I was reading this meta article and came across the idea of assuming good faith. It is mentioned in the podcast body:

So I think as long as everyone is operating from a place of assuming good intentions, good faith on the part of the other person. So you know, I'm not fighting with you, because I think you're a dingbat who's trying to ruin the site. I'm in conflict with you, because we both really care deeply, but have different ideas about what could be useful here. I think that is a productive conflict.

In addition, the community manager links to the wikipedia policy of assuming good faith:

Assuming good faith (AGF) is a fundamental principle on Wikipedia. It is the assumption that editors' edits and comments are made in good faith. Most people try to help the project, not hurt it. If this were untrue, a project like Wikipedia would be doomed from the beginning. This guideline does not require that editors continue to assume good faith in the presence of obvious evidence to the contrary (e.g. vandalism). Assuming good faith does not prohibit discussion and criticism. Rather, editors should not attribute the actions being criticized to malice unless there is specific evidence of such.

Notably, Stack Exchange does not have a strict assume good faith policy directly in its code of conduct, but based on this article it appears assuming good faith is considered a good thing. Assume good-faith also used to be in the code of conduct and it appears that there is very strong support to put it back in.

Which brings me to my question. Occasionally here on politics, we have posts being closed for being dog-whistles for something nefarious. From Wikipedia:

Dog-whistle politics is political messaging employing coded language that appears to mean one thing to the general population but has an additional, different, or more specific resonance for a targeted subgroup.

The argument goes like this. X is a dog whistle for Y thing that violates the code of conduct, therefore the post with X should be deleted. I'm all for strictly upholding the code of conduct and deplatforming bigotry, but this argument seems problematic. The good faith interpretation of any dog-whistle is always the "appears to mean one thing to the general population" clear meaning. If the user says X just means X and isn't a code for Y, then it seems like we're assuming bad faith. We're assuming the author is acting nefariously despite there being a perfectly valid "general population" interpretation. If something is bigoted, let's call it out for being bigotry. Directly offensive content is clear bad-faith and should be removed, but I'm not sure we should delete things for being dog whistles, suspected bad-faith content with a reasonable good faith explanation.

Does claiming someone is dog-whistling go against "assuming good faith"? Should we delete content we suspect has an alternate nefarious interpretation/intent?

  • 4
    Worth noting that assume good faith used to be in the Code of Conduct, but was taken out (though not in the most recent changes). See here for more info. Oct 13, 2019 at 18:54
  • 3
    @DavidGrinberg: and it's useful to note the date it was removed was 7/2018. And the context was "requiring people to assume good faith when interacting with a (throwaway) account which is a confirmed or apparent troll is merely a straitjacket that allows them to successfully flag everyone who complains about them for violations, then ditch the account when it gets suspended". So exclude trolls. The important debate about AGF is whether it should/ not apply to general users.
    – smci
    Oct 18, 2019 at 17:16
  • It's like anything in life, you have to conduct a contextual risk assessment. The internet has amplified political campaigning since the explosion of smartphone use, and we find online platforms like Facebook to be consumed by polarised and polarising campaigns. In that context, a lot of questions are asked to make the idea exist in cyberspace, not unlike how newspapers publish contentious content with contentious terminology to kind of entrain opinion and get particular ideas out there and gaining traction, so that people can cite them in debates and rants. AGF must be on a case by case basis Oct 22, 2019 at 2:09
  • Unfortunately, corporate decision makers ditched the whole concept and the one answer most closely resembling corporate goals & values We need "assume good intent" back in the Code of Conduct states that "assume good faith" is 'enabling people getting hurt'. Under these conditions this thread discusses a fictional CoC, and following our thoughts here actually violates the CoC in force? Oct 25, 2019 at 6:37

10 Answers 10


"Assume good faith" is not one of those decisions where either option is fine as long as everyone agrees to stick to it. It's a basic rule of politeness, very similar to the golden rule - if you assume bad faith and act accordingly, people will readily see bad faith in your actions too.

For instance, if I write a post complaining that busing is bad for the environment (because of exhaust fumes?), or that one should only wear pure wool clothes, and you will tell me to "fix the language", I'll be either confused or upset about being accused of a thought crime. Mind you, I'm not form the US or Canada, so those words have no special meaning for me. How inclined to you think I will be to actually fix my post? And if you downvote / close / delete my posts for what in my eyes looks like a made-up reason, how long do you think it will take me to complain about bullying on meta?

The lack of AGF is what (IMHO) recently happened with Monica in TL. Sometimes "I don't use third person pronouns" means just that - a particular style of writing, and not a profound lack of respect for transgender people, as SE staff hastily assumed.

  • 4
    But assuming good faith goes both ways – you also need to assume good faith in the people that are downvoting you. In your hypothetical, instead of complaining about oppression or “thought crimes”, you should work to address others’ concerns. If you explain: “Oh, I just mean that busses create toxic diesel exhaust”, then the down-voters can explain the connotations of what you wrote and you can change it to better convey your true meaning. You say that, as a non-American, you don’t understand these connotations. That’s fine! But to be a good-faith actor, you need to be open to learning.
    – divibisan
    Oct 10, 2019 at 15:16
  • 5
    @divibisan Assuming good faith doesn't mean "no downvotes". It means just that: that people reading an awkward sentence I wrote don't jump to conclusions that I did this intentionally, and that they take the time to explain what's wrong in the assumption that I will be willing to learn. When that's not happening, the interaction quickly becomes sour and the will to learn disappears, even if it was there in the beginning. Oct 10, 2019 at 17:05
  • 3
    That's pretty much what I wrote in my answer, so I think we're actually more in agreement than I though. We should start by assuming, absent other evidence or a pattern from the user, that the "awkward sentence" was unintentional, and try to fix it. A user acting in good-faith will take the opportunity to learn and improve their answer.
    – divibisan
    Oct 10, 2019 at 17:40
  • @divibisan, I can see why assuming good faith would be useful for written arguments, but it's not obvious why we should put quite so much faith in anonymous downvotes. Like many online communities SE's standards of identity management are wide open for resourceful bad actors whose sole reason for being is to manipulate social media online voting.
    – agc
    Oct 24, 2019 at 12:58
  • @agc I'm not sure I'm understanding your point (2 weeks is a long time in internet time). My point was simply that we can assume good faith in posts without being completely blind to connotations. AGF is a good starting point, but it's not a license to say whatever you want as long as you're clever about how you word it
    – divibisan
    Oct 24, 2019 at 16:28
  • @divibisan, Your prior comment asserts: "...assuming good faith goes both ways – you also need to assume good faith in the people that are downvoting you". These votes sometimes are, and therefore presumably may be, gamed by resourceful bad actors -- in which case an assumption of good faith would mislead faithful users into a bogus consensus.
    – agc
    Oct 25, 2019 at 10:41
  • @agc When I initially read Dmitry’s answer, I felt it divided people into the innocent answerers who get punished for dog-whistle language, and the bad-faith, PC thought police down/close voters that attack them. The point was that if you want people to assume that you’re writing in good faith, you need to show good faith by addressing criticism, not just saying “well, I didn’t mean it in an offensive way”. As I mention in later comments, I was misunderstanding their point. But it’s important to remember that AGF isn’t a blanket license to do whatever you want
    – divibisan
    Oct 25, 2019 at 14:16

The key here is to be shrewd about what the author is trying to do. Let me illustrate this by using spam.

We allow users to post links to SO/SE. There are good and bad reasons to post links, but there's also a grey area. On other sites a question like this would be permitted

I have some code not working

Some code here

You can see it here https://www.example.com

Sometimes this is legitimate, sometimes not. The first time, we tend to ignore it. One-off links happen all the time and new users don't know the rules. What we need to look for are the times where

  1. An action was clearly deliberate (i.e. this is obvious spam)
  2. There's a pattern of bad behavior

The key with dog whistles is that they're not obvious. Someone espousing, say, racist rhetoric is easy to spot. But what do you do when someone is trying to build a serious case that might be deeply controversial? Is that racism, or an honest attempt at discussion?

Dog whistles tend to be in the eye of the beholder with some seeing them as partisan jabs with others seeing them in common political speech. What we want here is this

  1. Proof that a post is really a dog whistle (i.e. explain that means something deeper). Meta is the place to do that so we can refer back to it
  2. Proof that someone has a pattern of making these types of posts.

In other words, if it's a dog whistle we need to assume good faith initially. Maybe it really is one and the poster is merely ignorant. Maybe it's just one person who feels that way. Either way, we don't need to act on the first infraction unless it's egregious.

  • You make a really good point that patterns and personal reputation is important here. The first time someone does something like this, we should assume good faith, but if they have a pattern, then they lose the right to that initial assumption.
    – divibisan
    Oct 10, 2019 at 15:30
  • 6
    @divibisan AGF doesn't mean you have to sympathize with a person doing something obviously bad, like robbing you. It means you don't start yelling "catch the thief!" when you see someone running. Oct 10, 2019 at 17:14
  • 3
    @DmitryGrigoryev That's true, but if the person running is a known thief, then its rational to be a bit quicker to yell "thief!"
    – divibisan
    Oct 10, 2019 at 17:52
  • Well, trying to discuss any highly ideologically contentious (much milder than ethnicity & IQ) means already heavy downvoting, and top answer telling you already well known arguments against, while second best is effectively how evil you were to dare to ask. So people asking in good faith ultra controversial issues generally give up such activity after a while.
    – Shadow1024
    Oct 12, 2019 at 7:35
  • 6
    This is strange: If it is a dog whistle, then it needs to be called out, criticised, downvoted and/or removed. Why should anyone here 'assume good faith' towards a post? That should never be a priority. What we should do is 'assume good faith' towards the user posting that. We do need to and should act on the first sight of a dog whistle for posts, and re-act on a pattern (>3?) of posts coming from one user? Can you clarify that? Oct 13, 2019 at 16:20
  • @LangLangC How can you possibly assume good faith toward the poster when your criticism is that their post is a dogwhistle? Oct 15, 2019 at 4:01
  • 4
    @LangLangC Again, dog whistles are not necessarily obvious. In some cases they require a lot of dot connecting to get close to an offensive point(for instance, Barack Obama want to take from the wealthy and help the poor. Just like they do in Chicacgo, where Obama was from. Oh, and there's a lot of black people in Chicago). The problem with taking "good faith" out of the picture is that this site will not survive because you can build a dog whistle case for almost anything.
    – Machavity
    Oct 15, 2019 at 13:07
  • 3
    Much agreed on often not obvious. And an addition: I tried the opposite: keeping 'good faith' in! If DW are obvious, harsher criticism than for ambiguous border cases. Helping avoid misunderstandings in more or less 'innocent' posts by pointing out the nefarious possibilities and correcting intentional whistlers with the tools at hand. Am not saying "jump on every glimpse and punch the hardest", just act on what you see and measure the response with flexibility. Oct 15, 2019 at 13:15
  • 4
    @sevenbrokenbricks Assuming good faith means assuming that they did not intend their post to be a dogwhistle and that they would be glad, as I think most of us would be, if someone pointed out that we were mistakenly giving the impression that we hold bigoted views.
    – divibisan
    Oct 15, 2019 at 14:43
  • "But what do you do when someone is trying to build a serious case that might be deeply controversial? Is that racism, or an honest attempt at discussion?" This is a false dichotomy. Many racists are sincerely and honestly interested in discussing their racism and appear to believe that they can convince others of their racist worldview via such discussion. May 21, 2023 at 8:51

How to deal with Dog Whistles

Are you sure it's a "dog whistle?"

If there's not an easily understood charitable explanation, it's just bigotry and violates the code of conduct. If it's not a real dog whistle, don't call it a dog whistle.

Be nice, assume good intent on the part of the other person.

Don't say:

This is a bigoted dog whistle. I'm voting to delete.

This violates the be nice policy by implying someone is acting in a bigoted way when they might just be a non-native English speaker or unaware of culturally specific dog-whistles.

Do say:

I'm sure this was unintentional, but I'm concerned that "this turn of phrase" has been used as code by < bigoted-group > to mean < bad thing >. Does changing it to "better phrase" still get your point across?

If the user doesn't answer, just do the edit. If the user responds with bigotry (as does unfortunately happen) down-vote and flag for moderator help/deletion.

If the user has a pattern of problematic behavior bring it up in meta and flag to notify mods. Regular users can do a lot, but these kinds of situations are better handled by designated moderators.

  • 1
    But then there's politics.stackexchange.com/questions/46507/…, which is very likely the same user as metasmoke.erwaysoftware.com/post/89406 and metasmoke.erwaysoftware.com/post/144309. I don't want to fix this question and give this user rep, and I definitely don't want to feed the trolls. Oct 10, 2019 at 17:16
  • @GGMG I don't follow, what links the users?
    – lazarusL
    Oct 10, 2019 at 17:26
  • 1
    Writing style, mostly. Short punchy sentences, over-fixation with Fascism and Holocaust denial, single-use account. I plucked the two that showed up in Metasmoke and explicitly mentioned Fascism, but here are a few more from the Holocaust angle: metasmoke.erwaysoftware.com/post/143243 metasmoke.erwaysoftware.com/post/143092 metasmoke.erwaysoftware.com/post/138260 And of these, these are just the ones that are logged. I've flagged at least a dozen more on this site that were removed without logging. Oct 10, 2019 at 17:30
  • See my meta question at politics.meta.stackexchange.com/questions/4021/… for more recent examples. Oct 10, 2019 at 17:31
  • @GGMG I have to agree with Philip's answer to your post. It'd be nice if we could catch all trolls, but most importantly we have to uphold standards and be welcoming to what could potentially be new users. If he actually does act bigoted, bring the ban-hammer and delete his stuff but not before.
    – lazarusL
    Oct 10, 2019 at 17:36
  • 2
    I voted to close two of those questions, for not being asked in good faith (not mentioning being carelessly written). The funniest thing: in those questions he did not use any alleged dog-whistles.
    – Shadow1024
    Oct 11, 2019 at 17:24
  • 1
    @GGMG OK, but all you are showing is that he is not asking in good faith but to promote his personal ideology, which is a clear (and good) reason to close question. There is already easily enforceable rule against that. Any rule expansion would be either superfluous or a dangerous way to stifle a discussion about any touchy subject, as you can easily argue that only morally repugnant person would dare to ask.
    – Shadow1024
    Oct 12, 2019 at 7:07
  • 1
    @Shadow1024 I think we're rounding back to the same point and I agree with you. A dogwistle is a bad faith question, and a bad faith question is very often a dogwhistle, and flagging like you do is a good solution. I just object to trying to fix the question, giving the troll a conversation, and giving the user rep to, say, leave comments. Oct 12, 2019 at 13:42
  • 1
    @GGMG I think we may have a semantic issue here. The quoted definition from wiki, implies a coded message, which is intended for in-group, while general population don't get it. In case of asking insincere questions to promote any ideology, you don't communicate with in-group (because they already share this view) but with general population which have to understand it, in order to by occasion smuggle some new piece of information favourable for promoted ideology or logic inconsistency in views to be undermined.
    – Shadow1024
    Oct 12, 2019 at 14:09
  • 2
    This is incorrect: "This is a bigoted dog whistle. I'm voting to delete. – This violates the be nice policy by accusing someone of being a bigot" It says the post contains a dogwhistle and that the post is therefore of class bigoted. Criticising an action is not the same as accusing the poster of being anything, just of having done something. 'Being' cannot be corrected, doing is easily corrected. One is essentialist accusation, the other operationalised conditional. Please revise that. Oct 13, 2019 at 17:22
  • 1
    Crucial semantics. ~ "Bigoted dog whistle" is a pleonasm, one would suffice, so choose one for the example. For the explanation: "This not a very helpful comment. 1. It lacks specifity. 2. While the criticism targets the possibly problematic content (which is (be) nice) and not the poster (which would violate the CoC 'Be Nice'!), it might even be misread as 'accusing someone of being a bigot', when eg they might just be a non-native English speaker or unaware of culturally specific dog-whistles." 3. Be concrete and criticise the post, not the poster, offer alternatives (a way out). Oct 14, 2019 at 13:38
  • 1
    @LangLangC I think I actually disagree with you. I think that this argument is saying "ackshually according to strict grammar rules you have no right to be offended" when, in practice, almost all real live humans would be offended by this claim. I think people should be courteous when talking about dog-whistles and bluntly claiming someone is dog-whistling is mean and offensive in basically any cultural context.
    – lazarusL
    Oct 14, 2019 at 13:47
  • 1
    It may be a fine line to draw, but won't you agree that "What you just said is racist thing" (stop it please) is fundamentally different to "What you just said? You are (a) racist" (typical, please go away). We must be allowed to voice factual criticism of actions with clear language. Actions can be changed in the future, easily (dumb thing, sorry won't do that again) while essentialist attributions (labeling) are just an insult to stay (you are bad, nothing good can come from you). One can be used to de-escalate, the other to escalate things quickly. So: "The word used is dog whistle" Oct 14, 2019 at 13:55
  • 2
    @LangLangC I think frequently people see being a racist as a precondition to saying something racist (for better or worse). It's such a stigma in modern liberal society that I actually think both phrasings just escalate the situation very quickly. When something is obviously racist, that's fine, sic semper racism. When something could have been unintentional, I think the only way to deescalate is to be as charitable as possible and emphasize that it could have been unintentional while still insisting on removing the offensive content.
    – lazarusL
    Oct 14, 2019 at 14:13
  • 2
    We agree on that: a pattern of offensive behaviour cements the congruence of both versions and that being polite and insistent is necessary. I agree further that people are both 1 sometimes tending to not realise the difference as receiver 2 unknowing the difference or ignoring it as sender. 1 can be soften by the comment language itself, 2 is what I try to get across here. The 4 modes still allow for facts to be transmitted. CoC mustn't mean: negative facts aren't allowed to write Oct 14, 2019 at 14:23

In this day and age, you can reasonably assume good faith on every Stack Exchange except this one. I agree with divibisan that "unintentional bigotry is just as hurtful as intentional bigotry". But I disagree with the idea that we should always seek to fix such posts, because doing so means that bad faith users may end up getting upvotes. If that happens enough, trolls can end up having a say on how the site operates (via close votes in particular).

FWIW my own mental model is closer to what follows. Consider a new user who posts a question or an answer that isn't outright egregious, but contains a dog whistle:

  • If it's just one instance that could have been accidental, then assume good faith. Add a comment about it and politely raise the issue. See if a fix is forthcoming. Revise the initial assumption and downvote if it's not.

  • If other signs point towards it being an effort to troll "normies", then assume bad faith. Ruthlessly downvote, close-vote, and flag as appropriate.

The case is simpler for repeat offenders: just downvote or close-vote at the slightest hint of bigotry. (My understanding is that automated bans kick in faster for users who accumulate downvotes quickly.)

  • 1
    Good points. My answer was based on the assumption that one chose to follow AGF, but, as you point out, there are risks to doing that
    – divibisan
    Oct 10, 2019 at 16:36

The thing about dog whistles is only dogs can hear them. With progressives claiming everything from Sponge Bob Square Pants to reading to your kids being racist [2 real world examples] means ceding the public square to today's censors and to those that are way too readily to cry racism just because they have lost the argument. If someone cries racism, they have the burden of proof requirement to demonstrate it so.


No, I don't think targeting dog-whistles goes against assuming good faith. First, as you mention, there's no rules about assuming good faith - it's just a nice thing to do. But secondly, words have meaning apart from the intentions of the writer, and we cannot (nor should we) force readers to interpret posts in a certain way.

StackExchange is not a forum where individuals discuss issues. The goal of this site is to make a resource that is valuable for everyone, now and in the future. Therefore, the text of questions and answers have to stand on their own. Words have meanings (both plain and implied) and the stated (or even true) intention of the writer doesn't change that reality. We can't expect readers to also be mind-readers and in the context of published writing, unintentional bigotry is just as hurtful as intentional bigotry.

So, let's say we're "assuming good faith" and we see that someone has written a post with a bigoted dog-whistle. We should start by assuming that that person meant well and that they used that wording by mistake. We should try to avoid getting angry, or lashing out, or making it personal. However, this doesn't change what they actually wrote in reality, which can be just as hurtful or damaging to others as a plain slur.

If they are, in fact, acting in good faith, then they won't mind removing or rewording the dog-whistle. If I unintentionally wrote something that was offensive, I would want someone to point it out so that it could be fixed. I wouldn't want any unintentional offense to distract from my real point. This is actually a core part of the StackExchange model – we allow trusted users to comment on and freely edit other's posts to improve them. AGF would push us to try to improve a post by removing dog-whistles before we consider deletion, but it doesn't change the fact that the dog-whistles are a problem.

  • 2
    I like it. Bad approach "X is a bigoted dog whistle for horrible thing Y. I'm voting to delete" Good approach "You probably didn't realize this, but unfortunately X has been co opted bigoted group Z to mean Y. If we rephrase it 'as so' we can still get your point across while avoiding unfortunate alternate interpretations."
    – lazarusL
    Oct 9, 2019 at 22:58
  • 11
    As the Ellen-sitting-next-to-Bush episode demonstrates, the "unintentional bigotry is just as hurtful as intentional bigotry" is unenforceable. There are too many who insist on viewing instances of tolerance (and insufficient enmity towards those they oppose politically) as a form of intolerance. This makes every position, however neutral, a form of intolerance to some group of people.
    – grovkin
    Oct 10, 2019 at 3:17
  • 2
    @grovkin That’s why we have a transparent (in the sense that every vote and edit is public and recorded – I’m not saying that the rules are necessarily clear) and democratic process by the users and moderators to address concerns and improve the quality of posts. I see no evidence for your “everything is offensive to someone” position, and your solution is just imposing your personal view of what is offensive on everyone, rather than allowing the community to determine that through this site’s democratic mechanisms.
    – divibisan
    Oct 10, 2019 at 14:26
  • 5
    The flipside here is that people can use this to shut down or even attack people. Chris Pratt was called a white supremacist for wearing a Gadsden Flag t-shirt. What about the "Betsy Ross" flag? The national NAACP and Colin Kapernick said it's a racist symbol while others disagreed. Who gets to decide who is right?
    – Machavity
    Oct 15, 2019 at 14:02
  • 2
    @Machavity The only way we can decide on anything: by looking at the context, the user’s reputation, and making a judgement call. There’s lots of talk of “obvious” bigotry, as if it was some fundamentally separate category, but it’s not. There are "rational explanations" for even the most blatant bigotry and identifying that requires a judgement call too, just a much easier one.
    – divibisan
    Oct 15, 2019 at 14:18
  • So, that’s my core point here: there’s no clear, objective definition we can use to identify all cases of bigotry. We should make a judgement based on the context and the user’s reputation and err on the side of informing the user of the issues with what they wrote and letting them fix the problems, if the in fact did not intend their message to be a dog whistle.
    – divibisan
    Oct 15, 2019 at 14:26
  • 3
    For your specific example: the Gadsen flag, like all symbols, had many different meanings. Some people use it as a symbol of libertarian politics, for its historical significance, as a “keep out” sign, as a racist symbol, or others because they just like the way it looks. What it means in any specific circumstance is a matter of judgement. If Richard Spencer wore the flag, we can probably assume he’s using it in a racist way, if it’s Chris Pratt, we shouldn't, as your Washington Post story ("Yahoo’s story on Chris Pratt’s ‘white supremacist’ T-shirt did us all a disservice”) argued.
    – divibisan
    Oct 15, 2019 at 14:32
  • @divibisan "democratic processes" only work in communities that don't inherently attract a biased audience. May 21, 2023 at 8:53

Posting too many comments is frowned upon, so here goes a proper answer:

Q Does claiming someone is dog-whistling go against “assuming good faith”?

No, generally, it does not. If done in the right way. A dog whistle is a dog whistle and needs to be dealt with. Clear language helps with that. The only problem with this question is that it may confuse criticising behaviour with criticising the person.

First, we need a common understanding of the term dog-whistle

Dog-whistle politics is political messaging employing coded language that appears to mean one thing to the general population but has an additional, different, or more specific resonance for a targeted subgroup.

Dog whistle politics usually refers to the use of certain code words or phrases that are designed to be understood by only a small section of the populace. Generally speaking, these are phrases that have special meaning to that subsection entirely independent of its meaning to others, and represent a particularly insidious use of loaded language.

Now for taking an example, fictional:

The United Nations is organised from the American East Coast.

That the UN headquarters is situated in New York City, on the American Eastern seaboard is factually correct. That an organisation 'is organised from' or 'controlled' from its headquarters seems logical. No problem with this sentence?

The post does contain something that is used as an antisemitic dog whistle:

Other antisemitic dog whistles in German language sources include bemoaning the huge influence of ("certain people" on) the "East Coast [of the US]".

Anti-semitism. Clear violation of multiple standards. So, how to address that?

A similar example that might be more clear to people from the US would be references to:

... the New York media ...

which could refer to the specific media environment in the New York City area, or it could be used as a dog-whistle to refer the:

... the New York [Jew-controlled] media

Analysing the situation: the example lacks context, we should judge posts by their content, not by who wrote it. We do assume good faith and even if the context still leaves this as ambiguous: that is a dog whistle, does the poster use it as such or is this an innocent slip done out of ignorance?

This is strange: If it is a dog whistle, then it needs to be called out, criticised, downvoted and/or removed.

Why criticise?

To help someone improve. To see a change that we would like. To further the discussion. (src)

Crucially by criticising the post! But not the poster!

Why should anyone here 'assume good faith' towards a post? That should never be a priority. What we should do is 'assume good faith' towards the user posting that.

We do need to and should act on the first sight of a dog whistle for posts, and should re-act differently on a pattern (>3?) of posts coming from one user.

Criticising an action is not the same as accusing the poster of being anything. Just that the user has done something.

'Being' something is a permanent label that cannot be 'corrected' easily, especially not by any action the poster may take.
("The poster is a bigot. The poster will be doing bigoted things in future")

'Action' or 'did wrong' is easily corrected. Fix the post, avoid the error in future: do not use a dog whistle term in the future (that depends: some of these terms are just evil, some just need proper context or explanation)
("The poster did something wrong. Whether intentional or not: The poster corrected that mistake, learned something, hopefully avoids that behaviour in future.)

One is an essentialist accusation, the other operationalised conditional. Since this seems so popular here right now: one is 'identity' (a bigot), the other is 'behaviour' (committed an act we don't want).

These are really crucial semantics.

One way to address the problem would therefore look like this:

'The United Nations is organised from the American East Coast.' could be [misinterpreted](ideally link to concrete whistle criticism) as an antisemitic [dog whistle](basic explanatory link), as this kind of ambiguous language is used by racist groups to imply some sort of evil Jewish conspiracy. We definitely don't want to allow for that kind of interpretation of your answer. Can we change the language to 'The United Nations headquarters is on the East Coast of the United States?'

A helpful comment

  1. Is specific, concrete, constructive, calls the problem by name
  2. Criticism targets the possibly problematic content (which is (be) nice, according to policy) and not the poster (which would violate the CoC 'Be Nice'!).
  3. Offers alternatives as a way out.

One problem with that a approach is admittedly: If such 'criticism of action' is misread as 'accusing the poster'. Another one is that sometimes it is really just a bigot doing bigoted things: if you read dog whistle terms repeatedly coming from one account, then 'assuming good faith' still applies, but runs out much quicker than for a first occurrence.

It may be a fine line to draw, but we should agree that

  • "What you just said is a racist thing" (stop it please)

is fundamentally different to

  • "What did you just say? You are (a) racist!" (typical, please go away).

We must be allowed to voice factual criticism of actions with clear language. Actions can be changed in the future, easily (dumb thing, sorry, won't do that again) while essentialist attributions (labeling) are just an insult to stay (you are bad, nothing good can come from you, ever).

One can be used to de-escalate, the other to escalate things quickly.

So: "The word used is a dog whistle, (perhaps you didn't know,) please re-phrase that!"

We agree on that: a pattern of offensive behaviour cements the congruence of both versions and that being polite and insistent is necessary. We should also realise that the above is a useful tool in principle, but that is not a perfect tool: it may fail occasionally. Some people just can't take any criticism at all.

  1. People sometimes tend to not realise the difference as receiver
  2. People may be unknowing of the difference or ignoring it as sender.

1) Can be soften by the comment language itself, but posters might be in need of being made aware of the difference.
2) When we post a comment calling out a dog whistle we are the senders. This difference is what I try to get across here. The 4 modes still allow for facts to be transmitted. CoC mustn't mean: negative facts aren't allowed to write

–– How to Give Kind Criticism, and Avoid Being Critical
–– How to Accept Criticism with Grace and Appreciation

  • "This post contains a common anti-semitic dog whistle. That is a problem. If that wasn't your intention please clarify the text by re-phrasing it." I really don't like this approach. Someone unfamiliar with anti-semitic groups and how they code language has no idea what you're talking about after reading this comment. This could lead to frustration.
    – lazarusL
    Oct 15, 2019 at 12:49
  • A better approach is, "'The United Nations is organised from the American East Coast.' could be misinterpreted as an anti-semetic dog whistle, as this ambiguous language is used by the alt-right to imply some sort of evil Jewish conspiracy. We definitely don't want to allow for that kind of interpretation of your answer. Can we change the language to 'The United Nations headquarters is on the East Coast of the United States, giving America an advantage when it comes to framing diplomatic questions?'"
    – lazarusL
    Oct 15, 2019 at 12:50
  • 2
    @lazarusL Thx. Now that you mention it, likely. I would probably include a link in such a comment that explains the actual dog-whistle used in the post. // For the text, I'd use your very suggestion if the 'who uses it' (alt-right? is too specific, group is larger, maybe works if an eg is set before?) is modified and the 'America-advantage' angle would be just dropped (too leading for most OPs. The example was meant to show that OP may have been really talking 'just geography': where is the headquarter?). If you agree on that, please edit the post directly. Oct 15, 2019 at 13:01
  • 8
    The United Nations is organised from the American East Coast. Thank you for illustrating what I meant in my post. The problem with this is as a "dog whistle" is this is mostly a statement of fact. As a native English speaker, I simply read this as "The UN does its work on the US East Coast". When you made your jump from fact to dog whistle, you completely lost me. Either the Germans have a gross misunderstanding of US demographics (the US East Coast is not primarily Jewish), or this is a highly localized "dog whistle", specific to some Germans.
    – Machavity
    Oct 15, 2019 at 13:36
  • I think “East Coast” might be too German specific. A better example would be to use “New York” instead of “East Coast”. I have definitely and repeatedly encountered people using “New York” as a dog whistle for Jewish, but as @Machavity said “East Coast” is much more vague. A better dog whistle here (and one I’ve encountered many times) is people talking about “The New York media”, even though many of the organizations are not from New York (CNN is based in Atlanta).
    – divibisan
    Oct 15, 2019 at 14:39
  • 2
    @Machavity It's not localized to some Germans, the domestic security agency informs about "Ostküste" as antisemitic dog whistle, as do the German wikipedia, national newspapers such as the Zeit, the bpb, and organizations fighting against antisemitism such as the Amadeu Antonio Stiftung. Dog whistles don't necessarily have to be based on correct geographical facts. But it does seem to be a specifically German dog whistle; US examples might be "globalist cabal" or "New York media" as divibisan mentioned.
    – tim
    Oct 15, 2019 at 14:59
  • 1
    @Machavity Exactly my point. Invented example shows: linked to a wiki that already refs 'East Coast' as AS code multiple times. It may be perfectly innocent, strictly geographic, but it can be used as the most vile code. Perhaps indeed more for Germans than Americans (haven't checked) but since we are an international community this gets even more important. We shouldn't let any German antisemites spread their code in secretly loaded posts here, just coz we do not recognise it immediately. But Germans might perfectly well discuss the Atlantic seaboard here (when about politics?!)? Oct 15, 2019 at 15:34
  • @divibisan If you like, add your example as well with an edit (the RationalWiki has even more for 'controlled' etc) Oct 15, 2019 at 15:36
  • 1
    Hm, I don't think this East Coast thing is completely German exclusive Smearing Steil With False Anti-Semitism Charge… (I don't endorse here any side from that article, but in any case it shows a nice trail of controversy around the topic) Oct 15, 2019 at 15:46
  • "New York media" is a great example of the hazards of calling out apparent dog-whistles: New York City is where all the big-name newspapers in the United States are published (New York Times, Wall Street Journal, etc.). Is a reference to the "New York media" an anti-Semitic dog-whistle, or a reference to the closest thing the country has to a "newspaper of record"?
    – Mark
    Oct 31, 2019 at 22:56

There is one big problem with fighting alleged "dog whistle" so "assume good faith" is implemented not without good reasons, as different echo chambers use slightly different vocabulary.

Is glass half full or half empty? Well... apparently if we fight against potential dog whistles, we'd have to decide which one is the proper one, and which one is to be combated.

Even term "dog whistle" seems for me predominantly used by left wingers who witness some expression to which they are not used to. In mirror situation, people on right wing would accuse other of using... "newspeak".

If one hope that I'm exaggerating, regrettably not. As a person fully aware of this semantic problem, I actually asked a question for reasonable name for one school of thought within the left wing, in order to avoid pointlessly emotionally charged language.

Nevertheless, later it turned out not be sufficient:

There is no double standard here. The narrative that "identitarians" (another dogwhistle term, by the way)

1) Generally speaking yes, unless someone makes it absolutely blatant like putting surnames in brackets, using numeric anagrams of one particular German leader or using Latin phrase meaning "it's the will of God".

2) How would you like to distinguish "dog whistles" from "phrases from different echo chamber"? The simplest way is - in most cases there are not dog-whistles, as there is not much point for hidden code words for co-conspirators. Really, if people who you ideologically detest can communicate openly at ex. 4chan, what's the point of using so much effort on other websites? Thus generally removing content with alleged dog-whistles is a way of removing content that you personally dislike ideologically, but you have no other good excuse.

As extra bonus part of right wing started already trolling by claiming milk or OK symbol as some secret far right code words, and later having lot's of fun with when media took the bait and started freaking out.


The phrase "dog whistle" goes against good faith because the epithet "dog whistle" totally begs the question. Dog whistles imply an intentional human summoner of dogs. It also compares the opposing side to dogs, which is an insult in most every culture. So that violates the "Be nice" rule.

Imagine the worst case: a non-manipulative user posts some entirely innocent answer, (for the sake of this argument grant that it's genuinely innocent, and not even obliviously or accidentaly offensive), and it gets downvoted, and comments claim it's full of obvious dog-whistle phrases. Now what? A political answer is sidetracked into a question of Political English usage, or worse the Philosophy of Language, where the accusing side construes all replies as further evidence of guilt.

The accusers themselves might be ignorant, (they can't distinguish between honest opinion and manipulation), or malicious, (they wish to make communication as difficult as possible). In either case the question ceases to be about the actual topic, and becomes, if the accusers prevail, about the allegedly sinister motives of the innocent user, or if the accusers do not prevail, about the evasive ignorance of the accusers.


Jumping into a controversial topic without doing the basic reading is not good faith.

Demanding people spend time explaining basic things that you could research on your own is not good faith.

Also from the Wikipedia AGF page:

Be careful about citing this principle too aggressively. Just as one can incorrectly judge that another is acting in bad faith, so too can one mistakenly conclude that bad faith is being assumed; exhortations to "Assume Good Faith" can themselves reflect negative assumptions about others.

The underlying principle behind Wikipedia's AGF is that by not assuming good faith, you might lose a potentially useful contributor.

But consider the editor who insists that X just means X and isn't a code for Y, even though X means Y to everyone else, for them, X just means X and isn't a code for Y and they should be allowed to assign their own meaning to X and Y.

That person isn't a potentially useful contributor.

It doesn't even matter if they're telling the truth, they aren't a useful contributor.

A useful contributor needs to understand that you don't get to make up your own meanings for X and Y that contradict everyone else's meaning for X and Y.

I would say that AGF is something that you can remind people of, if and only if you are not an involved party.

Otherwise, you're in the position of saying "Oh gosh I'm so totally new here and I didn't realise that X is a dog whistle for Y and that violates the code of conduct, but even though I'm totally new here I do somehow know about this obscure rule called AGF, so why don't you explain to me all about X and Y or I'll accuse you of not acting in good faith".

That's not good faith.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .