Having productive political discussions is hard because the world is staggeringly complex. We need better maps to help us navigate the byzantine maze of political views and to improve how we talk about our differing worldviews.
In this post, I’ll demonstrate a format that uses elements of path diagrams and systems thinking to add structure to how we communicate our worldviews. For people already interested in policy, this new format would make it easier to have productive and clear discussions.
Steven Pinker believes that the world is getting better. In his book Enlightenment Now, Pinker explains the reasoning behind his view. I’m going to use a small part of the book as an example of a political worldview, but I’ll be making some simplifying modifications so I’ll refer to the viewholder as Pinker-Like-Person (PLP), and not Pinker himself.
PLP devotes each chapter to a single topic, such as health, environment, inequality. When taken together, they encompass the entirety of his worldview. Let’s put these in a diagram, and indicate that they impact the overall view:
In these diagrams, causality flows from bottom to top — for example, if the level of inequality changes, PLP’s overall view of the world will change as a result.
The plus/minus signs indicate the direction of the relationship — for example, if inequality decreases, PLP’s overall view will be that the world is getting better, so that relationship is negative.
For each topic, PLP selects some measurements to judge the direction the topic is moving in. For this example, I’ll only add the ones for the environment to the diagram:
So, for example, according to PLP — if land use by humans increases, the environment will be negatively impacted, and thus the world will be in a worse position than before.
The grey text indicates the precise definition of the measurement (the ^ symbol in “Population” indicates that no precise definition needs to be stated, as population is already a precise definition).
PLP also lists the things that have a causal impact on the measurements. By adding these, the diagram ends up showing the structure of PLP’s worldview — it displays the topics he values, how he chooses to measure those topics, and his beliefs around what impacts those measurements.
This format could be used to display worldviews belonging to people, political parties, think tanks, non-profits, etc.
The Activist Lens
A worldview like PLP’s is a way to monitor the world. Lenses can be applied to a worldview to communicate extra information.
Imagine a non-profit that is focused on doing the most good with the resources it has. The non-profit has a broad worldview — beliefs about how the world should be:
From this worldview, the non-profit selects channels to spend their resources on. In this example, if the non-profit chose to focus on bed nets to combat malaria, the “activist lens” on their worldview would look like this:
The activist lens communicates where resources are being allocated, and the paths through which allocation impacts the monitoring view.
Another example of an activist lens is a politician’s campaign platform — from a broader worldview, they select a few key topics to focus on and a plan for how to impact those topics.
Now that we’ve seen a few examples of the format, let’s dive deeper into the details.
Types of elements
For the viewholder, the topics are intrinsically valuable. For example, if I have the environment as one of my topics, it means that I value the environment inherently, and not because it might lead to other benefits such as improved health for people.
Topics are nebulous — two different people can say they care about “equality” but actually be referring to different things. The topic measurements are where the worldview gets specific. Here’s an example of the process of converting the “equality” topic into a precise topic measurement:
- What type of equality? Economic.
- Definitely economic equality, and not poverty? Yes, economic inequality.
- Is it equality of opportunity or equality of outcome? Equality of outcome.
- Is it equality of wealth outcome or income outcome? Income.
- Which of the nine possible measurements Wikipedia lists for measuring income inequality of outcome will I use? Labor share of GDP.
Determinants are the elements that the viewholder believes have a causal impact on the topic measurements. They are the viewholder’s beliefs about how the world works. They represent the structure of a particular field of knowledge — for example, an economics-related topic might include how labor force participation, migration, GDP growth, and interest rates impact the topic.
Let’s say I believed that when a small number of firms gain a substantial share of their market, labor’s share of GDP decreases, and thus inequality increases.
We can add more information to the connection going from the determinant to the topic measurement, explaining more about my beliefs and providing the evidence I’m using (ignore the information overload here — this is just an example of how the data is structured, not how it would be surfaced to users):
Breaking down entire worldviews into these atomic units makes it easier for someone reading the view to focus their attention — if I want to find out about someone’s values, I can look at their topics and topic measurements. If I want to learn about their views on how certain policy levers impact certain outcomes, I can look at their determinants.
The policy wonk gradient
As you move from top to bottom, people’s worldviews become less fleshed out. Most people have a worldview, and thus a few topics they care about. Fewer people have selected topic measurements for monitoring those topics. Even fewer people have strong beliefs about what the determinants are for their topic measurements.
The choice/logic gradient
As you move from top to bottom, things become less of a personal choice. People can argue about what our choices of topics and topic measurements should be, but they are personal choices at the end of the day. But determinants aren’t personal choices — if you want to be correct, they have to reflect reality.
The leading/lagging indicator gradient
Indicators towards the bottom are more leading and input-like, while indicators towards the top are more lagging and output-like.
What are some ways that worldviews can be different?
People care about different things:
(They can also care about the same things but weight them differently — this is discussed later in the post.)
Two people can both have the equality topic in their worldview but select different ways to measure it:
Two people can have the same topic and topic measurement but differ in their beliefs about what has a causal relationship with the topic measurement (note the different direction indicators in the links between the murder rate and gun ownership.):
Benefits of the format
The benefits for the viewholder are:
- A more rigorous worldview
- It becomes easier to process news and analysis — for example, if someone writes an article saying labor force participation is too low, how does that manifest in the things I care about? Another example is displaying the projected impacts of a particular policy proposal across all aspects of your worldview
The benefits for a reader of a worldview are:
- Faster and more comprehensive understanding
- Easier to compare worldviews
So far, I’ve just been talking about a format — you could use it just by writing your worldview down on a piece of paper. But, like many things, it would be better to have these worldviews centralized online, where people can keep an always-up-to-date, canonical version of their worldview available for sharing.
Pinker’s book Enlightenment Now provides data for the topic measurements and determinants that make up his worldview, at the point of the book being published. There is currently not an easy way to track this data over time. I shouldn’t have to wait and rely on a new edition of his book to be published to see how Pinker’s worldview is interpreting the new state of the world. This is something the website would address — in addition to just displaying the worldview in the format, it would also show the current (and historical) data for the topic measurements and determinants.
Here’s an example worldview held by an Australian, with some actual data being displayed (a DALY is a “disability-adjusted life year” — a measure of overall disease burden, expressed as the number of years lost due to ill-health, disability or early death):
The website could allow for easy delegation — for example, if you care about the environment but don’t have the time or desire to select topic measurements and determinants, you could easily delegate those to an organization you trust, such as the Yale Center for Environmental Law & Policy. Or someone who doesn’t know much about economics could delegate the selection of economics-related determinants (e.g., would an increase in the minimum wage have an impact on unemployment?) to the IGM Economic Experts Panel. You could even delegate certain things to prediction markets (something like “if the prediction market indicates that increasing the minimum wage would have an impact on unemployment, add it as a determinant”).
The website would make it easy to quickly compare two different worldviews, and show you exactly where they overlap and where they differ.
The website would also provide the fundamental benefits of digitally structuring information. Worldviews, and every component thereof, would have a canonical home, easily linked to and shared — imagine being able to say “here is a link to my view on inequality — it will always be up-to-date”.
Lastly, the website would track any worldview changes over time.
Now that we have established the concept of displaying the actual values for the topic measurements, we can talk about another way that worldviews can differ — two people can have the same topic, topic measurement, determinant, but differ in what they think the value of the determinant is. A recent example is the debate on what percentage of bankruptcies are caused by medical problems:
Topic measurements and determinants have a number, by definition. Topics themselves don’t, by default, have a number, but it’s possible to give them one by combining the topic measurements into an aggregate number, otherwise known as an index. Here is part of The Heritage Foundation’s Index of Economic Freedom (IEF) for Australia (truncated for space):
Indexes can also encompass an entire worldview — here is the UN’s Human Development Index (HDI) for Australia:
Indexes make it much easier to track progress over time and compare regions. They also make it much quicker to create the first draft of a worldview — for example, you could simply add the Democracy Index, Human Freedom Index, Environmental Performance Index, and the Inequality-adjusted Human Development Index as your topics, and over time update the weightings between them and add/remove items.
Indexes also force the viewholder to assign weights to the elements in their worldview, which makes them more rigorous. Currently, most indexes avoid this rigor by simply assigning equal weights to their components, but at least that creates a starting point for discussing trade-offs. The website would make indexes clearer and easier to understand by displaying these weights. It would also allow viewholders to easily edit the weightings and details of existing indexes to create their own indexes, or build their own from the ground up by asking questions about the trade-offs they are willing to make.
Freeform words, with little to no structure (like an essay), make the act of communicating a worldview easier, but make reading and comparing worldviews harder, and leaves room for ambiguity. Documenting a worldview in this new format will be hard — you lose the flexibility that freeform words provide. But the clarity would be worth the effort.
The new format focuses on desired policy outcomes, and totally ignores politics and politicians altogether. It’s not that those things are unimportant, but rather that the format is intended to serve as an alternative way to engage with and think about politics, not as a replacement. We’ll always have long essays, soaring rhetoric, and personal political attacks. But as long as those things also include numbers, we should experiment with new ways to communicate these views.