Member preview

I’m a troll, and I didn’t even know it. Or am I being gaslighted?

Bad faith, civility, sea-lioning, and modes of argumentation

Yonatan Zunger posted this on Twitter and I had to reply in longer form:

Yonatan’s tweet that got my brain working overtime.

The linked thread from Tom Hatfield is a good one, I think, and I wouldn’t necessarily run afoul of its recommendations—I don’t argue “like I’m on Reddit” or demand citations for everything. Just a few months ago, I’d have retweeted the link Yonatan did myself, with much the same preamble.

But Tom’s remark on the importance of not having set rules so that bad-faith actors don’t game the system gave me pause in a way it wouldn’t have in the past—because I got caught up in some accusations of bad faith argument recently—and I still don’t know how I could have behaved differently, short of not attempting to engage with difficult concepts at all.


In late January, I joined an LGBTQ Slack someone recommended to me. I’d been on work Slacks and event-related Slacks, but never before on a non-work-related, ongoing Slack.¹

After reading the politics channels, the work-oriented channels, and the “fun” channels for a day, and feeling like I got the “vibe” well enough to participate, my first non-introductory post immediately got me into hot water over what I thought was an innocent (though certainly not flippant or glib) comment. In replying to someone who said they hoped, if there was war with North Korea, that Donald Trump wouldn’t “go nuclear” or it would be “really bad”, I linked to some casualty estimates showing that the prediction of deaths in Seoul in the first two days from a conventional (non-nuclear) exchange could be well over a million.

The person I’d responded to asked me to CW the comment (hide under a “content warning”, so that the comment won’t be seen without explicitly clicking on it). They said some might find it upsetting. To which I replied, “war is a terrible and upsetting thing, and we need to keep it terrible when we talk about it, or it becomes too easy to treat it as just another political tool.”

I was quickly direct-messaged (DM’d) by an admin of this Slack—summoned by the person I’d replied to—who told me that this Slack’s Code of Conduct (which I’d read, but not really fully ingested, not literally) required that I move to another channel or thread under a content or trigger warning anything someone else asks me to, no questions asked.

“Doubling down” on something… somehow

Okay… lesson learned. Or so I thought. I moved the comment under a trigger warning (“raw casualty estimates”) and then asked in open channel, “so that I understand: is it just numbers that I should thread, or mentions of casualties, or what? I’m not entirely clear on what I need to thread.”

Hoo, boy…. I immediately got an invite to a dispute-resolution channel, and in a discussion with two admins and the person from the politics channel, who complained that I was “doubling down”. I asked what that was, and the non-admin replied “tripling down!” (Which didn’t really help me understand what “doubling/tripling down” was.)

So the admins asked me to refrain from “doubling down.” I had re-read the Code of Conduct and other rules carefully just minutes before after my prior admonition, and it made no reference to this, so I asked for where I could read to learn about that. “Rules lawyering!” my (I now realized) accuser responded. The admins said “no questions asked” applied to after, as well as before (re)moving the objectionable content.

“But, I was just seeking some clarity on what content the request encompassed—I mean, war in Korea is apparently not covered, but my comment was, and I’m not sure where the line is. Is requesting clarification unacceptable if done after the content is already threaded or moved?”

“Not when it’s doubling down.”

“But what’s ‘doubling down?’”

I was told there wasn’t a written rule—yet—about “doubling down” because they couldn’t figure out the wording, but there would be, and in the meantime I must not do it now I’d been warned — and no, they couldn’t tell me what it was because like they said, there was still a dispute on the wording. (As of today, some four months later, there’s still no written policy nor any attempt to define even roughly what this disallowed behavior is.)

At this point a kind soul watching these proceedings DM’d me to tell me that, as a newbie, without even realizing it I’d just stepped into an ongoing dispute that had been weaponized by bad-faith actors who’d been recently banned. The well-wisher advised me to just drop it for now. They further advised that in the future, after someone made a request to me that something be threaded, moved, or reworded, to do that, and then to just exit the conversation.

Okay. I didn’t understand it, but I could comply, and I did.

A couple more weeks went by without incident, when I was on the US politics channel and someone remarked that they were a member of the DSA (Democratic Socialists of America) and their purpose as a DSA member was the ending of the Democratic Party as a whole and, ultimately, the United States itself.

I found this idea fascinating — what does it even mean to be a member of the DSA who is opposed to the “Democratic”² and “America” bits? I have a lot of problems with my Democratic Party, but this viewpoint wasn’t just left of mine—it was unrecognizable. But I presumed it was coherent, I just didn’t understand it. So I asked a question. I’m not sure I recall what it was, but it opened another, which I also followed up on.

An algorithm for finding common ground

In high school and college, here’s how I learned to have an argument so that the result is an understanding (hopefully for both people) of one another’s positions, rather than just yelling or tossing talking points back and forth like tennis volleys. It’s an algorithm, of an informal sort. (Apologies for the formatting below, but Medium doesn’t allow for nested lists, outlining or indentation. I’ve tried to use block elements and “1.b.iii.” item numbers to show the structure I’d otherwise indent.)

A flowchart representation of the dialectic algorithm (©Trey Harris CC BY-SA 4.0)

1. Asking the question:

Ask a question that gets to something you don’t understand about the interlocutor’s position, but that is a smaller claim than the claim of the original misunderstanding or disagreement. It can be a question about facts — how the interlocutor sees their world and the information they have to support their position — or a question about values — how the interlocutors’ morals and ethics align with the issue at question. (It can also be a question about definitions, in case the misunderstanding is just one of semantics, but you have to keep those to an absolute minimum lest the argument devolve into pedantry.)

2. Responding to the answer:

If your interlocutor’s response:
 a. is one you agree with
 Try another tack. You haven’t found the core disagreement. Move “laterally” to a new question getting at the same thing from a different angle or trying a different potential source of disagreement. Go back to 1.

b. is one you disagree with, but is about values
Reassess your understanding of their position. If you feel you now:
 i. understand the core disagreement of values:
 Go to 4.
ii. do not understand the core disagreement of values:
Dig deeper. Formulate a new question that subdivides or sets the boundaries of the values question in a better way. Ask it, and go back to 2.

c. is one you disagree with, but is about facts:
Go to 3.

d. is none of the above:
Your interlocutor is confused, not discussing in good faith, is hostile, trolling, or in some other way is not going to help you towards understanding. 
i. If you think they have a legitimate confusion, or you are confused:
Seek assistance. Ask they turn the tables, and dig into your own positions on the question to see if they can help locate the core disagreement.
ii. If you think they are arguing in bad faith:
Abort the discussion. Further discussion will not lead anywhere useful. END

3. Reconsideration of facts:

This part I will write in looser, less algorithmic prose, because it’s more improvisational. You’ve identified a disagreement with them about facts. There are several possibilities now: are they wrong, are you wrong, are you both wrong, or is the matter under dispute? This is a good point to turn the tables and let them lead the discussion, since critically examining your own factual understanding of the world is (almost, see “gaslighting”) always a good thing. At some point one or the other of you will (hopefully) adjust some facts so they’re in agreement with ground truth and you’ll reach common ground (4), or you will still have the disagreement on different terms, in which case you go back to 1 using the new disagreement.

4. Common ground:

You feel you understand the core disagreement is one of values. Paraphrase it in a way you think both of you can agree with. (Example: “It seems the basis of our disagreement is this: I think many guilty people going free is worth avoiding punishing even one innocent person, but you think some innocent people may have to be punished so that criminals won’t act with impunity.”)

a. If they agree:
You’ve reached common ground and can walk away, both understanding and respectful of where the other is coming from. END

b. If they don’t agree:
Invite the interlocutor to describe the difference instead.
i. If you agree with their wording:
Go back to 4a.
ii.If you disagree but think you’re getting closer to understanding:
Go back to 4, and try a new formulation or phrasing. 
iii. Otherwise:
Discovery of a deeper issue. You’ve agreed you’ve hit a new disagreement — but one that is more circumscribed, “deeper” in terms of worldview or values, than the original one, dependent on a smaller claim than before. Using the new disagreement, go back to 1, but invite your interlocutor to begin questioning you in the same way.


Note three things about this algorithm:
1. If allowed to complete, it has only two exit points: you conclude your interlocutor is arguing in bad faith, or you both have a better understanding. (The latter can include either or both of you changing your position(s).)
2. Except for those two exit points, it continues refining and increasing understanding and common ground until it’s fully achieved.
3. It assumes you initially control the exchange and will ask the questions until they choose to take control, unless you explicitly reach a “turn the tables” point where you ask for them to take control.

I’ll come back to these three points a little later.

The algorithm backfires

To return to the question I attempted to apply this algorithm to for understanding: My broad misunderstanding of the DSA member’s position was how one could expect a part of a major political party to work for its own disestablishment, and in turn for that major political party to work for the disestablishment of its country? Was this person under the impression that this was a legitimate goal for a major political party to have? (This could hinge on a disagreement of facts, of values, or both.) Or were they self-professing to being a sort of fifth columnist, acting against the interests of their organization? (This would be a difference of values — a rather extreme one.)

At first, I felt like I’d made some progress in understanding this person’s position, but then — before I felt I’d reached anything close to actual understanding — they became hostile, and said I was “exhausting them”, “moving the goalposts”, and “repeatedly derailing”.

Note the use of the verb “derail” without a direct object. This, I would learn, is a term of art, but I never quite understood it, since in practice it always seemed to refer to times I tried to focus the discussion or try a different angle, but never when others who used the word “derailing” to apply to me brought up irrelevancies or opened up entirely new avenues — for instance, in the DSA-vs.-America discussion, I was told that asking why a revolutionary would belong to a major political party at all was “a derail” but suddenly the people arguing with me brought up the 1973 overthrow of the Chilean government as an example of why one might legitimately seek the ending of the United States (which wasn’t something I was even wondering about—I’d presumed from the start one might honestly think the United States was more trouble than it was worth), and when I asked if that wasn’t “derailing”, was told it was not.

This ended with the other person exiting the conversation after accusing me of bad faith, specifically of “sea-lioning”. (If you’re not familiar with the concept, I’ll have more about it below.) And, in the coming weeks, in other arguments on this Slack, I was again and again accused of bad-faith tactics—always by the same three or four people, but tacitly supported, it seemed, by others, because even as I got sympathetic and encouraging DM’s, almost no one publicly defended me in the channel. I was repeatedly accused of the same things: “exhausting them”, “setting traps”, “moving the goalposts,” “derailing”, “sea-lioning”, and general bad faith.

This greatly troubled me; not only was I not, to my knowledge, doing these things (at least, the ones I understood), general bad faith was something I’d never been accused of before.

An airstrike, a comment, and a critique

Then, one day — specifically, on April 15, two days after the US airstrike on Syrian installations — someone posted angrily in the US politics channel about Democratic leaders seeking “process solutions” rather than “mourning the children we just killed”.

Note that today it seems that no one — certainly no children — died in that airstrike. Two days after, it was possible initial reports were wrong, but there were no positive reports of deaths, and only 9 injuries (3 to civilians, none children). I knew this, but chose not to bring it up, since someone who was upset about a use of military power probably didn’t want or need to hear it. And I thought it was possible—though unlikely—that there had been deaths of civilian children that hadn’t yet been reported out.

So I said nothing about this, and accepted that mourning would be something some people might think was the appropriate thing to be doing. But how did this context (in my opinion, a mistaken one, but—at that time—still just my opinion) make moves to limit war powers a bad thing?

Trying to understand this—or get the writer to understand my feeling that action was more useful than a political party “mourning” (a concept I had trouble envisioning), I replied that the Democratic moves — including by Rep. Barbara Lee, who finally had gotten a motion out of committee to request the Congress reconsider the now 17-year-old Authorization for the Use of Military Force (AUMF)— may seem like “process”, but that they actually represented an important attempt to reassert the constitutional authority to wage war granted to the Congress rather than the President. And further, that it was something a party in a legislature could do by right—there was no constitutional basis for “mourning”, whatever that was in practice.

I wrote,

“I’m not sure what a political party ‘in mourning’ would look like — but Democrats are attempting to take action in response.”

This sentence I have saved and quoted, because it’s the one that blew up.

I thought it was anodyne, and possibly might be an answer for the person who was so angry; I was trying to show that it wasn’t all “process”—in the sense of squabbling over prerogatives—but actually could, eventually, make a real difference in politics.

But this comment ended up in a firestorm, with the same small group up in arms, and a couple demanding I be banned for being so callous about the children we’d killed (who, remember, don’t exist, and who I never said on that day didn’t exist, just that my callousness was that I’d chosen to focus on things that we knew did exist instead).

This resulted in a long sub-channel being created to critique and post-mortem my “continued bad behavior”, and in which the conclusion was that I was “either a troll, or someone whose behavior is indistinguishable from trolling—and intent doesn’t matter.”

So I left the politics discussions on this Slack. The other channels reflected interests of mine that weren’t likely to result in argument, and I’d had mostly good experiences with those.

Except for that person I’d had that first DSA argument with, who shortly after that dust-up had demanded that I thread and add a content warning on the linguistics channel to any discussion of my specialty, computational linguistics, since it wasn’t, they claimed,“real linguistics”. And, since threading requests had to be followed without question, I left that channel too. I didn’t like that my specialty, on one person’s assertion, was demoted to a second-class status that no other subfield shared.

Bewilderment and reckoning

Those of you who have followed and interacted with me on Google+ over the past few years perhaps are as bewildered as I was — in over 25 years in various discussion forums online, I had never been accused of trolling (except in the most playful and jokey ways against bigoted and/or wretchedly cruel trolls) or of arguing in bad faith. In fact, on G+, at least a dozen times (grepping through my Takeout archive), others have remarked on how willing I am to continue to assume good faith on the part of commenters who are (apparently to others) clearly trolling me.

I just couldn’t figure it out: how could I be accused of bad faith so consistently and so suddenly, no matter how civil, forthright, and sincere I tried to be? Why had my attempts to explain, reconcile, apologize and listen been received so badly? Admins and even some of my accusers agreed that I had heeded every actionable criticism—but when one of them wrote, “every time you write, it doesn’t matter what, I just want to scream”, three of the others quickly added “+” reactions. What do I do with “criticism” like that?

Someone I’d never spoken to on this Slack before, a silent observer of all this, DM’d me shortly after I left these politics channels to ask if I’d considered that this little cadre I’d been tangling with repeatedly was gaslighting me, because it looked to this observer like they were, and they’d seen these same people gaslight others in the past. I thought about it, slept on it, and decided they were probably right. Since then I’ve seen this little group attack others in similar ways and sent my own little DM’s of encouragement to the subjects of their ire.

But. Let’s go back to my little algorithm, and my three points I made about it, and consider them in light of the trolling behavior I was accused of:

1. If allowed to complete, it has only two exit points: you conclude your interlocutor is arguing in bad faith, or you both have a better understanding. (The latter can include either or both of you changing your position(s).)

This method of reaching common ground was one I learned in high school forensics and in academia, in the language of scientific and rhetorical rigor. Every argument must lead, somehow, to some greater understanding (ideally for both parties, but even if for just one)— otherwise it’s pointless.

But in another context, the ongoing and unyielding nature of this style of argument could very well seem “exhausting”. And if someone is accosted with constant verbal micro-aggressions in, say, school or the workplace, the idea that they may be judged to be arguing in bad-faith and my conclusions about their words will determine this, could very well feel like I’m “setting traps”.

2. Except for those two exit points, it continues refining and increasing understanding and common ground until it’s fully achieved.

This sounds like “moving the goalposts”. It’s not — the goalposts aren’t being moved, they’re fuzzy and being brought into focus. Being asked, “can you run 45 feet?” or “can you run in this direction?” when previously asked, “can you run around 40 or 50 feet?” isn’t moving the goalposts; it’s just clarifying where they already are.

But if you’re accustomed to a style of argument that doesn’t seek understanding of one another’s positions but rather cements your own argument and seeks to contrast it and compare it positively to your interlocutor’s, it may very well feel like “derailing” when you ask a more-focused question rather than just volleying back the same argument you’ve already formulated.

3. It assumes you initially control the exchange and will ask the questions until they choose to take control, unless you explicitly reach a “turn the tables” point where you ask for them to take control.

This is the point that, with some introspection, I’m quick to agree is problematic. I expect my interlocutors to understand and share the same desire I have to reach common ground, and to take over the process of refining the disagreement whenever they feel it would be more helpful to examine my understanding.

But it doesn’t feel reciprocal to some. It feels like “sea-lioning”, especially when paired with my near-reflexive response to an argument’s getting heated of becoming more careful and civil with my words, to be quicker to apologize for even the smallest misunderstandings, and to reply only with questions or proposals, never accusations or bald statements of conjectures.

The terrible case of the sea-lion

Sea-lioning” is an insidious method of trolling. (If you’re unfamiliar with it, there’s a write-up at Know Your Meme that covers it pretty well.) I despise it. But how can you tell the difference between a troll sea-lioning and someone honestly trying to reach common ground?

The origin of “sea-lioning,” http://wondermark.com/1k62/

I honed the style of argument I described earlier in academia, at a time when I was also an activist involved in nonviolent protest, and I did weeks of training on nonviolence in the example of Martin Luther King, Jr. and Gandhi. One of the principles of nonviolence is that “giving up on your adversary, ‘agreeing to disagree’, is itself an act of violence against them.” I absorbed these ideas together and synthesized them so deeply that I often feel like I can’t ethically stop an argument until I’ve reached one of those endpoints: understanding and common ground, or a realization the other party isn’t arguing in good faith.

I can’t rid the world of sea-lioning. But I honestly don’t know how to change my style so it no longer resembles sea-lioning, but yet doesn’t turn argument into simply an identity-group touchstone, an opportunity to come up with clever quips and burns, or a performance for others watching rather than an actual attempt for those disagreeing to get something out of an argument.

I brought up some of those other tussles I had—the specifics of content warnings, the “mourning” issue, and not knowing what to do when asked to coöperate with a denial that my own training was “really” in my own field of study — because the person who warned me of gaslighting suggested that, for me, they’re all related.

To this observer, they suggested that my ability to have functional emotional conflicts with other people is hobbled by childhood trauma that caused me to be extremely suspicious of emotional arguments—because, in my child-mind, arguing with heightened emotions leads to yelling, yelling leads to verbal abuse, and verbal abuse leads to physical abuse. (This isn’t conjecture, it’s a fact about my psyche that my husband is all-too-aware of.)

They suggested I needed to learn when a link to a projection of casualties of war would be seen, not as a logical rejoinder, but as dropping an emotionally heavy shocker into a “light” conversation. (Which means, in turn, I need to learn how “I hope Trump doesn’t go nuclear” is “light” in emotion — in either sense of “light”.)

That I need to learn how to “performatively” mourn the (yes, non-existent) dead children because that was what the group needed from me right then — mourning — even though I didn’t see any real children to mourn for and thought constructive analysis would be more helpful (when it wasn’t welcome at all).

That I needed to recognize the person arguing computational linguistics isn’t linguistics was harboring a grudge against me, and to challenge them on that basis, rather than being confused at how anyone could honestly think that my subfield was not part of the larger field and just slink away.

I came inches from a ban for “civil” behavior that was believed to be in bad faith, and I still don’t know how to act differently so that I come across as acting in good faith (though I now understand a bit about why it looks that way). I’ve avoided the ban simply by avoiding further conflict entirely.

I’d like to learn a “non-sea-lion-y” constructive process of argumentation, but I haven’t found any good pointers when I’ve asked. When I was at Google, there was a class called “having difficult conversations” that taught a method almost indistinguishable from mine. (Actually, it’s more fraught than mine, since it requires both people produce data to back up their assertions.) When I’ve asked people on this Slack for suggestions of “good arguments” I could learn from, I’ve been directed to examples of what I’d consider takedowns or call-outs, not arguments. (One person even directed me to an encyclopedia of bigotry, which just made me feel bad — if my method of argument is fundamentally racist, misogynist, homophobic, transphobic, classist, ableist, or otherwise bigoted regardless of the subject matter, I’d like a bit more of a hint as to how it is so than a link to an encyclopedia.)

Someone else told me that you can have argument (where common understanding is not the goal, proving you’re right is the goal), you can have discussion (where the goal is not common understanding, but rather sequential statements of feelings and beliefs with little or no crosstalk and no direct conflict), or you can have teaching (where common understanding is being passed consensually from teacher to learner), but you can’t have two at the same time.


Is this really true? Is there no way for two people from different backgrounds who disagree about something to dig into why, and come out with some answers, without being offensive?

I feel like by his tweet, Yonatan may have a key to unlock a door I can’t even see.


¹ I won’t name the Slack since I don’t seek to draw attention to them and—certainly not!—to suggest they have some blame for any unpleasantness experienced by others on my behalf.

² Large-D Democratic; they are part of the Democratic Party.