Welcome to the Planetary News Show with Bryan White. The date is May 24th 2019. It’s about 2:38 PM, [and I’m] broadcasting today from Corvallis, Oregon. And this is the first episode, or at least the first episode in this format that I am attempting to record, so I don't have any show notes or anything planned for this first episode, but I do want to talk a little bit about what is the planetary news and what is the plan and why am I doing this.
So my name is Bryan White. I'm formally trained as a biologist. I have a bachelors and a master's degree in biology, and really, I just have a general love of science. And a lot of this came from my father, who was a middle school science teacher when I was growing up. Now he had multiple other careers. Before that, he was in the Air Force. He worked as an engineer in the aerospace industry. He was a lawyer. Later in his life, he was a pilot and he was training to be a flight instructor. But when I was growing up, he was a middle school science teacher, and that influenced me very heavily. He involved me a lot in the struggles that he had teaching science, making the curriculum and things like that. Controversies with evolution, those types of things.
So I grew up in that atmosphere, and that really is what led me to become a biologist was my dad's search for understanding evolution, because any time he would teach an idea, he himself wanted to understand it all the way. And so he really wanted to understand evolution and understand why it was so controversial. And so that led me to pursue education in biology, specializing in evolutionary biology, and I was very successful at that. On the research side, I worked in environmental genetics and human genomics for a number of years, and I really enjoy doing that. But, my real passion is kind of that same science education passion that he had. And so, while I love working in the field of genetics, I love to share science with other people.
I love to teach when I can. I've never been formally employed as a teacher, but I've done tutoring and taught as a grad student, and I just I just teach in general all the time with my friends and things like that. So I view myself as a teacher, sort of a Socrates of the sorts. And so that's why I'm here today, as a teacher, but also as a learner. So I am still learning. I'm learning new science, I'm learning new business, how to do a podcast and things like that. And so I'm learning and I'm teaching at the same time. And I think that feedback mechanism is important.
And so that's a little bit about me a little bit about why I like science. So why the Planetary News? Why am I doing this project? Well, so a lot of this this idea originated out of the fake news controversy that's been springing up since 2016, and I really see the need now more than ever to combat things like pseudoscience and fake news when it involves a scientific topic. Climate change, to me is one of the biggest ones. Right now it's probably one of the biggest threats to humanity and is one of the biggest areas of targeted misinformation. But what I really want to do is to be able to prove that, and not only prove that, but identify specific examples [of misinformation] and counteract those examples. So that's where the idea for this broadcast originated from.
And so I will talk a little bit today about how I plan to do that. So one of the things I could do is I could go, you know, I could interview people, which is something I might do later. Or I could go and find articles. I could spend a lot of time watching TV, finding things that I think your misinformation and things like that. But really, this day in the Internet, misinformation is so prevalent that I don't have time to do that in a way that would be effective. I would need a team of 20 or 30 people. An entire newsroom dedicated to researching, finding, identifying misinformation, and I'm not saying that I couldn't do more with that. But as only one person, I need a way to do this effectively. And so I'm developing a set of algorithms for analyzing texts and written text and language to identify sources of this information, and that's really my goal is to find the misinformation at the source. Find out who's repeating misinformation, identify the topic that is being repeated, and offer counterclaims. So when say, a new climate study is published, and there's a misinformation campaign against that, sort of a rebuttal to the science, I wanna have a rebuttal to the misinformation prepared already or as quickly as possible.
So that's the goal with the algorithm with the news, and so but how can I do that? So it turns out there's a really interesting body of literature out there in the scientific literature, and so I guess the easy way to call it would be the deception detection literature. And a lot of this originated from, the early, as far back as the seventies when law enforcement and legal, like lawyers in court, court proceedings and things like that needed a way to identify deception. So there was a lot of studies done on non-verbal cues for lying, and I'm not an expert at this, so I'm not going to claim to be, but a lot of these, nonverbal cues are effective queues at detecting deception.
Now the issue is that's great if you can see the person. So if someone's talking and you see they have a tick or something nervous and nervous emotion, they might make they look their eyes look in a certain direction, you might be able to confirm whether or not they're telling the truth, or at least whether they're accessing their short term memory of their long term memory. So you could have all these by seeing a person, so seeing a person is much easier [for] detect[ing] deception. But if all you have is written text, you lose a lot of those nonverbal cues that we [the observer] can't control. So in other words, the average person hasn't trained themselves to control their eye movements while they’re lying, [but] some people might have. And so some people might be very effective liars in public and be able to deceive a lawyer or law enforcement, or forensic expert. So that's where you have things like the polygraph test, then, where even if you've trained, it would be very difficult to trick a polygraph test - given that now, on the other end of the spectrum, you're collecting too much data, perhaps, and you might have false positives. So a polygraph has its own set of problems.
Now again, back to the Internet. So looking at only written text, you do still have something [in terms of linguistic information]. You have language, so you have grammar, so you have linguistic cues. And so there's been a lot of work on this in the last 20 years and now recently it's been escalated because of this just huge surge of fake news and a lot of it probably able to be generated at such a fast rate because it is also being done with the assistance of computer algorithms. So we need an algorithm to fight the algorithm. That's when I'm working on and not just me, other people, but I'm doing it in my own way, since my background is genetics and bioinformatics.
The way that I'm, approaching the problem [of fake news] is from a genetics and informatics approach, so I'll just draw some examples here. So this is from a study. Let's see, this paper is from a contributed article <a href=“https://cacm.acm.org/”>Communications of the ACM</a>. I don't have the full name of the journal. The title of the article is <a href=“https://m-cacm.acm.org/magazines/2008/9/5312-following-linguistic-footprints/abstract”>“Following Linguistic Footprints: Automatic Deception Detection and Online Communication”</a>. And this is just a brief article and its a summary [published] in 2008. So some examples of the linguistic cues, quantity, the amount of information in text messages that might give you a clue. Deceivers might tend to use more words and sentences, so they're they're throwing out a lot more since they might be trying to withhold information. They may be trying to make it seem like they're actually giving a lot of information, but really, they're just a lot of empty sentences. And so that kind of ties into the next one, [which] is word diversity.
So the ratio of unique words in a sentence with deceptive messages will have lower lexical content - diversity. Um, they'll also be less complex. So more sentences that have less diversity and are less complex. They might have greater expressivity, they may be using more emotional words, use fewer self references, more informal words, and having a higher cognitive complexity [overall]. Which [is] interesting because they have less grammatical or content complexity but have higher cognitive complexity. And the way that I think about that is the example of Scientology. And so, if you read, open up a Dianetics book, it will clearly state in the book that you have to read everything and understand it. And so I could imagine a person throwing out tons of kind of low, complex, empty, passionate sentences, but that you have to read every single one of them and keep them in memory and refer back to each sentence [to validate your understanding]. So that would have a higher cognitive complexity overall for the document.
Perhaps, you know, maybe you have 20 sentence article might have a higher cognitive complexity because it requires you to accept their ideas, and their ideas are spread out sparsely through the document. The other part of this is time. [Deceptive messages tend to take longer to construct]. And so since I'm looking at news articles, there isn't really a time factor. Since it's published, it's been edited and reviewed and things like that. But on the another example, the time issue might be example would be like read it like a forum or something like that, where you have people responding to messages. And so if they respond very quickly, well, then they're probably telling the truth because they already knew the answer, or as if they are taking time to respond. It could be they need to you think more about how to create a deceptive message. But on the other hand, they could respond really quickly with really low, complex, empty sentence.
So if you don't have any other information and someone's responding very quickly with low, complex sentences, that could be a deceptive message or if they're taking a very long time with very high, complex message it could be that in order to construct that high-complexity messages, it takes more time for a deceptive statement whereas a person telling the truth could very easily, very quickly create a high complexity sentence. So that's kind of the idea for some of the algorithms I'm working on, which are in beta testing. But I'm going to start using that and kind of see where it takes me. That's kind of my approach to data analysis, and so this will be news, but also be a [discussion about the analytics and informatics algorithms I’m developing] because I will be developing the software to produce to run analyses [to support the reporting of science news]. So I hope this is entertaining. Maybe not very exciting right now, but potentially [it will be]. And as I uncover more, this could be exciting. And so I hope that people listening to this will subscribe on [their favorite podcasting app], [and] like I said earlier, [this] is a learning and teaching journey. So thanks for listening. That's Bryan White with the Planetary News signing out. Goodbye.