The Connection Question

Two rooms, same engine

Between 2017 and 2022 I ran what is, to my knowledge, the only natural experiment of its kind on the polarization era. I built two large online communities on the same engine, with the same essays at the welcome page, the same moderation tools, and the same founder at the desk. One had a name that selected for people coming out of religious households, the other a name that selected for people who had been pushed out of institutions for asking unwelcome questions. I posted the same prompts to both rooms. I saved everything. I waited five years to look at it.

The headline finding is the one I will pay for the rest of my life. Each tribe was civil everywhere except the few topics it would not let you question. The same human, with two browser tabs open on the same engine, reached for the report button twenty-three times more often on the agnostic side. The same human posted political content roughly five times more often on the IDW side. Each room had a different set of defended topics, and each room used a different mechanism to defend them.

The room shaped the voice more than the person did. That is the headline. Which then opens onto the deeper finding, which is the one this page is about.

Each room had a topic it could not talk about. The two topics were the same topic.

On the IDW side, race, ethnicity, and immigration was the second-largest category of user-against-user reports across the lifetime of the site, basically tied with the room arguing about the moderation rules themselves. On the agnostic side, that category accounted for almost none of the reports. Not because the agnostic side did not care about it. Because the room had a working in-group consensus on it, and the report button was the mechanism the room used to keep the consensus from being challenged. One room was fighting about the question openly. The other room was fighting about the question by quietly reporting anyone who got near it.

What the medium can and cannot do

Social networking, as a medium, is architecturally incapable of harmonizing conflict around the questions a tribe has chosen to protect. The medium converts every careful question into a moderation argument within hours. The medium rewards the post that fights the other tribe. The medium starves the post that thinks carefully about the question both tribes are circling. The medium is not neutral. It has a structural bias toward identity defense, because identity defense is the format that holds attention, and attention is the asset the medium is built to capture.

I tried to falsify that sentence three times. With my own money, on my own platforms, with thirty years of engineering experience in human-connection systems. The medium beat me every time.

The bedrock question

When I sat down with the report logs at the end of the engineering experiment, the question both rooms had been fighting around for five years had a clearer shape than it had when I started.

To what extent can multicultural societies harmonize without coercion?

That is the question. It is one of the hardest questions in current social science. It sits at the intersection of population genetics, behavioral psychology, demography, history, immigration policy, and civic philosophy. The honest answer, in my reading of the literature, is somewhere in the middle, and the honest middle is uncomfortable in different directions for different readers. That is part of why the question had become unrunnable in public.

When I went looking for the working academic research that should have been there to inform the question, I found that the research had stopped being done. Not because the questions were uninteresting. Because the people doing the research had stopped being able to do it without losing their jobs. A graduate student who picked the question for a thesis got warned off it. A working academic who published in the area got attacked. A grant officer who funded the area saw her budget dry up. A journal editor who ran a piece in the area got the kind of mail other journal editors did not get. The topic did not get banned. The topic got quietly defunded, and the literature thinned out, and the question my rooms had been fighting around for five years became one almost no working scholar was free to study in public.

The next layer was a publication

I did not need another comment thread. I had run two comment threads to the bottom of what comment threads could do. The honest research that should have been there to inform the bedrock question had thinned out as a literature, because the working academics who would normally have done the research had been losing the jobs the research required. The institutional gap was the actual problem, not the platform.

What I needed, and what I funded, was a layer that could sit with a hard question for two thousand words and a citation list and a calm prose voice, in a tone an actual reader could finish. A publication, not a platform. One editor with clean editorial judgment, paid by a funder who would not interfere.

That is what Aporia is. I funded it from 2022 through the end of 2024, and the publication has continued under its editor since. The corpus is 832 articles. The most-cited sources in it are The Lancet, Nature, Science, The New York Times, and major academic journals. The contributor cohort runs across the political spectrum. Some contributors argue that the differences we measure between groups are primarily heritable. Others argue that they are primarily environmental, cultural, and historical. Most argue for a mix. The publication runs the debate, with citations on every claim. In the years I funded it, I never killed an article. I never asked for an article. I never redirected the editorial calendar. The editor picks. The funder does not.

I funded the publication because someone had to build the room where the bedrock question could be asked carefully, with citations, by working scholars, and almost nobody else was willing to.

That is the affirmative reason. There is no defensive version of this paragraph.

It is the same question I have been working on for thirty years. How do people connect without lying to each other.