The Planetary News Radio – Episode 14: Impeachment and Science in America – Introducing the TruthScore

Welcome to the Planetary News Radio with your host, Bryan White. I’m here recording today outside in a little bit of rain in Corvallis, Oregon. It is December 24th 2019 so that would make it Christmas Eve, among other things. And I am recording today because I actually have time to record finally. It’s been several months since I’ve had time and a lot has happened since, not just in science news, but in the world, in politics and in life. And so I’ll start today with the big world news. 

We know that Donald Trump has now been impeached, which makes him the third president in the history of the United States to be impeached. But it’s different. It’s different than the other ones, which I suppose every [impeachment] is different. And it’s different for a couple of reasons, or at least a couple of major reasons. Probably the first biggest reason that is different is that the evidence that impeachable offenses were committed is not really disputed. We know that these things occurred in terms of the factual, physical trail of evidence and occurrence of events that happened. There’s no question that Trump had contact with Ukraine and that this series of events unfolded in the way that it did. There’s some questions as to actual transcripts of phone calls or not, which is not really a big deal. The overall intent and the procedural occurrence of what was done is very clear. This is quid pro quo, where a United States president asked a foreign country to investigate a political rival. 

I was trying to explain this [situation] while I was talking to this with some friends of mine, and I basically [used this analogy]. Imagine if Thomas Thomas Jefferson had hired mercenaries, say Prussian mercenaries to investigate Benjamin Franklin, or something like that. If there would have been a point in time where Thomas Jefferson had hired French or Prussians or Native Americans, you know, some foreign entity, to investigate Benjamin Franklin, that would have been a major historical event. I think people would have said, “Well, that was unacceptable and probably went against [the Constitution]”. I mean, I’m being extreme in the example of Thomas Jefferson, but certainly this would have gone against what we thought the founders were creating as presidential. And so I think you’ll see Trump being the least “presidential” president, at least in the modern history of the United States. 

It does [make clear] all of those areas where we kind of expect people to follow tradition, and then all of a sudden, they don’t. We’ve learned a lot about what is really law in America.The other reason why the impeachment is different or unique is that [we are] really going back and forth between a constitutional crisis. We thought there was going to be one and there wasn’t. And then there was, and then it wasn’t. And now it seems like we’re at that point again because you have the Senate basically refusing to give a real trial for Trump. And so again, the example there is, say, [the trial of] O.J. Simpson. Popular belief is that O.J. Simpson was guilty of the crimes he was investigated for, but he had a trial. There were witnesses, there was evidence presented, and he was acquitted. And so, as the defendant, obviously O.J. Simpson believed, or presented [to believe], that he was innocent. In the impeachment trial, Trump presents that he is innocent. But O.J. still had a [full] trial. [Granted] a murder investigation is very different [than an impeachment inquiry], murder trials happen all the time, [so their procedures are well established]. Impeachments don’t happen all the time, but we all agree that impeachments should have a fair trial. 

What we’re seeing now is Senate Majority Leader McConnell is basically saying up front that it doesn’t matter what happens in the trial. Trump will be acquitted. Imagine if O.J. Simpson had gone to trial and the judge announced the beginning, “It doesn’t matter what you say. All of the evidence that you present any witnesses that you call. I am going to acquit O.J. Simpson.” And so there’s no reason for a trial [if the end has already been decided]. So imagine if a judge had said that, that is the death of the justice system in America. So what we’re seeing is the Senate, as a leadership entity in the United States, is creating a precedent for a judicial entity to decide the outcome of a trial before it happens, and that is a violation of the Constitution. You’re innocent until proven guilty in America, and so we have a right to a trial. And so really denying Trump a trial is taking away his right to a trial. 

And of course, it works out in the favor of a person who knows they’re guilty to be declared innocent before the trial happens. But imagine if, in another case, say, a circumstantial evidence case. Say someone had had drugs found in their car or something like that where you really don’t know before the trial, you need the trial because you need to do the investigation. Imagine if judges began declaring someone guilty before the trial and said, “It doesn’t matter.” This is a major constitutional crisis, I believe, and this is relevant to science because we can’t do science, we can’t function as a society in the midst of a constitutional crisis. We need stability to do science, and so that’s part of my job. My job is to bring about stability and create the atmosphere that science can be done in. 

And so that’s what leads me to my next development. The major thing that I’ve been working on, which is my truth score algorithm, which when I initially thought about it, when I initially had this idea, I did think that it would be a truth score, but it’s more than that now it’s been released. It’s different, but I’ll still use the word truth score. In my reading of the primary literature I looked at the history of deception detection, a linguistic science of which the major applications are in law enforcement [and criminal justice]. In this review I found that there had been studies done where linguistic patterns associated with deception were found. And so I spent a lot of time thinking of a way to scan news articles so that I could score these articles using these metrics. 

And so I’ll talk a little bit about some of the metrics. It’s not super complicated. It seems complicated when you add it all up together, but individually, it’s not really difficult to understand. The first metric is a complexity measurement. That’s the Shannon Index, which comes from a strict information theory background. The Shannon Index is a measure of complexity. You can use the Shannon Index to understand the complexity of an ecosystem or the complexity of a sentence. Since it’s a general information theory [formula], [it can be used on] any type of information. One of the things I found is that deceptive language will tend to be less complex, and truthful language will tend to be more complex. A good example is probably this podcast, which is probably using a higher level of complexity language than might be used if someone were trying to be purposely deceptive. My goal is to inform and so I’m using robust language so that you can understand what I’m saying. If someone is being deceptive, they might be leaving out details, and that’s where you see a reduction in complexity.

That leads me to the next metric, which is another information theory metric. You could also use this on ecosystems or language: it is an Evenness index. An evenness index is going to tell you what’s the spread of the type of something in the measurement. So, for example, if you have an ecosystem that has 500 worms and one rabbit, that’s not very even because most of the population is worms. Everything’s concentrated in worms, [and so the local population isn’t evenly distributed amongst the possible species]. The language example would be, for example, a sentence that had 10 nouns and one verb. Well, that wouldn’t be a very evenly spread sentence [in terms of grammatical components], so that’s telling me something about the language that’s being used. Either it’s non-standard grammar or it’s not really a sentence or something’s going on [in terms of truthfulness]. And again, the idea is that it’s hiding the ability to fact check to understand what’s being said in the sentence [by using uneven grammar]. [What this suggests is] that sentences with higher evenness are again being less deceptive. 

The third metric is actually the closest thing to a truth score in the system, and this is actually matching the grammar of sentences to a database of known deceptive sentences. I have a database of Amazon reviews and hotel reviews where participants in a study were either told to write a fake story or told to write a true story, and so we know that we know which stories are lies in which stories were true, and these stories also have a positive or negative impact. You could have a deceptive positive review, which might be someone who’s trying to pay a writer to get his or her hotel ranked higher than others. You might have a deceptive negative review. Maybe it’s someone’s paying a competitor to down rank that competitor’s hotel. You have truthful, positive or negative reviews. Someone had a truthful negative experience or a truthful, positive experience. So again, we’re saying that in terms of science, a truthful, positive ([or neutral sentiment]) sentence is closer to the truth. In other words, scientific language tends to avoid negatives or emotional words. In critical thinking, you see a separation between logos, which is logic, and pathos, which is emotion. You want to say in scientific thinking that we minimize the pathos and maximize the logos, and ethos, which is credibility. 

Then there’s a final metric, which is a similar to the positive or negative sentiment analysis, which is the objectivity or subjectivity ranking. Another type of sentiment is objective or subjective. Using clear terms, clear language that is, again, not hiding, not using lots of extra adverbs and adjectives [that dilute] descriptive language and don’t give a clear meaning. Some words are much more concise and clear in their meanings and others. So a sentence with higher objectivity might be considered again, more truthful. The overall “TruthScore” is just the sum of these metrics. A higher complexity, higher evenness, higher truthfulness, higher positivity, and a higher objectivity sentence is going to score higher on the TruthScore, and a lower [sum of scores] will score lower overall. 

I’ve done this [analysis] now on a couple thousand news articles, and I’ve seen a very consistent pattern amongst them. I’m publishing a list of news organizations that publish feeds related to science, [and their rank via the TruthScore]. I’m seeing a [pretty clear] pattern. What you see is that, there’s a couple of organizations that are consistently highly ranked. Most notably, the BBC, NPR, WIRED, and the Guardian. Surprisingly The Daily Maily UK [was also highly ranked, which is strange because the Daily Mail] might be generally be considered a tabloid in its political news, but maybe in their science news they might have really good writing. I think that’s fascinating because that tells me people are looking at politics as entertainment. But when it comes to science, they want to know the real truth. And that is amazing to me, because the truth itself is fascinating. I want to learn more about the universe, and so it’s telling me this is good. There are people out there that want to learn more, and then you see some of the consistently lower ranked news organizations (e.g., Fox News), which we historically know has weaker science news. 

And so the question arose, “Why is Fox News consistently ranked the lowest out of 20 or so major broadcasting networks and science news?” And the [follow-up] question is, “Are they being purposely deceptive or what’s going on?” And so, so far, it looks like Fox News isn’t being purposely deceptive. They just don’t dedicate a lot of resources to science. And so the take home message there is that if you only watch Fox News, you’re not getting good exposure to science [writing]. You’re not getting good exposure to science and again, it tells me if people are okay with that, then people who are just watching Fox News are ok with having poor exposure to science. They don’t have that curiosity. They don’t have that drive to understand the world and learn new things that are true that we all agree or true.It tells me something about the[Fox News] readership, and as a [journalistic] source. So again, what does this boil down to? I would make the recommendation not to use Fox News as a source for science because I know off the bat they’re not dedicating resources to it. 

That would lead me to believe that, while they are publishing science articles, that I don’t know what the decisions are being made on which articles are being published. So there could be some bias there, and in science, you want to publish a diverse array of articles. So again, I’ll keep investigating this. And when I’ve learned more specific facts, I’ll share them. And then just one more interesting piece. It’s funny, because BuzzFeed news again is similar to Daily mail UK, you would think it is like a tabloid, but, BuzzFeed news consistently ranks higher than Fox News, in it’s science reporting. And so again, it tells you they’re people who are reading these tabloids for entertainment, they still have this desire to have science news, and maybe they see politics as more of entertainment. Whereas Fox News is presenting politics as facts and science as the entertainment, which to me, is flipped. And with that, I won’t say any more on this broadcast. So I hope you enjoyed this podcast and continue learning about science and searching for the truth. That’s Bryan White with the Planetary News signing off.