How should conspiracy answers be treated?

There recently was a Meta Q&A where a discrepancy in the handling of so called "tin-foil hatted" answers was brought up, i.e. answers which claim that e.g. the NSA can break AES easily or answers that the NSA is behind a certain weak cryptography standard.

This Q&A has shown that even within the mod team there's no clear consensus on how to handle them precisely, so we'd like to ask you, the community, for guidance on how you want such answers treated.

The different approaches will be provided as answers. Please upvote an answer if you feel like this is a good way to handle things, down-vote if you don't like this approach and don't vote on it if you feel neutral.

After 14 days, that is on the 5th of april 2018, we will consider the answer with the most upvotes our official policy on this matter.
As of now, the highest-voted answer is our current policy in these matters. However it will not be applied retro-actively.

• You may want to note that I if you have your own idea on how to handle this issue, you are more than welcome to post another answer here! – SEJPM Mar 22 '18 at 19:21
• I kind of wish the answers' orders were randomized, otherwise the answer that gets the first upvote will almost always become the answer that keeps the highest score. – forest Mar 31 '18 at 4:42
• @PaulUszak I think it's a per-site configuration, as different sites seem to have different behavior in that respect. – forest Apr 1 '18 at 1:03
• I can't wait to tell everyone what the real purpose of AES is... – Paul Uszak Apr 5 '18 at 20:24
• @PaulUszak i.stack.imgur.com/edzVe.jpg ;) – e-sushi Apr 5 '18 at 21:25

Let such answers stay (at the discretion of the poster) if they're not blatantly not answering the question.

That is, if an answer even partially answers the question, exclusively using conspiracy theories, it shall stay and be left to down-votes. Not-An-Answer and similar flags on these answers will then be mostly declined.

More detailed explanations can be found in e-sushi's answer to the above linked Q&A. Here's an except:

Two things to remember:

1. Mods should intervene as little as possible… aka we do not decide if a wrong or conspirancy-loaded answer is an answer in the first place — that's the job of the community (they have down votes for that). This is underlined by #2
2. Flags should not be (ab)used to kill incomplete or altogether wrong answers. Remember the mod flag rejection option that says something along these lines — which OP here points to too. Simply killing a conspiracy- loaded answer feels wrong as this somewhat contradicts what we usually do (and SE expects from us - as mods - in such situations).

flags should not be used to indicate technical inaccuracies, or an altogether wrong answer

As we know from the past, fueling conspiracy by abusing flags with the goal to suppress the related Q&As can and will result in weird suspicions targetted at Crypto.SE as a whole. We've been there before. Let's try not to reboot such drama.

• I support this approach for another reason as well. Downvoting conspiratorial rants encourages downvoting and reduces the reputation of the poster (whereas a delete, unless wielded as a spam flag, costs nothing). If the answer is subsequently deleted after some time, I believe the reputation loss stays (although I may be incorrect about this). In addition, having a sufficient number of downvoted and deleted answers eventually results in an answer ban as well, without the moderators needing to do any further work and without expiry. – diagprov Mar 22 '18 at 15:12
• While encouraging downvoting can be good, there is the issue of tripping the (secret) vote fraud script if someone has a large number of terrible answers, and heavily-downvoted answers make the site appear less... clean. – forest Mar 22 '18 at 20:02
• @forest Trying to answer all your points on a single comment: (1) Heavily downvoted As get less focus when reaching (I think) -5 votes and are faded out accordingly (example answer) to keep the appearance of the site clean. (2) Vote fraud triggering is more of an issue in cases where a single user drops many votes on a single user's Q&As. It doesn't tend to kick in when a dozen of users downvote a single A because it's incorrect/bad/whatever. (3) Answerers getting many downvotes should enhance their answer (or delete it if they prefer that). – e-sushi Mar 23 '18 at 1:42
• Would moderators reverse delete voting one these answers? – Braiam Mar 23 '18 at 15:07
• I think it's -3 votes that causes the fading out. – forest Mar 25 '18 at 21:43
• @Braiam Unless there's good reason to undelete answers — which is not the case in 99.9% of the time — mods do not undelete answers which have been deleted as the result of delete votes by the community. Remembering the few times mods undeleted a community-voted-to-delete answer within the past couple of years, it was mostly done after a related meta discussion, which clarified the community consensus on some special case. So, in the end, such undeletion was decided by the community; the mod merely clicked the button after the majority of the community decided that that's what should be done. – e-sushi Mar 26 '18 at 20:19
• @Braiam Also, moderators will not undo single delete votes dropped by one or more users either. (Just in case you wanted to know about that too.) – e-sushi Mar 26 '18 at 20:38
• @e-sushi wait, you can revert delvotes? How? – Braiam Mar 26 '18 at 21:14
• @Braiam mods cannot revert actual votes from other people. However they can single-handledly delete / undelete posts and can view deleted posts (10k rep feature) and actually search for deleted posts (mod-only feature), so what sushi meant there is that "if a post has been actually deleted by community votes / explicit community wishes, that is not solely by a mod, we won't reverse that". – SEJPM Mar 26 '18 at 21:28
• @forest basically what @ e-sushi said in reply :) The vote fraud detection script is supposed to catch potential voting rings and potential "i went through the profile and voted up/down" type behaviour. A single answer getting downvotes in due course shouldn't trigger the voting fraud script, unless you happen to only downvote that user or something. Unlike when the site started, there are now enough people to clobber these answers, hopefully. It also makes the decision process more consensus-based. – diagprov Mar 29 '18 at 14:23
• Note that by this I do mean a single answer receiving many downvotes from different users should not come up as voting fraud, even if it hits -25 or something. If you feel compelled to downvote everything by a specific user because it genuinely is all rubbish, that's probably time to flag for a moderator so they can wield the somewhat satisfying "destroy account" option (if it still exists). – diagprov Mar 29 '18 at 14:27
• @diagprov Well downvoting everything from a user ,and downloading everything that sucks from a user when a fair portion of their content sucks are different scenarios. – forest Mar 30 '18 at 1:45

Upvote if NSA claims contain links to documents

In addition to keeping the question/answer, users could tend to upvote the Q/A if links to NSA documents are included

• the links might be useful for them
• it certainly shows research effort
• (it does not make the post less clear)
• Oh certainly, saying that Dual_EC_DRBG has a backdoor is not the kind of conspiratorial answer in question since you can actually provide references to the factuality of the claim. Saying, on the other hand, that AES-NI is a backdoor or that P-521 can be broken by the NSA in seconds without the slightest bit of evidence is just silly. – forest Mar 25 '18 at 21:41
• Do other sources count like Der Spiegel, Washington Post, BBC, WikiLeaks, Amnesty International, Privacy International, Electronic Freedom Foundation, United Nations High Commissioner for Human Rights, Netflix etc? – Paul Uszak Mar 29 '18 at 22:00
• @PaulUszak: I would say that journalistic sources count exactly to the extent that the sourcing for the stories linked does. If it's a story sourced from reputable cryptographers or cybsersecurity experts, I would count that. If the story is reporting on claims by cranks, then not so much. There are also more difficult cases, like The Guardian's controversial reports on WhatsApp security last year. – Luis Casillas Mar 29 '18 at 23:57
• If WikiLeaks or the EFF claim that WEP was backdoored by the NSA (rather than just being really badly designed from the start), or that RDRAND is a confirmed bad RNG, I'll be more liable to take the claim seriously. Until a leak is published detailing the NSA's involvement in something (especially something who's poor security has a more logical and evidence-based explanation), I reserve my right to call bull. – forest Mar 30 '18 at 2:13

Remove such answers more likely if they fail other objective criteria.

That is, if an answer e.g. only tangentially or partially addresses a question and solely relies on conspiracy theories, it will be removed. Not-An-Answer and similar flags will then be more case-by-case when it comes to declining and accepting.

This would have the benefit that we remove most of the "bad" conspiracy theoretical answers, while keeping the ones which would be "really good if they were not based on conspiracy theories". It is then left to the community to appropriately vote on such answers. Of course this has similar implications as the complete removal of such answers when it comes to "outsider"'s perspective on us, i.e. we could expect more accusations to be supporters of the conspiracy, even though there probably would be less than with the full-on approach.

• This is exactly what I would like to see. If an answer has a bit of a conspiracy feel to it, then let it stay (and get downvoted), but if the substance of the answer is nothing more than a conspiracy and not a real answer, then delete it. – forest Mar 22 '18 at 18:15

Remove such answers more likely if they fail other criteria and convert them to comments on the question instead.

That is a question, which solely relies on conspiracy theories to make its point and fails some other objective criteria to be a good answer, e.g. only tangentially or incompletely answering the question, would be turned into a comment under the question. Flags would be decided more on a case-by-case basis again.

The implications being that we remove answers that are only based on conspiracy theories and with that the ability to downvote them. But we also take the possibility of them being presented as a valid, potentially good answer. Of course this would be the least likely approach to get us targeted as conspiracy supporters as we keep the content around, just at a different place. This would also mean that we push (partial) answers into the comments.

Remove answers which are exclusively relying on conspiracy theories.

That is, even if an answer fully addresses the question it will be removed if it solely relies on conspiracy theories to make its point. This also means that all Not-An-Answer and similar flags will be approved on such answers.

Of course, this will keep the site "clean" from such conspiracy theories, but then a similar result is achieved using downvotes which hide answers below a threshold score and sort them at the bottom of the page. This would also mean that we as a site are more likely to be a target to accusations of being supporters of conspiracies which may scare away new users.

• I think this is a dangerous option, because who is to say what is and isn't a "conspiracy theory". The term "conspiracy theory" has been used to smear legitimate information that wasn't group consensus when it was brought up, but in certain cases became established and publicly accepted fact years later (e.g. NSA mass surveillance + Snowden disclosures). That isn't to say that totally unfounded + paranoid accusations should be fostered/encouraged, but using "conspiracy theory" to control what can/cannot be discussed at all is a bit dystopian for my tastes (even if the answer is wrong). – Ella Rose Mar 22 '18 at 17:44

I hope that no one minds if I repost a couple of things as an answer $\smile$ I believe it deserves consideration, referring to the removal of answers and general tin foil hatness:-

"I think this is a dangerous option, because who is to say what is and isn't a "conspiracy theory". The term "conspiracy theory" has been used to smear legitimate information that wasn't group consensus when it was brought up, but in certain cases became established and publicly accepted fact years later (e.g. NSA mass surveillance + Snowden disclosures). That isn't to say that totally unfounded + paranoid accusations should be fostered/encouraged, but using "conspiracy theory" to control what can/cannot be discussed at all is a bit dystopian for my tastes (even if the answer is wrong)." – Ella Rose

and this. Please reflect carefully on the implications of the highlighted word, and how it might relate to the above comment:-

"My problem with tinfoil hats is that conspiracy theories are almost always utter nonsense. Speculation on my beliefs is similarly unhelpful :-)" – Rory Alsop

• Mods can't (and won't) see nor transfer individual votes and certainly won't fake nor tamper with votes when it comes to defining policies or community consensus. – SEJPM Mar 28 '18 at 13:39
• @PaulUszak There's a difference between something that involves a conspiracy and something that lacks evidence or even common sense and goes to extreme lengths to explain something with a perfectly reasonable explanation as being the result of some deep state conspiracy. Yes, it is almost always utter nonsense. If you claim that AES, Blowfish, Twofish, Serpent, RC4, 3DES, Anubis, Camellia are all fatally broken because [insert conspiracy here], you'd be wrong, even though it is true that RC4 is really quite bad. Rory Alsop was not promoting your conspiratorial rants with that one word. – forest Mar 29 '18 at 5:42
• Exactly what @forest said - my "almost" does not refer to any conspiratorial theories Paul, or anyone else across SE for that matter, has come up with. For purposes of SE please strike out the word "almost" in that sentence. – Rory Alsop Mar 29 '18 at 10:48
• Why is the bold text "remove answers" if the body of the text is asking to not remove answers? – forest Apr 2 '18 at 2:01
• – forest Apr 25 '18 at 7:02
• @forest It's meant to mean that things like NIST's Randomness Beacon, random.org and Intel's RDRAND cannot be considered to be random at all. They're probably PRNGs. Do you think it's clear enough? – Paul Uszak Apr 25 '18 at 12:17
• @PaulUszak It's just you taking Rory Alsop's quote out of context that I found amusing. – forest Apr 26 '18 at 13:28