Value Driven Data Science

[00:00:00] Dr Genevieve Hayes: Hello and welcome back to Value-Driven Data Science, where data professionals become strategic experts. I'm Dr. Genevieve Hayes, and I'm here again with Professor Rob Hyndman, co-author of over 200 research papers, more than 65 R packages and five books on time series forecasting, and one of the world's most influential applied statisticians.
[00:00:27] Last week Rob and I discussed why traditional statistics still matters in the age of ai. Today in this value boost episode, we're exploring how selectively giving away your work for free can be one of the most powerful strategies for building authority and influence as a data professional. Welcome back, Rob.
[00:00:48] Prof Rob Hyndman: Hi, Genevieve. Good to be back.
[00:00:50] Dr Genevieve Hayes: Now Rob. Your textbook forecasting principles and practice is one of the most influential textbooks on the topic. The online version receives over 25,000 page views per day. The our packages you co-authored have been downloaded over 144 million times since 2015. If you received $1, every time someone made use of your work, you'd be a very rich man.
[00:01:19] Yet, you've generously given this work to the world for free. This approach has clearly helped position you as a global authority in forecasting. However, was that your intention when you first decided to open source your work or was making your work freely available?
[00:01:38] Nothing more than an act of pure generosity.
[00:01:42] Prof Rob Hyndman: I, no, I don't see it as pure generosity. I'm paid as to be a researcher, like I have a job. I'm a professor of statistics. My job is to do research. The area of research I work in is methodological statistical research. So I develop new methods, new algorithms, new models in statistics, and I write papers about those ideas which get published in journals.
[00:02:06] But if I left it there, a few people might cite the papers. Almost nobody's going to use the models or use the algorithms or use my new methods unless they also have some easy way to implement them. And so I decided right at the very beginning of my career, when I was still a PhD student, that if I was gonna do this type of work, I also needed to make.
[00:02:32] Implementations available so that people could use it. Otherwise, what's the point? I really see no point in writing papers that don't actually get used. So I decided to start writing some software and sticking it on my website. So I finished my PhD in 1992.
[00:02:50] The web was invented around 1993, which was perfect timing for me to start putting stuff on a website. And people started to use it. So I'd come up with a model for forecasting, for example, and I'd try it out. I'd write a paper about it. I'd stick some code on my website, and people started.
[00:03:10] Downloading the code and using it. And over time I've just got a little more sophisticated. I now package it up into our packages. It appears on cran and I maintain them, and I'm a much better coder than I was back in those days. So the software is much better, but it was really just about trying to get my work used rather than finish the process in writing a paper.
[00:03:31] Dr Genevieve Hayes: There's a joke among academics that if you write a paper, only 10 people ever read it, and two of those are your parents.
[00:03:40] Prof Rob Hyndman: Yeah. And I see a lot of academics do that. Like they think that the paper is the end of that line of work and I just do not understand that I'm not in this business to write papers. I'm in this business to come up with new ideas that hopefully change the way people work. So that's why I've done open source software.
[00:03:59] It's also why I ended up doing an open access book. I have two commercially published textbooks before I did anything open access. One of them was an undergraduate textbook, which I used for my own teaching. But it was published through a commercial publisher and they.
[00:04:16] Had a ridiculously high price on it. It was about $150. This was 15 years ago. And of that $150, you'd think, I'd be getting some money for that. I'd get about $1 for every book they sold. Even though the contract said I was getting, I think it was 5%. Never was, because there was all these other clauses that meant they didn't pay for this or didn't pay for that.
[00:04:39] So I wasn't making money off it. And the students weren't buying it, it was too expensive for them. So they would hop online and try to find some information, usually from some dodgy website full of total nonsense. And I thought what's the point? I've had this book that they don't buy and they don't use, and even if they.
[00:05:00] Did buy it. It's not about money anyway. I might as well write a book that they can access for free. So I decided to update my old book after some arguments with the commercial publisher about me doing a competitive book which I eventually managed to win the legal argument. I stuck it online for free.
[00:05:19] And that's gone through multiple additions and it basically. Has become the textbook in forecasting because it's free, it's open, everyone can use it. And there's more than 2000 instructors around the world who now use that for their forecasting class. Presumably there's some other books that get used in some classes, but it's extremely dominant.
[00:05:39] Dr Genevieve Hayes: Did you expect your work to be as popular as it's been?
[00:05:42] Prof Rob Hyndman: No. Not at all. It was very surprising to me both the popularity of the art packages and the popularity of the book. I would never have imagined that.
[00:05:52] Dr Genevieve Hayes: How has giving your work away for free helped shape your influence as a statistician?
[00:05:58] Prof Rob Hyndman: So it's had lots of great side benefits. So for a start, my papers get cited far more highly because there's open source software available that implements them. So somebody is looking around for some method and they go here's a method. I can try this out. And then they write a paper and then they will cite the paper on which the package is based.
[00:06:18] So my citations for the methodological papers a much, much higher than they would ever have been if I didn't give away the software that was related to them. Secondly I get a lot of requests for consulting or for speaking at conferences or for doing various things that academics like to do because I'm known and I'm known because of the free stuff that I give away as much as anything.
[00:06:42] Dr Genevieve Hayes: Do you do training workshops on your free packages?
[00:06:45] Prof Rob Hyndman: Yeah. So for the last. 15 years or so. About once a year. I'll do a workshop over two or three days where we do some training around the packages. And that's mostly aimed at people in business and industry who are forecasters who need to upskill. So we teach them a little, basically, it's a shortened version of what I teach my third year undergraduates.
[00:07:06] Dr Genevieve Hayes: Obviously there's a limit to how much you should give away. You do still have to pay for things like food and electricity. How do you decide what to share freely and what to hold back?
[00:07:17] Prof Rob Hyndman: I'm fortunate that I have a full-time salary from Monash University, so I don't need an income from software or books. 'Cause I have an income, so I. Give everything away for free. That I produce. As long as I can. I think that's a good strategy. But it's only possible because I'm fortunate enough to have a job as a researcher.
[00:07:40] Dr Genevieve Hayes: If you didn. Job as a researcher, how do you think that might change things?
[00:07:44] Prof Rob Hyndman: So I'd have to find some other way to earn income, which might be through book sales or it might be through. Running training. I still think open source software is the best model for software. But you can run training around that and charge people for the training and people do that. I imagine I could make a living that way if I wanted to, but I'm actually pretty happy doing what I do.
[00:08:04] Dr Genevieve Hayes: Not everyone has a textbook or an R package to give away. For a data scientist who wants to build their authority by sharing their expertise, what other forms might that take?
[00:08:16] Prof Rob Hyndman: You could start a podcast like you blogging or a YouTube channel or there's lots of things people can do to bring attention to what they're doing and to try to promote a certain way of. Thinking around the topic. So for me it was, I have views on how forecasting should be done and I wanted to try to get those views across.
[00:08:35] And I've done that through software and through papers and through my books. I have not done podcasting or YouTube channels. But that would definitely be other ways that you could get messages out.
[00:08:48] Dr Genevieve Hayes: Do you think the type of work you give away matters, for example, is a higher effort output, like a textbook, more valuable for authority building than say, smaller, more frequent contributions like LinkedIn posts, for example.
[00:09:04] Prof Rob Hyndman: Or if you can write a textbook and no one reads it, and then it's not really contributing much to your authority. So it, you need to build. A bit of a reputation and then they feed off each other. My books get used because people use the software and they recognize my name because of the papers.
[00:09:22] There's a whole lot of feedback going on with different things that I'm doing. And it takes time I finished my PhD in 1992 and I've been doing this type of work ever since then. So that's a long time to build up a reputation. And definitely for the first 10 years or so, probably no one in forecasting knew who I was.
[00:09:42] Dr Genevieve Hayes: So it's about starting that virtuous cycle and then it compounds over time.
[00:09:47] Prof Rob Hyndman: Yes. Yeah. I.
[00:09:48] Dr Genevieve Hayes: So for a data scientist who wants to start building their authority and starting that virtuous cycle, where would you recommend they begin?
[00:09:57] Prof Rob Hyndman: What have you got to offer? What are your ideas? What are your distinctive things that you think you understand something better than other people and you'd like to get that across? And then think about how am I going to best share that? For the most impact. And then see what happens.
[00:10:14] So everyone told me when I was either an undergraduate or a PhD student, I needed to get overseas. No one was gonna take any notice of me if I'm living in Australia. And that wasn't an option to me for family reasons. So I thought how am I going to build a reputation when I'm not in the main places where stuff's happening in North America or in Europe?
[00:10:35] And so I thought there's this new thing called the internet. I'll set up a website. What do people want? I thought maybe people looking for data for their teaching. So the very first thing I did that attracted any attention at all was I just collected about a hundred time series and stuck them on my website and called it the time series Data Library.
[00:10:54] And I made it available for every other academic in the world to use for their teaching. And they did, and I would go to conferences. I had not written a paper that was widely known at that point. I definitely hadn't written a textbook that anyone had heard of. And they'd go you are the guy with that website.
[00:11:11] That's fantastic. So it was just something that sort of got the ball rolling. That still exists, and nobody much uses it anymore, but they use other things.
[00:11:21] Dr Genevieve Hayes: The first thing I ever made public was after I finished my PhD, I started writing the puzzle page for Actuaries Magazine and I would go to conferences and people would walk after me and wanna shake my hand because I was the puzzle
[00:11:36] Prof Rob Hyndman: Yeah. Yeah just find something you can do and do it and see what happens. And then after a while, some other opportunity will come along or you'll get some other idea about something else that you might be able to work on.
[00:11:45] Dr Genevieve Hayes: Yeah, opportunities give rise to future opportunities.
[00:11:48] Prof Rob Hyndman: Exactly.
[00:11:50] Dr Genevieve Hayes: Okay. So that's it for today's conversation with Rob. If you haven't already, listen to our previous episode where Rob and I discussed why classical statistics still belongs in every data scientists toolkit.
[00:12:04] You'll find it at value driven data science.com or on your favorite podcast platform. Thanks for joining me again, Rob.
[00:12:12] Prof Rob Hyndman: No problem. Thank you.
[00:12:13] Dr Genevieve Hayes: And for those in the audience, thanks for listening. I'm Dr. Genevieve Hayes, and this has been Value-Driven Data Science.