Anchoring to Other Perspectives

“Stepan Arkadyevitch subscribed to a liberal paper, and read it. It was not extreme in its views, but advocated those principles which the majority held. And though he was not really interested in science or art or politics, he strongly adhered to such views on all these subjects as the majority, including his paper, advocated, and he changed them only when the majority changed them; or more correctly, he did not change them, but they themselves imperceptibly changed in him.” – Anna Karenina, Leo Tolstoy

When I was younger I noticed that I agreed with mainstream public opinion on all the subjects I knew nothing about but that I disagreed with the mainstream opinion on all subjects that I considered myself informed on. When I realized this I inferred that my agreement with the majority on those subjects was evidence of my ignorance and that learning about these subjects would almost certainly reduce my proximity to the status quo.

To clarify, I didn’t just think that expertise provided one’s views with more detail and accuracy. I believed that expertise necessarily led to one holding different beliefs from the general public and the public experts that inform them. In regards to cinema, for instance, I started off by watching everything on the IMDb Top 250 list and using IMDb extensively. I took popular movie reviews seriously and my favourite movies were the same kinds of movies your average teenager would have listed. After watching a few hundred movies I outgrew newspaper reviews, popular websites and forums, Hollywood in general, and held mostly minority views about the philosophy of art. Moreover, most of the other people that had seen thousands of movies also seemed to be more likely to hold these sorts of minority views, such as a distinct disinterest in the Oscars.

Now I have an increased respect for commonsense opinions as tethers to straightforward thinking. I’ll first present the case for skepticism of public opinion and then argue for the utility of anchoring to it.

Experts and amateurs don’t always disagree. Take this list of mundane facts:

• The USA contains 50 states.
• We live in a galaxy known as The Milky Way.
• Most of our milk comes from cows.
• Hockey is a sport usually played on ice.

These are all trivial facts that are believed by both the general public and by experts in these respective fields. It seems that the split between expert and amateur opinions does not begin at the level of surface facts. An incorrect amateur public opinion is more likely to arrive from:

1) Ignorance of subtle facts

Media corporations are in the business of selling content, not in the business of maximizing education. Highly nuanced views do not necessarily sell as well as simple ones that can be understood by a broader audience. In hockey, for example, fans and journalists rely on basic statistics such as goals, assists, points, +/-, hits, blocked shots, PIM, and, when evaluating goalies: Save %, Goals Against Average and shutouts. These stats are useful at gaining a general understanding of which players are better than others but they have limitations. Amateur bloggers, with no space limitations and less pressure to entertain a wide audience, have used in-depth statistics such as Corsi, Fenwick, shot quality, quality of competition, WOWY, and others that greatly increase the accuracy of models of player performance despite receiving literally no mainstream attention. Insofar as the in-depth statistics change our opinions of player performance and value, we should expect differences between public opinion and the very informed. Note that the “experts” with the greatest media exposure are not necessarily the most informed.

2) Lack of technical knowledge

Sometimes boring technical distinctions are crucial in shaping opinions. When epistemology was still a debate about a priori and a posteriori knowledge, Kant opened the door for new positions by establishing the concept of synthetic a priori propositions. Similarly, probabilistic epistemology resolves disputes where the arguers are thinking in binary terms. Just like in-depth information, technical knowledge often isn’t very interesting or marketable to the general public and is thus left out of mainstream media coverage. On issues where there exists technical knowledge known to experts but not to the general public, we should similarly expect disagreements.

3) False normative inferences

Sometimes two people are privy to the same information but still disagree. This might be because they are making different normative inferences. A pair arguing about the acceptability of gay marriage may be in agreement about all the relevant descriptive facts. Where they differ is likely to be with regards to emotional reactions and moral statements such as, “Homosexuality is wrong.” It is often possible to reduce normative questions to descriptive ones but pundits often aren’t equipped to recognize and defuse these situations when they occur. There is much disagreement on ethics even among moral philosophers. Political pundits could not please them all even if they were educated on the relevant issues, which they are unlikely to be.

Recap:

Imagine that Person A and Person B are having an argument about abortion. Person A believes that fetuses feel enough pain that they should never be aborted. Person B believes that abortion is okay because fetuses don’t feel pain. They are having a factual disagreement.

If they resolve this issue, they might move on to another area of disagreement. Here, Person A believes that abortion is wrong because a fetus is a human being and we shouldn’t kill human beings. Person B believes that abortion is okay because fetuses do not qualify as human beings. This dispute is about the nature of a technical distinction.

If they resolve this issue, they might move on to another area of disagreement. Person A believes that abortion is bad because her deontological code of ethics tells her that, “Murder is wrong.” Person B believes that abortion is okay because her consequentialist code of ethics tells her that murder is only wrong when it leads to more suffering than it cures. This is a normative disagreement. Sometimes this normative disagreement is beneath the level of self-consciousness and manifests itself only as a strong emotional reaction.

Despite these shortcomings, the status quo is generally a safe opinion to tentatively anchor to. A complete dismissal of public opinion is more likely to lead one toward bad ideas than good ones.

Effective altruists are concerned with maximizing the amount of good they can do in the world. Unfortunately, this is easier said than done. The long-term and varied consequences of actions often lead down unexpected paths. In addition to the difficulty of tracking consequences, there’s the problem of the quantity of directions from which horrible suffering could occur. This means that to generate a good model of the future and to come up with effective ways of reducing expected suffering, we often need to synthesize our knowledge of many subjects in creative ways. As a result, effective altruist concerns can be so far removed from public opinion that they risk coming off as science fiction to the average person.

A quick scan of some popular EA blogs reveals anxiety over possible destructive uses of nanotechnology, the moral relevance of the suffering of agents within hyper-realistic computer simulations, projects attempting to reverse aging and death, wild-animal suffering on other planets caused by space exploration, the economics of a society of emulated agents, etc. Although some intelligent people dedicate their time to examining these possibilities, to the average person they aren’t worthy of any attention due to how unlikely any of them are of occurring. My younger self would have ignored this disagreement with the majority on the grounds that the majority simply doesn’t know what it’s talking about. Why value the average Joe’s opinion on the probability of a singularity occurring when the average Joe knows little to nothing about AI? At the same time, on average, contrarian views are less accurate than standard views. Contrarians should therefore feel pressured to have especially good justifications for their views.

Although in any individual case it might be true that public opinion is ignorant, on average, staying within a certain radius of public opinion is likely to keep you from getting disastrous results. There is also the rational view of tradition to consider. If one stumbles across a fence, it is best to find out what that fence is doing there before tearing it down. The default view should be that if a fence is standing, somebody had a reason for putting it there. Standard ways of thinking and doing things probably have some kind of upside that makes them so popular.

My current technique is to occasionally anchor to alternative perspectives that prevent myself from straying too far off course. When I start thinking about issues like the singularity or whether parking lots reduce insect suffering, I try to put myself in the shoes of a more conventional thinker and wonder what they would think of me. When people don’t do this, they risk putting nerdiness and fetishism before rationality and altruism.

Anchoring to public opinion (or alternate perspectives) can also be helpful in everyday life, particularly in identity construction. People easily gravitate toward cookie cutter identities like “hipster,” “nerd,” “hip hop,” “emo,” or “jock” without realizing it. They simply fall into a subculture and start to act, talk, and dress in ways typical of that subculture. Once in one of these subcultures, it is easy to prioritize that subculture’s norms above what is rational. Making friends in other identity categories can help you understand how your own identity appears from outside perspectives. This allows you to distance yourself from associations that hold you back. Periodically, I still check myself against other identity categories in a kind of reflective equilibrium among the insights contained within each perspective. “Effective altruist” and “aspiring rationalist” cultures are subcultures like any of these. The insularity of these groups leads conversations down paths that appear obviously ridiculous to the members of most other identity categories. Anchoring to alternate viewpoints reduces this problem.

This might seem painfully restricting but I think it’s helpful in preventing me from spending huge portions of my time thinking about things that may turn out to be useless. Some EA causes may do very little good if, for example, the catastrophe they forecast never occurs. When one theorizes about insect suffering or artificial suffering or the suffering of countless future generations, one deals with quantities so vast and error bars so large that (1) Pascal’s Mugging and (2) being plain wrong are likely. The further ahead one speculates the less likely speculations are to be accurate.

None of this is to say that any one of these issues isn't worth exploring. But the most productive approach to effective altruism is probably one that focuses on a cause very likely to actually prevent lots of suffering. EAs that try to maximize their positive impact by swinging for the fences ignore the fact that on average, conservative bets will accomplish more good than will championing high-risk/high-reward causes.