← All sections

Independent work

You don’t necessarily have to work with a campaign or another existing organization. You can just do things yourself. Independent punditry and public activism are possibilities.

Punditry

Substack, YouTube, Instagram, etc.

The main questions are: do you have anything to say that everyone hasn’t heard many times already? And, how many people will listen if you say it?

Writing

Social media posts / videos

Comment section

Comment sections on news stories and social media are probably a waste of time. I’m not totally sure. I’ve occasionally felt compelled to respond to certain comments on the AP News site. [How big is the audience there?]

Witnessing / documenting

If you see government abuses happening, get it on video. I think ideally you would live stream it, so that if your phone gets confiscated or broken, the video still lives in the cloud; but there are some disadvantages, like bandwidth limitations, and the extra time it may take to start a live stream. [Are there apps that let you instantly start live streaming and also recording high quality on the phone?]

It may be impossible to measure, but I assume videos like the ones from Minneapolis in winter 2026 make a difference in elections.

1:1 or small group talking

If you’re not working for a campaign or some kind of political organization, you may have more credibility, especially with people who know you.

Talking to strangers

This is hard. There is one known method that evidently can change minds: deep canvassing. It involves having fairly long conversations (10-20 minutes), either in person (door-to-door) or by phone, in which both people talk about their experiences. The canvasser is trained to listen non-judgmentally to the person’s stories.

Several studies (mostly by Broockman and Kalla) have shown that it can work to increase acceptance of LGBTQ people or immigrants. One experiment (also by Broockman and Kalla, working with People’s Action Institute) tried to sway people’s choice of candidate in a presidential election (persuading people to vote for Biden instead of Trump in 2020).

I say “allegedly” because I see several reasons to be skeptical. For one thing, Broockman and Kalla apparently never published a paper on their results⁵, suggesting that they may not have had much confidence in their data. People’s Action Institute published this document, which is more PR than journal paper; its description of the experiment isn’t very complete. It claims an increase (in Biden’s margin over Trump) of 3.1 votes per 100 completed conversations (with a fairly wide 95% confidence interval of 0.29 to 5.9). It glosses over some potential confounders, as well as the fact that it seems to have probably taken thousands of person-hours to complete 695 “completed conversations,” resulting in an estimated 21.5 votes⁶.

First they did extensive preliminary experiments to tune their messaging; then they selected 460,000 people who were “likely to answer the telephone and have a conversation;” mailed them a survey; got 25,936 responses; dropped all the strong Biden or Trump supporters; sent another survey to the remaining 15,787; got 7798 responses (not counting people who responded but asked not to receive further surveys); tried to call all 7798 of them on the phone; and had full conversations (probably at least 10 minutes long) with 695 of them, and partial conversations with 322.

In addition to all that work, the 104 phone canvassers received extensive training, then additional coaching between shifts.

All 7798 respondents to the second survey were then sent a third survey; the results of the experiment are based on the differences in survey responses between those who had phone conversations vs those who didn’t.

The experimenters tried to rule out any effects of self selection at the last stage (actually consenting to a full conversation) using “placebo questions” on the surveys. As far as I can tell they didn’t have any way to rule out those effects on the earlier stages of selection. Could it be that people who initially were judged likely to respond, then fully responded to two surveys, and were not already strong supporters of either candidate, were more persuadable than the average person?

Another thing that occurs to me: In 2020, at the height of the pandemic, people may have been more persuadable than usual (maybe particularly on the idea that someone other than Trump might have handled things better). This would fit in with evidence that postcards were much more effective for get-out-the-vote efforts in 2020 than 2024.

Deep canvassing is certainly still a promising technique; the experiments involving increasing tolerance are great. I think going through the training from deepcanvass.org or peoplesactioninstitute.org would probably be a great thing for anyone interested in political activism (though I personally have not done it (yet)).

But for getting votes in a particular election, it seems to be far less time-efficient than standard get-out-the-vote canvassing.

Friends / family / acquaintances

The easy part: if you know people who tend to agree with you politically, but might not vote, send them a reminder.

The hard part: changing people’s minds. Most peoples’ political opinions are formed by mixing and matching the opinions of the people closest to them. To influence someone’s opinions, it helps to be one of those people.

But trying to change the opinions of a friend, relative or coworker is fraught, for four reasons that I can think of:

1. We can become attached to beliefs, so that having them challenged feels like an attack.
2. Some issues are inherently emotional.
3. People prefer to “win”, especially if other people are watching.
4. Some opinions are deemed to be off limits by our social circles, to the extent that even considering them might end friendships.

Because of reasons 3 and 4, I think it is best for discussions to happen in private, in very small groups (and preferably in person, not online or by phone). The optimal number of people may be two in most cases. I’d be very hesitant to encourage groups larger than four. Even four might be a terrible idea unless there are very strong bonds among some of the people, for example two married couples.

So, let’s say you have the opportunity to talk with someone one-on-one. How do you go about changing their mind?

I think that is the wrong question. It seems likely to me that, to change one person’s mind through 1:1 discussion, it helps a lot to go in not with the goal of winning an argument (or changing someone’s mind, which may or may not be the same thing), but with the goal of figuring out what’s true. Then there might be a chance that the other person can assume that attitude too.

It’s totally likely that you’re right and the other person is wrong. It’s also possible that they’re right and you’re wrong; or that you’re both wrong; or that you’re both right but haven’t communicated well; or that there is no “right” answer. If you’re 100% sure that you’re right, for one thing that’s obnoxious, and also you are wrong to be 100% sure you’re right, even if you are right. There is no valid way to be 100% sure of anything. (I think...)

It can be hard to admit you might be wrong. It requires epistemic humility, which I think is a vastly underappreciated virtue. Everyone should work to achieve it and to help others achieve it.

If you can go into an argument with humility AND the other person does too (which of course you can’t force), then there’s a chance that one or both of you might come out of the argument with better understanding than when you went in. Otherwise, not much chance.

(This is consistent with the research on deep canvassing mentioned previously, and the techniques developed there are probably useful here.)

And it probably helps if you already have some level of comfort and trust with the other person. I’m trying to imagine knocking on doors and trying to change strangers’ opinions on political matters, and I have a feeling it wouldn’t go well. I CAN imagine talking to my friends and changing their opinions, or changing mine. In fact it’s happened many times. Though maybe less and less as I get older.

A possible objection: one person goes in with an open mind and honest epistemic humility, while the other pretends to, but is actually playing to win using subtle rhetorical tricks. They might even consciously think they have an open mind, but really just be echoing the arguments they’ve heard in their bubble. Or maybe both people are honest and open-minded, but one knows a lot of the standard arguments on one side, and the other doesn’t know as much. This is a real possibility, but in any given disagreement it can work in either direction. Over time it probably mostly cancels itself out, with a slight bias towards truth.

Look things up online as you’re talking, if necessary. Rather than saying “no, you’ve got the facts wrong,” say, “let’s find out.”

So who do you talk to? There is likely to be a tradeoff: the more comfortable you are talking to someone, the more likely they are to already agree with you. I personally have trouble thinking of anyone I know well enough who is either (a) likely to vote Republican or (b) a voter in a swing state, let alone both. But if there’s anyone you can think of who’s even a remote possibility, please consider it. (I think my best prospects in 2026 might be high school classmates who still live in Michigan; I will have to make a social media post to try to find them.)

If you know anyone who is likely to not vote at all, they might be persuadable. Maybe the arguments under Vote (below) would help.