Timo Denk's Blog

Influencing People (Essay)

· Timo Denk

“The creatures outside looked from pig to man, and from man to pig, and from pig to man again; but already it was impossible to say which was which” writes George Orwell in his novella Animal Farm, published in 1945. I came across this book because my compiler construction lecturer mentioned it: “If someone wants to really understand democracy”, he said while taking a short break from LL1 grammars and Look-Ahead LR parsers, “they should read the book Animal Farm”. I ordered and read the book, and understood why he had made the recommendation. The story goes that animals live on a farm on which humans treat them badly. The animals revolt, take over, and establish their own hierarchy… I don’t want to give away too much. But in essence the book figuratively shows how precious and fragile a democracy is.

Democracy has been important to me ever since I remember. Presumably because I have consistently been told that democracy is the “best known form of organizing a large group of people”. That might have happened through people making it clear that it’s an “absolute obligation to participate in elections” or simply in school where we talked about our democracy and its importance. It surely did not happen because I sat back and objectively approached the topic of organizing a state and thought about all the possible options.

Now when trying to empathize with Orwell, one could ask what he had on his mind in 1945, when he wrote the book mentioned before. It was probably a strong feeling of “something like the events leading up to the Russian Revolution of 1917 should never ever happen again”. I speculate that he thought about how he himself could have a share in preventing such political development. And then he wrote Animal Farm. He saw the possibility of sharing his thoughts in the form of a novella, a short novel which shows in a tangible way how fragile political systems are. People must feel it, not just hear “democracy is important, better protect them”. With this undertaking he was (and still is!) influencing people. He wrote the story in such a way that my lecturer would – almost 70 years after Orwell’s death in 1950 – still spread the word and thoroughly recommend the book to a room full of students.

I’d like to zoom out and take a moment to appreciate the difficulty of the task of influencing people. Suppose it was on you to bring it to people’s minds that, say, eating animals is wrong. Whether this is the case or not is not the subject of this essay, but it serves as an example here. Try to really put yourself into the situation of someone who wants to make a number of other people think that the statement is true. You will notice that it’s not an easy task. Probably an effective way is to record a dramatic documentary about the brutality of animal slaughter (lots of effort, potentially having a somewhat high impact). One could alternatively consider sharing some pro-vegetarian posts on social media. With a bit of luck the posts convince one or two people to become vegetarian. But making millions of people change their eating habits is not easy at all. Approaching such a challenge in a systematic manner requires a specific kind of intelligence, which I’m keen to understand better.

We like to complacently compare our Western freedom to the more autocratic Russia or China. In doing so we often fail to zoom out a bit and acknowledge that our freedom is also limited. For example, we all go through the same education system. It is a form of streamlining. In school we learn that “we are free”, but do we ever seriously discuss the disadvantages of topics like democracy or immigration? I did not during my school time. And I believe that’s good! Because it aids the stability of our entire system. We limit our mental freedom in order to have a stable society without tumult and revolutions. With a broad support for democracy and no discussion of alternatives, people do not come up with ideas like taking over the Bundestag to establish a new order. Our school system is part of a mechanism that ensures this order prevails.

The education system that we go through influences the entire society significantly. I see that also as the main argument in favor of a federal (rather than a nationwide) education system as we have it in Germany. Modifying what is being taught in school is much harder that way. It is purposely set up in a way that the control over this influential tool (education) is distributed. A newly elected chancellor cannot just cross out the education about the cruel Holocaust, for example, and thereby influence an entire future generation. This gives the school curriculums stability and resilience. Even though the federal setup is often being criticized because of its more immediately evident disadvantages.

In school we are also being told that it is important to express one’s opinion and that the right to demonstrate peacefully is a fundamental one. I find the act of demonstrating particularly interesting to analyze in this context. If one decides to participate in a demonstration one commits to spending several hours of walking (and shouting). The contribution or influence that one has is extremely small, but all people combined sum up to a mass that can surely have an effect. I see this kind of activity as positioned at one extreme end of the spectrum of ways of having influence. It is the little (mental) effort, little influence end. On the opposite side (high influence and large effort) is probably something like the creation of a book, so well-written that it captivates numerous readers and leaves a lasting mark on them. Yes, we’re back to Orwell, albeit he is just one example. Becoming a political leader would be another one.

Is there also some tool in the low effort, yet high influence regime? Hard to say. The closest to it are probably social media influencers. As the name suggests, influencers are also influencing. Without knowing anyone famous in this area, I assume that many stumbled into their influencer role rather than actively thinking about how they could have an impact. Sometimes seemingly randomly people end up with reach to an audience larger than the population of an entire city. That’s a new form of influence which is made possible by the internet and platforms built on top of it. Probably often overlooked is how much power the providers of the platforms have, on which the influencers operate. I mean YouTube, Instagram, Twitter, and such. Adjusting recommendation algorithms slightly in one influencer’s favor decides over who gets followers and reach – probably more so than the content itself.

In the algorithmic field I am fascinated by one specific idea in this context. Let me start by going back to Orwell. His book was captivating my lecturer. I enjoyed reading it too and so did many other people. Orwell said he wrote it “to fuse political purpose and artistic purpose into one whole” (Orwell, George (2014). Why I Write). People are receptive to different things though. And people with a different cultural background, level of education, etc. might have enjoyed a different version of the book. They might have wanted it to use easier language, use more figures, or be more formal. The point is, the book is written as one single piece of work for everyone out there to read it. It disregards the reader’s personal preferences, their personality, expectations, and background. What would happen if one was able to write, say, five versions of the book? They would all convey the same message, but do so in different ways, targeting different audiences. Wouldn’t this be even more impactful and influential?

You might see where this is going: Instead of writing five books by hand (which requires high effort), one could automatically generate one single book per person (!) with a computer algorithm. The buzz word that I hear too often in the wrong place seems to fit here: it is artificial intelligence. Suppose one was able to develop a model which, given a message (for example the book written by Orwell) and a person’s context (their level of education, age, interests, etc.; basically everything about them) would generate a specific version of the book tailored exactly to them. The book’s wording would differ from Orwell’s book, but it would convey the same message. It would differ in such a way that it has maximum appeal to the reader and influences them the most. The utility of the model would be to influence people and it would be enormously powerful.

One could spin this thought further and say the personalized model output could just as well be of a different modality. It could generate a music piece, a picture, or a video as a means to influence. And the effect it has on the person for whom it was generated could reach beyond societal topics (like the ones mentioned, i.e., growing faith in democracy or convincing to become vegetarian). The model could be given the task to output something that makes a person smile, gives them goosebumps, or it could write this gripping motivational speech.

Does such a model exist? No. Do I know how to build it? Surely not. Is the idea fascinating? Yes. Is it similar to artificial general intelligence (AGI) – a problem studied by many ambitious machine learning labs in the world? Yes; at least it feels similar to AGI so it’s probably very challenging to discover a solution. Observing how humans solve the task of influencing each other seems like a good starting point.