Wednesday 4 December 2013

Cereal Box Thesis

[This post is intended to be a miniature, popular version of the thesis paper I'm currently writing for my MA. I began working on it 3 months ago. My paper is a work in progress and this post will not reflect everything in the finished product.]


Effective altruists don’t just want to satisfy moral guidelines and do some good deeds. They want to maximize the amount of good they can do over the course of their lifetimes. Some effective altruists think this is morally obligated, but others just think it’s the most exciting puzzle one could possibly try to solve.

As one might expect, there’s a lot of uncertainty about how to maximize the amount of good you can do. For one thing, the game plan we decide on is going to depend on our ethical views. Subtle changes in our moral views can drastically alter the effectiveness of a cause.


Suppose Adam is a classical utilitarian – he believes that “good” and “bad” are determined by weighing the costs and benefits of an action against each other. He realizes that hundreds of millions, if not billions, of the world’s population live in very poor conditions. So he decides to spend the rest of his life making lots of money and donating the surplus to these people.

Another utilitarian, Betty, believes that intelligent animals should fall into our ethical sphere of equal consideration. If an animal can feel pain and depression then it carries a relevant amount of utility for us to need to be careful how we treat it. Most of the world’s sentient creatures are non-human. Betty researches animal suffering and finds that the ones raised on factory farms live lives not at all worth living. She discovers that the ones in the wild are likely to regularly experience running for their lives, fighting for their lives, starvation, injury, and disease until, at last, they are eaten alive. We now have reasons to prioritize animal suffering over human suffering.

But wait. Carlos thinks that just because the lower animals fall outside our sphere of equal consideration, doesn’t mean they fall outside our sphere of any consideration. There are a billion billion insects in the world, as well as invertebrates and other less intelligent animals. Most of these species lay thousands of eggs in expectation of only a few offspring reaching adulthood. These creatures only experience a fraction of the pain that more intelligent creatures are capable of experiencing and they can’t experience psychological trauma, but many of them can feel some kind of pain. Considering the farcically high number of these creatures out there, the total amount of suffering on Earth might be dominated by insect suffering. Carlos makes this his focus.

But it all depends on whether one even decides to take insect suffering seriously. I mean, one insect is so much less important than one human and who’s to say that a million little insect pains are capable of “stacking up” to equal a single big hunk of human pain?

Thus our three utilitarians have different priorities because of ethical differences. But ethical differences aren’t the half of it. Even if two people can agree on what to value, they still might have different ideas about how to acquire the things they value. Say you and I both want something like “peace,” “freedom,” “equality,” and “happiness.” I might think the best way to do that is to go into politics and create policies right for human flourishing. You might think that’s a waste of time and that you can do 100 times the good by just making lots of money and donating it directly to the poor. Someone else interested in the same values might decide the best way to have a mass positive effect on society would be to find a cure for cancer. In order to make the right decision, we need to understand several fields of knowledge and understand how they interact with each other. If we overlook just one crucial piece of information, we might spend our lives optimizing in the wrong direction. 


There are four major focus areas of effective altruism: global poverty, animal suffering, existential risks, and meta-effective altruism. Each of these focus areas represents a cause or many causes that have been hailed as the most important thing one can possibly do with one’s time. Of course, the truly optimal use of one’s time will be divided among multiple endeavours, but effective altruists generally pick a single cause to be their major focus.

People that prioritize global poverty reduction think like Adam. They see millions of people struggling and think, “this extra bit of money I have won’t do me much good but it can change an extremely poor person’s life. I might as well give it to the poor person.” Adam then extends this concept to all the extra money he has and gives 10% of his annual salary to charity. He specifically chooses a charity that is highly cost-effective so that he can get the most bang for his buck. Remember, he isn’t trying to do some good – he’s trying to do as much good as possible. So if he can save 1 life or 100 lives with the same amount of money, he will donate to the charity that saves 100 lives.

Betty believes saving humans is a great thing as well, but thinks that, due to the staggering numbers of suffering animals and the incredibly poor treatment that some of them receive both by humans and by nature, that we can do more total good by prioritizing non-human animal needs.

Others agree that there’s a lot of suffering in the world today, both among humans and non-humans, but think focusing on present suffering is ignoring the big picture. In the big picture, the present is just one generation out of potentially countless future generations. These effective altruists believe that it’s more cost-effective to focus on catastrophic risks that threaten to greatly curtail or even wipe out the future of humanity. If an asteroid is set to hit Earth in one week, our first instinct wouldn’t be to donate money to starving people in Africa – it would be to Stop That Damn Asteroid! There’s not much point trying to solve the world’s current issues when there’s a future issue coming that might wipe us out for good and render our hard work moot. Sometimes the risks they worry about seem terribly remote – but because the stakes are so high and they involve so many billions of people, they still have what’s called a high expected utility. This group worries mainly about technological advances that threaten to have disastrous consequences if not properly designed or controlled.

Lastly, there must be a place left for talking about effective altruism, movement building, establishing the philosophical foundations of the movement, surveying the effectiveness of various causes, making useful recommendations to other effective altruists, etc. Spreading effective altruism is one of the most cost-effective uses of one’s time.


In my MA thesis, I’m investigating the possibility of a fifth area’s being able to rival these other areas in cost-effectiveness. I’d estimate that there’s only about a 20% probability that this fifth area actually belongs in this class, but because of how underexplored it is and how many people it could potentially help, I think the expected utility of researching it is high. [04/02/14 - EDIT: As I continue to research this area, my estimated probability of it rivalling the other four focus areas is lowering. I still think it's a useful area to explore and improve, however.]

The area I’m targeting is the cultural landscape. Now, worrying about cultural issues is nothing new, nor is worrying about how media content influences those issues. But nobody has ever tackled this field from a strictly effective altruist viewpoint of wanting to maximize the world’s utility. Most people are quite satisfied when a piece of media does more good than harm, but few people ask for anything more than that. What would happen if we developed a better understanding of how media content affects people and then used this knowledge to optimize cultural output?
I'm especially interested in the persuasive power of fiction to alter viewers' beliefs and attitudes, as I think that's the most obvious way for a work of media to have mass effects on society. If we learn about how specific kinds of effects on audiences are generated, we can optimize our content by designing it so that specific kinds of consequences are created.
Someone’s always whining about how a certain kind of media is having +Bad Effects+ on society. If we properly understood the relationship between media and culture then why wouldn’t we use media to engineer +Good Effects+ on society?


The main reason to prioritize the optimization of the cultural landscape is the fact that resolving cultural issues will foster the resolution of most other issues. It's the same sort of thinking that might lead one to prioritize education.

Adam can spend his whole life donating 10% of his salary and he’ll wind up donating a finite amount, let’s say $300,000. If over the course of his life, he convinces one other person of equal wealth to follow his example, he’ll have doubled the amount of money going to charity. If he could broadcast meta-effective altruism to a wider audience, then that would multiply his impact many times over.

The fact that people don’t pay enough attention to the world’s worst off individuals is a cultural problem. Speciesism, the belief that humans are superior to non-humans on the sole basis of species, is a cultural problem. The fact that only a handful of the world’s leading scientists are thinking hard about existential risks is a cultural problem. The fact that meta-effective altruist messages don’t permeate or interest the mainstream is a cultural problem.

Raise the general level of rationality and knowledge in a culture and watch rates of speciesism, racism, sexism, homophobia, bias, and hostility drop. Watch more people donate to charity, more people take animal suffering seriously, more people research existential risks, and more people take initiative in making the world a better place.
The media is the battleground on which cultural battles are fought. Due to the dominance of the media in people's lives on a mass scale and the amount of time people spend per day making sense of media content, we should expect the media to play a strong role in shaping cultures - which indeed, it does.
There's also the fact that so much money and power are invested in the global media corporations. They rank among the most powerful and richest corporations in the world. These corporations are cavernous potential utility mines. 


There are many possibilities for using media to do good in the world. One approach is to identify changes in the content of the media that would benefit the world. Such an approach might include the optimization of artistic output, engineering the work such that it leads to the best effects on viewers that we know how to create. The content approach also encompasses statements such as "We should have less fiction programming and more non-fiction programming," "It is beneficial to have lots of competing voices in the media," "Too much violence in the media leads to a violent society," and "Fiction should avoid oversimplifying depictions of the self."

This approach can rely on a lot of very narrow scientific findings on media effects. For instance, suspense can increase the persuasive power of a story. That sort of thing. By combining a thousand little pieces of knowledge, we can start to gain an understanding of what media with positive consequences would look like. Marketers and advertisers do exactly this except that they target a different behavioural output in their audiences.

The content approach can also refer to a list of memes that are safe bets to have good consequences if circulate through the media. For instance,

Rationality is great
Science is useful and non-arbitrary
Morality is non-arbitrary
Atheism is acceptable
Breaking conventions is acceptable
Understanding cognitive biases is good
Emotional control is good
Learning is good
Receiving criticism is good
Racism is bad
Sexism is bad
Speciesism is bad
Homophobia is bad

It would also be a good idea to circulate more concrete memes specific to current issues:

Abortion should be legal
War X should not happen
Apartheid X should be stopped
Gay marriage should be legal
Marijuana should be legal
Capital punishment should be far less common
Gun control laws should be stricter
Climate change should be taken seriously
Party X should win the election
The rich should be taxed more

Alternatively, we might care about the structure and dynamics of the mass media. The relationship between content and effects is so complex that it might be too difficult to take advantage of in a hugely beneficial way. But we might be able to look at the role media plays in our lives, how we engage with it, how it's affecting us for better or worse, and how specific aspects of media connect with specific consequences on culture. This approach relates to statements such as "Writing is going to ruin our memories," "Talking on the phone and online hurt our ability to communicate with people face to face," "The optimal amount of time to spend watching television a day is 1 hour," and "We're bombarded with so many media messages that we've become overstimulated, desensitized, and alienated."

For example, media content can be useful for stimulation, relaxation, and catharsis. These are non-trivial factors to remember in our evaluation of media's worth. Perhaps TV watching is stimulating up to a certain number of viewing hours per week, but trails off beyond that point. Perhaps listening to mainstream radio in the car is more mind-numbing than it is relaxing or stimulating.

Together, these two approaches can help us optimize - or at least, know how to optimize - our system of media. The next obstacle is to find ways to influence policies, audience habits, and creative decisions toward greater wellbeing.


In order to recommend specific media uses that I think lead to good consequences, I need to have a way of recognizing a good piece of media content when I see one.

I came up with a methodology for roughly evaluating media effectiveness and I've been applying it to various uses of media. I've also been researching multiple fields in order to find information that informs my application of this methodology. If I pull all the answers out of my ass, then the methodology wouldn’t be adding any semblance of objectivity to my judgements.

Thus far some fields I've covered are the psychology of fiction, the psychology of persuasion, experimental aesthetics, neuroaesthetics, neuromarketing, multimedia learning, audience reception studies, cultivation theory, various cultural and communications theories, etc. The more I know about how media content affects people, the better I'll be at evaluating which ones have good consequences and which ones have bad ones.

My method involves breaking down difficult, mysterious questions into smaller, more answerable ones.

To borrow from a past post:

Effectiveness = Strength of Impact x Quality of Impact

In other words, we want the effects of our media uses to be both as large and as good as possible.

How might one evaluate the strength of a media project's impact?

I’ve come up with this list of 5 questions we can ask ourselves to determine a work’s impact. In answer to each question, we might answer “High,” “Medium,” or “Low."
  1. How many people does the project reach?
  2. How significantly does it impact the people it reaches?
  3. How likely are the people it impacts to spread this impact?
  4. How long lasting is its impact?
  5. How grave was the issue pre-impact?*
[*The fifth question is only relevant to some consequentialist views, but many effective altruists would value this question as an indicator of the value of a media project.]

Answering these questions gives us a rough idea of the degree to which a media project can impact the world. But none of these answers tell us whether the impact in question is positive or negative.

How might one evaluate the goodness or badness of a media project’s impact?

I’ve come up with this list of 3 questions we can ask ourselves to determine the quality of a work’s impact. In answer to each question, we can choose a value from a 7-point scale: -3, -2, -1, 0, 1, 2, 3. This is because we don't want to just distinguish between very good and slightly good initiatives. We also want to distinguish between good and bad initiatives.
  1. How much does it increase the accuracy of people's models of reality?
  2. How much does it improve people’s quality of life?
  3. How much more likely does it make people to act altruistically toward others?
This is a very rough method of determining effectiveness but it at least reduces the problem to answerable questions. I’ve now applied this method to 8 case studies, with justifications for my decisions posted on this blog.

[04/02/14 - EDIT: I no longer see the above method as particularly useful to me and don't intend to include it in my paper. I think my approach can be useful for clarifying and organizing intuitions but it doesn't elevate art evaluation above the level of intuitions and heuristics.]


My analysis of these case studies has helped me clarify my intuitions and realize what I still have left to learn.

The research question currently on my mind is, “How reliable are traditional criteria of artistic greatness at predicting good consequences?” In my case studies, I found myself scoring an art house film, Guy Maddin’s My Winnipeg, identically to James Cameron’s Avatar – except My Winnipeg reaches less people and thus gets less done. I could think of no real basis in the context of my method for scoring the art house film extra points for artistic greatness. I’m currently looking for research on whether traditional criteria such as profundity of ideas, emotional impact, timelessness, popularity, originality, and others actually have concrete benefits on viewers, or whether they’re entirely useless.

I’m also looking to develop concrete recommendations that dictate specific actions. One aspect of that is developing a list of unusual ideas that are safe bets for effective altruists to disseminate through the media. For instance, utilitarianism, anti-speciesism, Thaler and Sunstein’s notion of “nudges,” and rationality. These ideas might be misunderstood and used in bad ways, or they might raise the general level of rationality and thus help resolve cultural issues. Do we have reasons to favour one prediction over another?

Additionally, I would like to get others to apply my method to the same 8 case studies I chose, so that I can compare my answers to theirs. I suspect that answers would largely converge but I would like to prove this.

Finally, I plan to use many of the facts I'm learning about the psychology of fiction and multimedia learning to my advantage. I intend to weave psychological insights from the fields I'm researching into my paper. For an example, read the introduction of my post on the construction of disbelief. I may complement some sections of my paper with bits of fiction that help illuminate the key points and demonstrate how the fiction version affects readers differently.


  • Effective altruists want to do as much good as they can possibly do.
  • People aren't sure exactly how to do that.
  • Until we're sure, it's worth exploring new possibilities.
  • The possibility I'm exploring is the media-culture relationship.
  • I think it has a lot of potential because of how resolving cultural issues leads to the resolution of other issues.

No comments:

Post a Comment