Opinion change and propagation

Main article: Opinion

It is well known that opinions change over time and propagate differently depending on cultural or circumstantial factors. Can we predict these factors and their effect on opinion? Are there algorithms we can build that will give us accurate predictions or at least explain opinions after the fact? But let's start with the difference between short-term and long-term opinion changes.

Opinion changes

Short-term opinion changes

Short-term opinion changes are clear and can be traced to an event or, perhaps, an influencer. The attack on Pearl Harbor changed US opinion on whether or not to enter WWII. The 9/11 attacks provided an impetus for direct US intervention in the Middle East and expanded government powers to surveil private citizens. The Great Depression and FDR's policies changed how American's view government intervention in the economy.

For the case of short-term opinion shifts caused by an event, it is usually easy to trace how and why the opinion changed. The infrastructure of the ratings system will have software to track changes in opinions, both on the personal and community-level, and correlate them with events. Obviously, the software will be able find the origination point of an opinion in cases where it is clearly discernible (such as a scientific paper).

Opinion shifts caused by influencers are a little harder to track but still doable. Influencers have an inordinate impact, especially given current technology, to spread their opinions across the world. It is for this reason that they are sought out by political organizations for endorsements. Both Democrats and Republicans are seeking an endorsement from Taylor Swift, for example. When and if it comes, we can track the impact it had on the race through polling, especially targeted polling of her fans.

This type of influence is fairly easy to quantify. Famous entertainment figures, like Swift, are obvious influencers and are sought out by political campaigns, philanthropic organizations, and companies trying to advertise their products. We can relate the money invested in them to increased product sales, for example. The ratings system, with its considerable quantitative abilities, will have software to isolate the effect of influencers on opinions.

Things get a little harder when we’re dealing with influencers who are not that famous, at least not at first. Let’s look at one such influencer who had an outsized impact on the public, Dr. Andrew Wakefield, who in 1998 published a paper showing a link between the MMR vaccine (given to practically all children in the developed world) and autism. His paper, in the prestigious Lancet medical journal, was subsequently retracted and he was stripped of his license to practice medicine. But his influence is still felt in lower immunization rates, a robust anti-vax movement in the US and Europe, and episodic outbreaks of measles.

Since Wakefield was a medical doctor and published in one of the most respected medical journals in the world, his influence is outsized. But other facts about this case warrant attention, especially given the intense blowback he got from the conventional medical community. The first is that he had a ready-made community of would-be followers, parents of children with autism, or any parent fearing that their child might develop autism. A parent fearing for their child’s health is, needless to say, predisposed to making incorrect risk assessments. The second is the relatively low cost of following his advice. Not giving a child the MMR vaccine will probably not result in them getting any of the diseases the vaccine prevents (Measles, Mumps, and Rubella) because of herd immunity. Although measles outbreaks do occur, they are uncommon.

Since Wakefield was not famous before his paper, his influence rests on the factors just mentioned: publication in the Lancet, the low cost of following his advice, and the vulnerable group of people he was addressing. It should also be mentioned that Wakefield, realizing that few people would actually read his journal article, held a made-for-TV press conference to announce his findings prior to publication. It appears that he was a savvy marketer.

It is hard to know how these factors played together to give him the real-world “success” he was clearly after. Some scientists try to become famous through prestige publications followed by books for the lay-person, interviews, etc. But most do not achieve Wakefield’s level of notoriety. One finding that helps is the observation that people are rewarded with attention for stating eccentric or extremist views. “Normal” views do not get much attention. But this only goes so far. Well aware of this fact, social media is awash in extremist/eccentric content and only a small portion of it becomes truly influential.

Although it may be difficult to predict this kind of influence, it still isn’t hard to analyze it after the fact. As we’ve discussed, it is clear in retrospect what the reasons for Wakefield’s influence were. And if we’re able to casually figure this out, so can our software, perhaps aided by human intuition or AI.

Long-term opinion changes

But we’re not done yet. Another level of opinion change occurs over time without clearly identifiable causes. One such issue is attitudes toward gay marriage and gay inclusion more generally. Here’s a graph of Gallup survey results on this issue over the past couple of decades:

This question has a few advantages over many political questions. Almost everyone has an opinion on it and it does not require a detailed knowledge of policy to understand. It is also not an example of generational opinion shift – that is, it is not primarily the result of a new generation with more liberal views displacing an older, more conservative generation. Individuals, young and old, changed their minds on this issue even though, admittedly, the young changed more quickly and in greater numbers.

So public opinion has dramatically shifted over the past 30 or so, from solidly against to solidly pro gay marriage. Why? There doesn’t seem to have been a single, defining event but researchers have pointed to the AIDS epidemic as starting a movement in the gay community for inclusion. Popular culture then started normalizing gay people. Another big factor, perhaps stemming from the first two, was people coming out as gay in their families and communities. One of the strongest predictors of positive attitudes toward gay people is if someone personally knows someone who is gay. Given the number of gay people in the US (about 7%) and its relatively homogeneous spread throughout society, this is a fairly large number of people coming out. Alongside this phenomenon was the realization that no one chooses to be gay, just as no one chooses to be heterosexual. By all accounts it is a trait we are born with. People used to act as if gay people “chose” to be that way, as if it were a lifestyle. It is not, and the more people realized that, the more they were willing to accept gay people.

So, we can identify a few reasons for the shift, but there doesn’t seem to be one single point at which anything changed. The desire to be accepted in society (not just ignored), a persistent movement to gain rights, and the realization that gay people are just normal people, helped. We might also note the influence of popular culture which has depicted more openly gay roles in a supportive light. But this is a chicken or egg problem. Does popular culture drive or react to the society it is a part of?

We might also note that acceptance does not come with high costs. We are not asked to change our way of life because of gay people or the fact that they want to get married. We simply have to treat them the same as everyone else, a duty which doesn’t seem too stressful. We can contrast this with an economic policy, say increased gas taxes, which is never popular even when people know it is for some public good.

We might distill the reasons for change on this issue as follows: a relatively large number of people who are clearly discriminated against, activism on their part, pop culture influence, and a low personal cost of acceptance. Underlying this is a community asking not for special treatment, but to be treated the same as everyone else.

We might be tempted to conclude, as for the simpler cases above, that our software might easily discover these historical reasons for gay acceptance. However, the trouble is that we can’t be so sure anymore. Was AIDS the foundation upon which a successful movement for inclusion developed? It seems likely that the movement would have happened even without AIDS. What role did elites in media and entertainment play? Some, obviously, but it’s hard to quantify. And once gay people started coming out, how much did that encourage others to do so as well? We intuit that it played a large role, but we really don’t know how much.

Simulation software for opinion analysis

This seems like a good exercise for social simulation software, as we’ve discussed in the past. Here we can feed it data from the past 30 (or more) years and tune the parameters to fit the historical record. We might allocate influence factors to each of the suspected causal variables and see their effect together or separately. For each influential factor, such as coming out, we can simulate how that spreads based on proximity. It is far more likely that Linda will come out if her friend Jane has come out. So as more people come out, the number of others that do so will rise exponentially until everyone is affected. This in turn will lead to an exponential rise in people who know the folks who come out and thus shift toward a favorable view of gay inclusion. After some thoughtful human input, a well-designed simulation should be able to identify how opinions on this subject would have changed and why.

Incidentally, past-looking simulation will inform the more common forward-looking use case for simulation. As simulators are tuned to past experience they will be better able to predict future events and likely changes in opinion.

One thing that probably won’t work is surveys to ask individual people if they changed their mind and why. The research on this is not conclusive but appears to at least validate the notion that people are often unaware of their changing opinions and much less, why they changed. Other studies have concluded that people have few stable opinions and tend to answer randomly based on “ideological heuristics and cues”. This finding correlates with an observation made earlier that most people essentially vote at random.

If this is true then people are affected by external factors much more than they, and we, realize. This makes it all the more important that we understand the interplay of events, cultural shifts, or even the zeitgeist of a particular time in history on opinion changes.

The ratings system will have a number of advantages in discovering these root causes. One of these is the aforementioned analysis and simulation software. Another, perhaps more important, is a large population of people who can provide opinions on the mood of the people and take stock of what types of influence are likely to be successful. This is sort of like predicting what new product will sell. There is no better predictor than a large group of people with an opinion.

Influence of money on opinion

We might also note the influence of money on changes in opinion. Our society spends an inordinate amount of money on advertising and other forms of marketing. For the 2024 election cycle the US is forecast to spend $12 billion, an all-time record. This is 3.1% of total media ad spending. Obviously this has an effect and in politics ad-spending is a requirement for a realistic chance at winning an election.

However, money is an artificial and undemocratic form of influence. Money perpetuates the US two-party lock on our politics since no one is likely to spend on third parties with virtually no chance of winning. And money is hard to predict because any eccentric rich person can influence politics in strange ways. Fortunately, the moneyless ratings-based society does away with this problem. In a moneyed society, the ratings system will presumably track monetary expenditures carefully and rate them. Moneyed communities will also probably restrict money as a political influencer since it seems unlikely that a direct democracy would accept the distraction of self-serving political campaigns over some policy position or another.

Individual opinion analysis

All of the above covers opinion changes in the aggregate but what if we are interested in individual changes in opinion? What if we want to know why our own opinion, or that of a friend, has changed? The software built into the ratings system can certainly analyze that as well if equipped with enough background knowledge and up-to-date information about someone’s life. Some of this information will already be “in the system” due to ongoing ratings and opinions. Remember that ratings-based communities will require participation in the ratings system as an integral part of everyone’s existence. After all, those who don’t participate won’t be rated, or be rated poorly, and thus be at a disadvantage in society.

So, the simulation mechanisms available to societal analysis are also available to individual analysis. If we want to know if our opinion changes, we will easily be able to produce a graph over time showing the change. In fact, such graphs and other analyses should be produced for our review automatically. If we want to know why an opinion changed, we can plot this against historical events, new scientific knowledge, the rise of some influencer or another, a technology change, or some personal event in our lives. For complex changes over time which require multiple variables to understand, we can perform the same simulations that are performed for all of society. Needless to say, these personal simulations might not have the accuracy of a society-wide simulation (where individual randomness is averaged out), but they can still provide insight. And perhaps their greatest benefit is simply to make people aware of their own opinions and why they hold them. We are aiming for a thoughtful, deliberate society based on reason, one where an individual’s introspective abilities are held in high regard.

Hacking and Security of Opinions

Another reason for plotting individual opinion, and being aware of changes, is security. People should be provided with constant feedback on their own input to the system as an integrity check and so they can formulate rules for their participation. If an opinion changes drastically, the user might have a rule to suspend their own participation until the incident can be studied.

Given the importance of the ratings system in any society based on it, folks are likely to treat the security of their accounts in the same way they treat their financial accounts. Indeed, in the RBS (Ratings Based Society) the ratings system, and the rated opinions within it, is the currency by which our economy functions. People will be highly motivated to use basic security procedures: strong passwords, two-factor authentication, etc. Higher security could be obtained through biometric authentication. The community (for the community based ratings system) will also be highly motivated to maintain secure servers and end-to-end encryption for communications.

One harder problem would be if an aggregate calculation is being hacked by inserting fake opinions of people. Some discussion and mitigation strategies have been explored for this problem.

Another important feature of security is detection of suspicious activity, much like banks do when monitoring anomalous credit card purchases. Basically, the system would learn over time, across the whole user base, to detect behavior patterns which suggest a compromised account and report them to the owner for verification. In cases where the user confirms hacking, the account can be shut down temporarily until the issue is rectified. Users will have a system of saved snapshots so they can travel back in time to recover their unhacked opinions and build back honest opinions from that point forward. To facilitate this we might envision secondary and tertiary backup systems which record people’s opinions at the time they are made and, hopefully, were not also hacked.