are you getting two days to do a test for every job interview that you have? It’s a real passion topic of mine, and I’m trying to restrain myself from losing my shit.

45:41 MJ: I totally get that. One of the… We have not worked on technical tests. I know that they’re proliferating. I would love to see some of the companies tell us what is the gender breakdown of the people who pass this test, and how… Who does well, if there are characteristics that are associated with them. Because they are a little bit like SATs, all types of standardised tests that you can train for them, and that’s… That gives a lot of privilege to people who have more traditional backgrounds, or who have time to train for them and things like that. That’s on one side of things.

46:16 MJ: What we found actually for outside of things, we’ve hired engineers and things like that, is we would come up with assignments, we would pay the engineers for those assignments that they would do, and they should take more… No more than 10 hours or whatever else of their time, and they have as much time as they need to get those 10 hours of work done. Usually, it would be quite relevant to what they were doing anyway. If it was like, “Okay, you have this problem… ” Like as in, “You wanna set this up in a new framework, we wanna see how you do it, and so talk us through,” so it tends to be a little bit more big picture. Like I don’t really care about coding, like as in the… Like voting or coding and things like that. It also gives them a sense of what they’re actually gonna be working on.

46:55 MJ: That is a little bit more time-intensive, it’s more expensive, and that’s actually really good that it’s more expensive, so we’re very thoughtful about who we give that to because the worst thing that you can do, and I’ve heard of this, is… Like HackerRank has become so cheap that companies are using it as a screening tool. You submit your resume, and then they give you a HackerRank test, and it’s now just a bar that you have to pass, and it’s like $1 or $2 from what I’ve heard of some companies that recruit a lot of engineers. It’s so cheap that you just throw that at everyone, and you’re not really respectful of the candidate’s time that it takes to get to the point where you could be successful at those.

47:31 MJ: Yeah, and I see lots of women on various Slack channels and things like that, being like, “How do I study for this thing, and that thing?” That’s one of those areas that I think is problematic. I think for recruiting engineers, it can vary. In our case, we’re looking for more senior ones, so the project system works very well for us, after we’ve screened through a bunch of questions and things like that, but we’re still writing a book on that sort of stuff as we go along. It’s still early days, but we found that that was fair and successful because I’ve done those assignments. They’ve taken up like… They say it takes up 10 hours, it really takes 20 hours.

48:03 MK: Whoa!

48:04 MJ: They’re really stressful, and you’re sitting there and writing and doubting yourself, and the least we could do is make sure that you’re actually getting paid for that at a reasonable rate, and that… It becomes a starting point of a conversation that goes beyond your resume, and that’s really what’s more important to us. One of the things that we noticed is that the people… These resume matching systems, things like the Amazon AI stuff that came out six months ago, where Amazon was trying to tie the resumes of the applicants, the successful applicants who got the position to what could they come up with… An algorithm that would predict who were gonna be successful, so that they could implement it as part of their recruiting process.

48:43 MJ: The problem with those sorts of systems, and I noticed this, is none of the people on my team, none of the people in my company, certainly not me, would have ever gotten to the positions that we’ve gotten to based on our previous record on an algorithm. Those sorts of algorithms tell you what historical people have gotten from those roles, and they also tell you… They sort of match for the experiences you have. It doesn’t match for growth. That’s very hard to do because if you… Those of you who are familiar with topic modelling or matching documents of diverse outcomes, the documents that score the best, if you were gonna match a job description and a resume, are the ones that have the most overlap, and that’s usually someone who already has that position. It doesn’t let you grow.

49:26 MJ: That’s the other problem with those sorts of systems that we think a lot about because that’s obviously a big inefficiency in the system, it’s a bias-ridden system, it takes way too long. There are a whole lot of problems with it, but we’ve gotta really think about, this is a very deep AI question and algorithms and ethics and that sort of side of things that we really need to think through.

49:48 MH: Okay, we do have to start to wrap up, and this has been amazing. Thank you so much, Maryam.

49:53 MK: Oh. Can we have her back for another full hour? I’m so not done, I’ve got a lot…

[chuckle]

49:56 MH: Yeah, I think if we could, maybe we do a show where we just go through job descriptions together and live critique them. [laughter]

50:03 MK: Totally.

50:04 MH: Anyway, okay, but one thing we do like to do on the show is go around and do a last call. It’s just something we’ve seen recently, something that’s of interest to us. My last call, there is a video I saw, and I think I saw it on LinkedIn, it was from Evan LaPointe, who’s a good friend of mine. It was a TEDx talk called “Why do so many incompetent men become leaders?” It basically walks through the guy who did it, I’m gonna mispronounce his name, but Tomas Chamorro-Premuzic. I don’t know how to say his name.

50:38 MH: He basically is a… He does research on this, and basically, it’s this idea of confidence, and conflating confidence with competence, and why that be sort of persistent in our world. Anyway, it’s a really great TEDx talk. It’s really short. Really highly recommend it. I think it offers a very great perspective on… It will definitely ring true when you watch it, you’ll be like, “Yeah, I’ve definitely seen this,” and kind of relevant to what we’ve been talking about a little bit. Okay, Moe, what about you? Do you have a last call?

51:10 MK: Okay, I’ve got two, but they’re both like eh, so it’s fine. They add up to be like one good one. I’ve just finished a really good book by my main man, Michael Lewis.

51:21 MH: Love it. I love Michael Lewis.

51:23 MK: I freaking love him. I love him.

51:25 MH: Yeah. He’s good.

51:26 MK: It’s called The Fifth Risk, and being ex-government, it’s also incredibly interesting. It’s basically about the handover from the Obama administration to the Trump administration, but he goes into so much detail about what a bunch of the different government departments in the US do, what their responsibilities are. Total page-turner. Jamie and I were reading it at the same time, and fighting over it when we went to bed, so we both had to take turns because it was…

51:52 MH: Wait, like one physical copy of the book, and you were like…

[chuckle]

51:55 MK: Yeah, yeah, yeah. With two…

52:01 MH: Oh, newlyweds. You guys are adorable.

52:01 MK: Yeah. But the other one is, I went on about this book recently called “Why We Sleep” by Matthew Walker, and I’ve been trying to get everyone in the world to read it. It’s my like pilgrimage to get everyone to be happier and have better lives by sleeping more. He’s actually done a TED Talk, so if you hate reading, and you just want a brief synopsis of what his book is about, you can watch his TED Talk.

52:24 TW: I’ve made it to the point in his book where he says he’s okay if you actually fall asleep by reading his book.

52:28 MK: Yeah, that’s the start.

52:29 TW: Yeah.

[chuckle]

52:31 MH: All right, Tim, what about you?

52:34 TW: I’m gonna have to build on Moe’s with a mini one, which is that Michael Lewis now has a podcast called Against the Rules, so if you’re a Michael Lewis fan.

52:45 MK: Oh!

52:47 TW: The first season’s all about the idea of neutral parties, refs, so some good episodes. But my main last call is going to be another podcast which I picked up on through a friend of the show, past guest, Walt Hickey, Numlock News. On his… One of his weekend interviews, he had Derek Thompson from The Atlantic, who has a podcast called Crazy/Genius. Season three, episode one is about privacy. There’s no mention of GDPR, no mention of CCPA, but one of the people they talk to is… Talks about how we’re using…

53:22 TW: Really, privacy’s the wrong term, that it’s really surveillance capitalism, which is kind of an intriguing kinda way to think about privacy. I’ve actually started working my way through all of the past episodes of that podcast. They’re short, they’re like 25 minutes long, and he gives kind of two different viewpoints on a subject; but season three, episode one of Crazy/Genius.

53:43 MH: Outstanding. All right, now, Maryam, you’ve seen us kind of do it, you understand it’s just like the think house, no rules.

53:49 MK: [53:49] She’s looking way more concerned now. [chuckle]

53:54 MJ: No, I have an example of every one of the things… I’m sure many of you have already read Bad Blood. It is a story of Elizabeth Holmes and Theranos, and essentially how for 11 years, they avoided… They raised money in ways that it was nonsensical. John Carreyrou’s book, it is a total page-turner. We had two copies in our house, but we were racing each other to finish it. We both have Kindles, obviously, so we were racing each other to finish it [chuckle] because I was like, “I’m at this point,” and my husband would try and catch up, and just be like… It was amazing. It has been extraordinary, I’m super excited, one of my best friends is moving to San Francisco. She’s living one building away from Elizabeth Holmes, so I can’t wait to stalk her there until she goes to jail, which she will be doing.

54:47 MJ: But it’s really fascinating to see how Silicon Valley, someone who’s total outside to that whole environment, sort of takes an idea, and as someone who’s a medical scientist, who kind of is like… What she was proposing is bananas, but how organisations can kind of continue on and persist in bases of fraudulent endeavours without knowing, and it’s so interesting from a managerial perspective to be like, “You don’t do this.” [chuckle]

55:13 MH: You’ve described half the field of digital analytics as well, so yeah. [laughter]

55:19 MJ: Yeah, I think there’s a lot of ways to echo around that, that it’s fascinating. I totally recommend that, if you haven’t read it, it’s total… You’ll read it so quickly over a weekend.

55:30 MH: Yeah, I haven’t read it, but it’s on my list, and I did see the documentary, I think, that HBO did, and it was wow.

55:37 MJ: That wasn’t as good as the book.

55:39 MH: That’s what I’ve heard.

55:39 MJ: I was disappointed. Yeah.

55:40 MH: I’ve heard the book is better.

55:42 MJ: Yeah, unlike the Fyre Festival documentaries, which are something else altogether.

55:47 MH: Yeah. Okay, you’ve been listening, and I know you’ve got questions. You may be someone who’s doing hiring for data scientists or analytics, and you’re just like, “Wow, I’ve gotta learn more.” We’d love to hear from you. You might be someone looking for a job, and you already got some advice as well on the show, so we definitely would love to hear from you. Please reach out if you’d like. The best way to do that is on the Measure Slack, or on Twitter, or on our LinkedIn page. We’d love to chat with you, and I’m sure… Maryam, I don’t know, do you have a Twitter that people could ping you on?

56:19 MJ: They can totally ping me on, but it’s… I usually just like panda videos on that, so I don’t think… [laughter] If you’re enjoying your world, it’s usually a good thing, but I will come and check in on all of those environments and things like that.

56:33 MH: Yeah, if you’ve got great funny panda videos, post them over on Twitter to Maryam, and then you’ll all be good.

56:40 MJ: And she’ll find you. Yeah. [chuckle]

56:42 MH: Anyway, but Maryam, what a pleasure. Thank you so much for being on the show.

56:47 MJ: Thanks so much for having me, guys.

56:49 MH: I think it was just great, and sort of like, as always, we never even get close to all the best stuff, and so I feel like there’s a lot of unspent ammunition that we’ve got laying around, not to use a deterministic militaristic description, but I just did. Anyways, okay, we’ve gotta wrap it up, and so I’m gonna do that, and the way that I do that is because I wanna tell everybody out there, no matter if you’re a senior data scientist or a junior analytics analyst, the best thing you could do, and I think I speak for my two co-hosts, Moe and Tim, is keep analysing.

[music]

57:31 MK: Thanks for listening, and don’t forget to join the conversation on Facebook, Twitter, or Measure Slack Group. We welcome your comments and questions, visit us on the web at analyticshour.io, Facebook.com/analyticshour, or at Analytics Hour on Twitter.

57:51 Charles Barkley: Smart guys want to fit in, so they’ve made up a term called analytics. Analytics don’t work.

58:00 Tom Hammerschmidt: Analytics, oh my God, what the fuck does that even mean?

58:08 MH: Take 100, 200 keywords and dump it into this thing, and it does some sort of rudimentary Markov thing…

58:15 MK: Maryam’s actually from Australia.

58:19 MK: Yeah, my sister’s like the original gangster of our family who’s in data and analytics.

58:24 MK: Oh my God.

58:26 MK: We’ve been trying to do this sister on sister thing, so we’re like, “Now we need to get another two sisters involved in these conferences.”

58:32 MH: Moe and Michelle.

58:33 MK: Yes.

58:34 TW: Moe usually just recommends thinking fast than slow.

58:37 MH: Hey, that’s not true.

58:40 MK: Hey, I have moved on to the full suite of Michael Lewis books.

58:44 MH: Data-stitions, am I right?

58:47 TW: Yeah. Like, dude, you could have brushed your hair.

58:53 MH: Oh, that’s cold. Oh my gosh, wow, Tim.

58:57 MK: Oh, says you.

[chuckle]

58:57 MH: Tim has very high hair standards.

59:01 MJ: You have to visit the US, Moe.

59:02 MH: No, just get a VPN.

59:04 MK: I will be there.

59:05 TW: Oh, great. You know what, instead of going to dinner in Las Vegas, you can go watch a documentary on Hulu.

59:09 MH: Oh yeah.

59:10 MK: Hey, I want my birthday dinner, dammit.

59:12 MH: Okay. [chuckle]

59:13 MK: Dammit.

[chuckle]

59:15 TW: Rock, flag, and senior data scientists at Theranos.

[music]

Podcast: Download | Embed

Subscribe: RSS