YouTube — Fix Your AI for Climate Change! An Invitation to an Open Dialogue

Image: Ganges River Delta. Image provided by the USGS EROS Data Center Satellite Systems Branch. Caption adapted from text provided by NASA’s Earth Observatory. Source: NASA/USGS Landsat 7; NASA Earth Observatory. From NASA Climate YouTube channel https://www.youtube.com/user/NASAClimate

An interview with Joachim Allgaier on his recently published study on how social media platforms such as YouTube have become hostile to climate science. When you search around climate change on YouTube the results are 50/50 climate science versus anti-science Chemtrails conspiracy theories. YouTube and Google Scholar have been strategically hijacked by groups posting anti-science content, while at the same time academia has neglected to use YouTube and recognise it as the vitally important channel for their scientific voices to be heard. The interview is a call for scientists to actively engage with the platform and for YouTube to reflect the values put forward by its CEO and Google co-founder Susan Wojcicki of an appreciation of the scholarly environment within Silicon Valley. These values could be actioned in YouTube’s AI to favor scientific factual content and by adopting Open Science practices of enhanced transparency across its platforms as anti-science Chemtrails content is also immediately found in search results on Google Scholar. Open has worked for Google’s support of its technology stack with open source, why not apply the same workings to its search indexing, dare it be said by applying a little open library science.

Simon Worthington, GenR (SW): First off could you introduce yourself and your area of study?

Joachim Allgaier (JA): My name is Joachim Allgaier. I have a background in sociology, and moved into science communication and technology studies, so quite an interdisciplinary background. To cut a long story short, I’m really interested in studying the interface between science and society. I’m based at RWTH Aachen University the biggest technical university in Germany, working at the Chair of Society and Technology at HumTec the Human Technology Center an interdisciplinary research institution.

SW: We’re here to discuss your paper ‘Science and Environmental Communication on YouTube: Strategically Distorted Communications in Online Videos on Climate Change and Climate Engineering’ published this July (2019) in the journal Frontiers in Communication, which you could boil it down to say is looking at the representation of Climate Change in social media and specifically YouTube. A good place to start would be to ask you to unpack the title of the paper, which looks like it breaks down into two parts?

JA: First of all I wanted to do something on YouTube because most of the science communications research is focused on other portals, particularly Twitter. This is because it is easier to study, you have these small segments of text and functions on the website where you can aggregate the data and sampling is really easy. Whereas, studying YouTube is somehow more challenging, in terms of methods and you also have more elements, not just text. You have: stats, audio, video, imagery, discussions, stories, and all kinds of different elements.

Also the reason I wanted to study YouTube is that the reach of YouTube is many times higher than the reach of Twitter. YouTube is available worldwide, with almost two billion users, which is impressive. We know from reception studies and user statistics from various surveys that many people say that they get much of their knowledge of science and technology from YouTube. My idea was to come up with a systematic way of studying content on YouTube so at least we could have an idea of what the people might find on YouTube. This is covers the first part of the paper’s title.

The second part of the title ‘Strategically Distorted Communications’ this is the last part of the paper where I look at the results of the sample of videos on YouTube when you look for climate related issues,  where you don’t necessarily find scientifically correct information but also a lot of conspiracy theories videos. If you look for the big climate terms like — climate change, climate science, or global warming — you mostly find scientifically correct information. If you move a bit further away from these terms, if you start to look into climate manipulation, or geo-engineering then content tagged with these terms is totally infested with conspiracy content and there is actually a strategy behind it. Importantly these are all the issues that need to be discussed pretty soon and that the IPCC is discussing as something we might need to do when we don’t manage to reach the emission targets. Somebody created a strategy, somebody was thinking about how can we manipulate the mechanisms of YouTube that people find our anti-scientific content and not the actual story that a scientist might tell you about what geoengineering is for instance.

SW: In the paper’s introduction you highlight the enormous challenge of climate change, the work of IPCC, and how climate change will affect us  — as well as the importance of communicating these issues to the public. It would be good to hear about your motivations for the interest in climate change and the need for information for the public?

JA: My interests started many years ago, you can see from the samples in the paper the work started in 2014/15. It took some time to get to the review and changes needed to be made before it could be published. Back then climate change was an important topic and it was recognised that something had to be done — especially to the scientists, but it wasn’t so pressing as it is today and not so controversial. With the introduction of the strategies of fake news and a post-truth more and more challenges (Cook et al. 2019) were mounted to the consensus on climate change. When studying YouTube I did not expect it to be so divided. I expected when looking for some terms I thought something controversial might come up and I was interested in what the ratio that might be, would it be 1/3 or 1/4 of videos that might have incorrect information, or unclear depiction of what climate change is, I wasn’t expecting that they would deny climate change. But the more I go into it, the more this clear pattern emerged and I realized there is a serious problem, not so much with the big terms, but rather with the related ones like ‘climate modification’, ‘geoengineering’, etc.

Back then I was reading the literature on conspiracy theory and I could see that various conspiracy theorists have done the same with different terms, for instance my colleague René König has written about the ‘truthers’ who don’t believe in the 9/11 tragedy and they are encouraged to look on the web for a particular term (e.g. WTC-7) (König 2013). There have not been so many pages on the web referring to this particular term so it could be taken over so that people will then be directed to the conspiracy websites. Reading the literature from the Geoengineering community, they also reported these attacks coming from the Chemtrails side. At their conferences they have activists there and the situation for them has at times got very serious. For most of us if we hear about Chemtrails we wouldn’t take it seriously, but according to a survey carried out in the US around 1/3 of the population in the US (Tingley and Wagner 2017) now believes in Chemtrails.

SW: It would be really interesting to hear about the process of the research, when it started and what methods and tools were that you applied?

JA: The research was started several years ago, but I was always interested in the question what might different user find if they really used YouTube as an information source for scientific or medical issues. Until then absolutely no work had been conducted about science communications on YouTube, with some conducted on health communication. What the researchers did there was to pick the 100 most viewed videos on a particular condition or disease and then a team of medical experts assesses the medical accuracy and said whether something is medically correct or not. This was the method used so far, but this method doesn’t address the question of what a user would find individually.

I had been making quite an effort to find out how I can study content on YouTube that is not just my own filter bubble. I wanted to know what different users might find. Eventually I found that the Tor tool could be used in the research, which re-routes your web browser, you can use different end users, IP addresses, different persona — basically each time you use the web you have a new identity. You still have filter bubbles, but they’re other people’s bubbles. I ran the searches several times, each with different personas. This is the major innovation.

Then I picked ten terms, first the really big terms like — global warming, climate change, climate science — and already back then there was the fear that the emissions targets might not be reached and that methods like climate or geoengineering would need to be used to remove greenhouse gases from the atmosphere. These fields are new epistemic communities within science and have not been applied yet. It is extremely risky technology because it can only be used on a global scale. But as soon as you type in the phrase ‘geoengineering’ into YouTube you are almost immediately directed to Chemtrails and conspiracy theories (Wagner n.d.). This is because a big effort is made in strategically occupying these terms, so that people do not find the real scientific information there, and this might have something to do with trying to undermine the trust with science as a whole.

This is also the reason I wanted to include the term Chemtrails as I wanted to see what happens if I type in Chemtrails into YouTube —  is it just Chemtrails coming up or is it also debunking videos coming up, say from the science side, or say other civil society actors who say ‘hey this is just a stupid conspiracy, don’t believe in that because there’s no evidence for it’. Because from the scientific side there are actually surveys (Shearer et al. 2016) that none of the scientists who work with this stuff actually believes in Chemtrails. From the point of view of science it is quite clear that there is no evidence base for the Chemtrails theory.

This is the reason why I picked these different terms. Then I repeated the search with different identities from different Tor users, this is where it gets opaque as well as we can’t really look into Tor, what is happening inside there. I can just see that by using it, I have a different profile now, a different end node in a different country and so on. So I have got an idea basically of what types of video and content comes up if I search for particular terms. The main question for me was, does the message of the videos stick or adhere to these scientific consensus view of climate change, or not, or is it something completely beyond that like a conspiracy theory.

SW: I wanted to ask about the viewing figures that came out of your study and what can be read from the search results of the 200 videos, with 89 supporting the consensus about anthropogenic climate change, 4 being discussions between the two disagreeing camps, and with 107 opposing the scientific consensus? But also how the number of views in each camp were almost identical, with only a difference of about 2,000 views between them — with the science views receiving 16,941,949 and the skeptic position receiving 16,939,655 views in total?

JA: If you would remove a few items like the comedian John Oliver’s videos then the climate denying content would have won in terms of views. So what we can learn from these figures is that the conspiracy theorists have done a really good job in terms of achieving their goals. Also it is mainly media content that is found supporting the scientific consensus position and very little from scientists, and so the lesson learned here is that it is the scientists and science communicators that have done a really bad job of recognizing this medium as important — being present, producing content, and taking this super influential communications channel seriously. So, when I talk with members of the science community I encourage them to make use of it as it has enormous reach, especially among young people. Many YouTubers tell me that their followers generally don’t watch TV, they don’t read newspapers and that YouTube is their prime information channels — and if you’re not present there you will simply not exist for them. And this is why I would also encourage people who are doing research on climate or geoengineering to work together with some people who know how to use this medium well, have a collaboration with a YouTuber, and get them to talk about your point of view, because if you don’t take care of this someone else will present their point of view and you simply won’t be heard.

SW: I wanted to ask about the details of the strategies at play by the climate skeptic groups looking to disseminate their content. In your paper you talk about examples like flooding, producing content from multiple sources designed to take over search rankings. Recently I attended the Video Vortex Conference about the aesthetics and politics of online video. There I learned that these groups who have an agenda to push on YouTube have sophisticated operations: not only understanding SEO and media strategies, but also being able to directly fund their operations out of YouTube’s revenue share programme. So I would be really interested to hear about a strategy breakdown of the climate skeptic YouTube channels?

JA: What is striking is how these two terms ‘climate engineering’ and ‘geoengineering’ have been targeted. If you browse around on the Chemtrails conspiracy websites you will find a call to action with the warning notice ‘not to use the term Chemtrail when posting as this is associated with conspiracy theorists and to instead use different tags, specifically climate engineering and geoengineering’. Also they chop up videos and use them in various chunks flooding all sorts of channels. This is important as its manifest evidence of the strategy. Another strategy is to attack the scientists, if you read the comments under scientific YouTube videos, especially people talking about climate change, the comments are sometimes extremely violent and aggressive. Under YouTube community guidelines they should be removed but sometimes they seem to slip through. 

If we zoom out of YouTube and look for example at Google Scholar under searches for Chemtrails you will not see scientific literature but instead Google Scholar directs you to the websites of the Chemtrails theory people. This is really horrible for Google Scholar pretending to be an academic database when it actually directs you to the most anti-scientific websites you can imagine. This is why I have become quite skeptical about Google and YouTube because I have tried to have a dialogue with Google, we have invited them to conferences, they have never really replied. Instead have just replied in relation to my work as it was picked up in the press — where Google press people have said that the sample is really old, that they have adapted YouTube, and that there is no problem anymore. This is easy for them to say and if the journalist or scientists asks them to tell them about the changes then they don’t receive an answer.

My solution would be to say we need more independent research to be able to verify whether if it is true what they say. Because I’m not willing to simply believe what they are telling us all the time because the climate skeptic videos are generating traffic and this is what YouTube runs on. Additionally some of the videos contain merchandise advertising and this points to there being an economy with its relationship to ‘fake news’ where people got the idea if you created outrageous stories people would click on them and that this is a way to make money. But this has really bad side effects for all of us and I think this is a matter of responsibility for YouTube.

SW: I started out this interview as a Google and YouTube skeptic and remain so as we still don’t have the levels of transparency that are needed to take any other position, but I appreciate that it’s a complex situation. Starting with Google, where are we at with Google and regulation and controls? Are there any controls over YouTube from the scientific community, institutions, and their supporting government bodies? The status quo appears to be that there is no regulation, accountability, or transparency. Also this problem of disinformation must be affecting a whole variety of sectors, as you mentioned with healthcare, but it can literally be anything else: a local city issue, economics, culture, or essentially anything that can be hijacked?

If we were dealing with public service broadcasting or broadcast media there is more regulatory control: gatekeepers, media organization, or parliamentary control, is YouTube being brought into those frameworks?

JA: So far they have been very good at avoiding these regulation issues, because of the international nature of this platform. The main angle by which the platform was grasped was by the copyright issue, especially in the EU. But whether content is true or not, or harmful or not are not covered unless it relates directly to harming someone, if people are asking for someone to be hurt. And then there is borderline content, which is notoriously difficult to regulate and this is how they have got off the hook, most of the time.

What is interesting is that Germany and France in the Treaty of Aachen (Jan 2019) agreed to make their own state run AV platform, a public service YouTube. It was tried before with the EU Search Engine and this didn’t fly. Even if they can get it running the question is how do they get the people to this other platform from YouTube. But we have to think about alternatives like this as YouTube has just become too powerful and too big — even if we criticize it, it doesn’t matter.

By the way have you heard of that Extinction Rebellion occupied offices of YouTube (Oct. 2019) as they really wanted to attack YouTube for spreading all this misinformation on climate change and funnily enough they referring my paper on their Extinction Rebellion website.

SW: Is there awareness of these issues in Governments. What are the ideas that are available to bring about transparency, could we have a simple things like a library catalog classification of videos, an inventory. But obviously it’s part of YouTube’s business model to not let people know what it holds and only show a temporary glimpse of what they have available AKA a search result?

JA: There is an awareness of hate speech so this is why Germany and France came up with an idea of a public platform, but they don’t know how to regulate something like this. If you use YouTube you have to completely rely on the algorithm that is gives you want you want. Their AI is one of the most powerful algorithms in use, just think of the incredible amount of data generated by the videos, by the users, the traffic, this is where the deep learning mechanisms learns. How users behave, what happens if I feed them right wing content, conspiracies, etc., do they stick watching? Back in the days when they started the whole deep learning exercise the algorithm just had to keep people watching, no matter what, even if its harmful content. Then advertisers said we don’t want to have our ads associated with extremist content and this is the reason they made changes by removing content.

SW: So maybe they need to be understood as what they are an advertising company reliant on an economy of attention, so only things affect that bottom-line mean anything to them?

JA: If you listen to the YouTube CEO Susan Wojcicki speak, she stresses that she grew up in the Silicon Valley surrounded by all these academics and professors, that she values and appreciates having all these creative minds around her. If this is her attitude I would like to see this reflected in how YouTube as a technology works: that it prioritizes factually correct information, especially about medicine and controversial scientific subjects, let’s say vaccination or climate change. It’s not a big deal to tweak the algorithm to do this, it’s just a question of whether it’s profitable for them and this is where the problem comes in.

If at some point they would actually really make that step, to say — look this is our culture, this is where we come from, we come from Silicon Valley with all these academic enterprises and so on, and we want to transport this spirit — then this would be a wonderful step. This is something I would like to hear, but unfortunately I cannot see that they act according to this spirit. This is the problem I have they seem to have lost this progressive spirit.

SW: So a concluding question would be how can researchers, scientists, or institutions engage with these questions. As on other topics you can find interesting YouTube content, I think of something like Sixty Symbols in physics and astronomy from the University of Nottingham in the UK. What would be your request or recommendation for creating content for YouTube?

JA: Scientists no matter what they are working on should treat YouTube seriously as a communications channel. They don’t necessarily need to post videos themselves, but they should consider working together with someone who knows how to deal with this medium. Normally every university or institute has a press person or a science communicator on board. Also in regards to subjects they talk about they should look on YouTube and see what content is there, if it’s good then OK, if not why not post a video that points to what is going on in their field from a scientific point of view.

In my field of study I have been disappointed to see that most of the research has been conducted on Twitter or newspapers for controversial topics, but very little on platforms like YouTube. Also there are many new visual channels, like Instagram for young people, and for really young people things like TikTok,  which uses very short clips. Now that these platforms are emerging we should immediately start monitoring what is happening on these sites, especially when it comes to conspiracy theorists and extremists content.

Thinking of people from libraries, archives, or scientific academies, why not have something like a label for videos or channels where it could say this is scientifically approved and this is a good channel if you want to know about topic X, so you have an indexing with a scientific quality mark and this could be on YouTube. Many universities have YouTube channels but they are mainly used to distribute image brochure videos and by doing that they are not really educating people but instead they are destroying their reputation and people will not watch this and think it’s not worthwhile, this is just advertisement for the institution, I don’t learn anything and why should I go to that site. Instead they should think about different types of collaborations, with other institutions, or with YouTube to reach their followers.

SW: On the labelling for scientifically approved content I think you’re totally right something should be worked out and there are in existence already an array of tools that could be combined to index content or provide a label: for adding persistent identifiers with DOIs, personal and institutional identifiers with ORCiD, blockchain cryptographic IDs, as well as web crawlers such as the prototype My Research Institute that can go out on behalf of an institution and harvest content from a variety of platforms like YouTube and other platforms.

Links

HumTec (Human Technology Center) – https://www.humtec.rwth-aachen.de/ Twitter @RWTHHumTec

Tor Project – https://www.torproject.org/ 

Video Vortex Conference – https://networkcultures.org/videovortex/ 

Aachen Treaty, Treaty on Franco-German Cooperation and Integration – https://en.wikipedia.org/wiki/Aachen_Treaty

Physics and astronomy from the University of Nottingham in the UK – Sixty Symbols, YouTube Channel  – https://www.youtube.com/sixtysymbols

Extinction Rebellion –  https://rebellion.earth/2019/10/16/extinction-rebellion-uk-mothers-and-babies-and-youth-blockade-google-and-youtube/

My Research Institute (web crawler academic prototype) – https://myresearch.institute/about/

References 

Allgaier, Joachim. ‘Science and Environmental Communication on YouTube: Strategically Distorted Communications in Online Videos on Climate Change and Climate Engineering’. Frontiers in Communication 4 (2019). https://doi.org/10/gf8rst.

Cook, John, Geoffrey Supran, Stephan Lewandowsky, Naomi Oreskes, and Ed Maibach. ‘America Misled: How the Fossil Fuel Industry Deliberately Misled Americans about Climate Change’, 14 October 2019. https://www.climatechangecommunication.org/america-misled/.

König, R., 2013, „Google WTC-7“ – Zur ambivalenten Position von marginalisiertem Wissen im Internet, in: Anton, A., Schetsche, M. und Michael, W. (Eds): Konspirtation. Soziologie des Verschwörungsdenkens, Wiesbaden: Springer VS, 203-220. https://link.springer.com/chapter/10.1007/978-3-531-19324-3_9 

Tingley, Dustin, and Gernot Wagner. ‘Solar Geoengineering and the Chemtrails Conspiracy on Social Media’. Palgrave Communications 3, no. 1 (31 October 2017): 1–7. https://doi.org/10/ggcpgh.

Wagner, Gernot. ‘Chemtrails Aren’t the Geoengineering Debate We Should Be Having (Because They Aren’t Real)’. Earther. Accessed 11 November 2019. https://earther.gizmodo.com/chemtrails-are-not-the-geoengineering-debate-we-should-1825171856.Shearer, Christine, Mick West, Ken Caldeira, and Steven J. Davis. ‘Quantifying Expert Consensus against the Existence of a Secret, Large-Scale Atmospheric Spraying Program’. Environmental Research Letters 11, no. 8 (August 2016): 084011. https://doi.org/10/ggcpgd.