Clarify problem, contribution in the abstract

[JamesPHoughton]

In reply to https://arxiv.org/abs/2010.02188v3

Reviewer 1:

Starting from the abstract, it's super unclear what's the paper exactly doing. Nearly the first 2/3 of the abstract is criticizing the assumption that we study diffusion of ideas one at a time. But it's not clear what's the criticism really, with a sentence like "this could be true, or could be wildly off the mark!"

So, clearly I need to be more explicit about the problem that assuming independence between diffusants is bad. Fair enough. Really there is a secondary assumption - not just that beliefs are independent, but that studying them as independent gives us the same understanding of social contagion that we would have had if we had studied them as interdependent. That could be more explicit.

R1 continues:

The abstract then claims "Interdependence between beliefs generates polarization, irrespective of social network structure, homophily, demographics, politics, or any other commonly cited cause." This is such a huge claim! I encourage the author to rethink this sentence through the lens of what assumptions are hidden behind it. For instance, suppose there's only minimal interdependence (very close to independent, but not exactly independent). Then results generated from the model should be closer to the standard model and, as such, the claim can't be right.

Yes, it's a huge claim. Thats the point. Thats why interdependence is important. Thats why I have a simulation, an observational study, and a lab experiment. To explain, contextualize, and test that huge claim. Not just some "that can't be right" intuition based on the fact that if you remove most of the intervention, most of the effect goes away. You don't reject the claim that advil cures headaches just because when you cut the pills into tiny pieces you see no effect.

I can say that "in the presence of strong interdependencies" we get an effect, but that sends people down a rabbit hole of saying "well, how much..." which is irrelevant when your prior claim is that there is categorically no effect!

Or, how can there we polarization regardless of politics, really?

There can be polarization without politics because of interdependence between beliefs and the emergent processes these interdependencies create. The fact that this is not obvious is why I wrote this paper. You can't claim that there is politics in the simulation or the experiment with a straight face, and yet, there it is in the outcome measures. If you want to take offense to the outcome measures, fine. But that's what you should do.

Perhaps I was not explicit that this is a claim my paper makes. The way I read this, it's as if the reviewer thought I was just saying this without backing it up. Maybe I need to introduce every sentence with "I find that..."

[JamesPHoughton]

"Politics" is a slippery word. I meant it to mean that people were deliberately trying to influence one another in order to build a following, but you could easily use it to mean a state in which people disagree - a sloppy synonym for polarization itself. Let's excise that and be more specific.

[JamesPHoughton]

rewrite:

Social contagion is the process in which people adopt a belief, idea, or practice from a neighbor and pass it along to someone else. For over 100 years, scholars of social contagion have almost exclusively assumed that only one belief, idea, or practice spreads through the population at a time. It is a default assumption that we don’t bother to state, let alone justify. The assumption is so ingrained that our literature doesn’t even have a word for ``whatever is to be diffused'', because we have never needed to discuss more than one of them.

But this assumption is obviously false. Millions of beliefs, ideas, and practices (let’s call them ``diffusants'') spread through social contagion every day. Is this discrepancy between assumption and reality a problem? Is the assumption that diffusants spread independently of one another a useful simplification, or is it leading us astray? I argue the latter. This paper demonstrates that independent diffusion is not a valid proxy for the spread of mutually-interacting real-world beliefs. I use simulation to show that when diffusants interact, new sociological processes emerge that are separate from the effects of network structure, homophily, demographics, partisan-identification, political organizing, or deliberate misinformation. I show that these emergent mechanisms promote polarization and belief coordination in comparison to independent diffusion, and that these predictions are consistent with large-scale patterns observed in textual and survey data. Lastly, I manipulate the level of interaction between beliefs in a 2400-subject laboratory experiment, providing causal evidence for the hypothesized effects of interdependence on contagion outcomes.

[JamesPHoughton]

rewrite:

Social contagion is the process in which people adopt a belief, idea, or practice from a neighbor and pass it along to someone else. For over 100 years, scholars of social contagion have almost exclusively considered the diffusion of a single belief, idea, or practice at a time. We then implicitly assume that this single diffusant is a valid proxy for the spread of mutually-interacting beliefs in the real world. It is a default assumption that we don’t bother to state, let alone justify.

Is this assumption a harmless simplification, or does it lead us to overlook important aspects of social contagion? I argue the latter. In this paper, I use simulation to illustrate sociological processes that emerge purely from the interaction between simultaneously-diffusing beliefs, ideas, and practices. These simulations predict that interdependent diffusion can 1) generate polarization and 2) coordinate shared structures of internally-consistent beliefs without reference to external "ground truth". These effects are separable from the effects of network structure, homophily, demographics, partisan-identification, political organizing, or deliberate misinformation. I then show that interdependent diffusion predicts macro-scale patterns that are consistent with large-scale patterns observed in textual and survey data. Lastly, I provide causal evidence for the hypothesized effects using in a 2400-subject networked laboratory experiment.

[JamesPHoughton]

Social contagion is the process in which people adopt a belief, idea, or practice from a neighbor and pass it along to someone else. To simplify our models and experiments, social scientists almost universally consider the spread of one belief, idea, or practice at a time. We then implicitly assume that this single diffusant is a valid proxy for the spread of simultaneously-diffusing and mutually-interacting beliefs in the real world. If this assumption is violated, much of what we have learned about social contagion may be invalid or of limited applicability.

Unfortunately, we have never rigorously asked whether interactions between diffusants have macro-scale consequences for social contagion, and so the assumption remains unsubstantiated. In this paper, I argue that assuming independence between diffusants is not a harmless simplification, but fundamentally changes the process of social contagion. I use simulation to illustrate sociological processes that emerge purely from the interaction between simultaneously-diffusing beliefs, ideas, and practices (hereafter ``diffusants''). These simulations predict that interdependent diffusion can 1) generate polarization and 2) coordinate shared structures of internally-consistent beliefs without reference to external "ground truth". These effects are separable from the effects of network structure, homophily, demographics, partisan-identification, political organizing, or deliberate misinformation. I then show that interdependent diffusion predicts macro-scale patterns that are consistent with large-scale patterns observed in textual and survey data. Lastly, I provide causal evidence for the hypothesized effects using in a 2400-subject networked laboratory experiment.