[RULE CHANGE] Markup and Boilerplate

[Haymarket Riot]

*** Public consultation on proposed rule ***

The GA Secretariat proposes introduction of a new rule as follows, directly under the existing language rule:

Markup and boilerplate: Proposals may not contain improperly parsed BBCode, Markdown (eg "**Article 1**"), or other markup languages. Nor may proposals repeat boilerplate repeal or category text in the style generated by the game as headers or footers.

Many generative AI programs output text in Markdown format. Players have repeatedly, in the past, pasted such output as proposals (eg, from both the GA and SC, 1, 2, 3, 4, 5). This rule is in part intended to filter out such proposals, which may nevertheless be legal if generated with nebulous generalities. (We are under no impressions that this will filter out all AI generated proposals.) The copying of boilerplate into resolution text, causing it to be repeated (eg 1, 2, 3, 4, 5) is also a practice which would not occur under the presumption in-universe that the GA would have professional drafters or counsel.

The Secretariat requests comment on:

1. whether this rule would be burdensome for new authors to follow,
   
   
2. what this rule should be called if not "Markup and boilerplate", and
   
   
3. whether this should be split into two separate rule entries.

We would appreciate comments and responses to be topical.

Those voting in favour of the given wording were Imperium Anglorum, Desmosthenes and Burke, The Ice States, Barfleur, and Haymarket Riot. Separatist Peoples did not express a view on the change. Under the procedures, this comment period will end in two weeks (31 Oct. 2024), subject to finalisation.

[Tinhampton]

Is this supposed to be GenSec's proposed anti-AI rule, or is it there while a more permanent solution is figured out?

what this rule should be called if not "Markup and boilerplate"

"Formatting" would be the most obvious alternative name. Should the decision be taken not to change the name, I'd note that rules have historically capitalised the opening letters of each word for emphasis: "House of Cards" is not called "House of cards" and "Game Mechanics" is not called "Game mechanics," for instance.

[The Ice States]

Is this supposed to be GenSec's proposed anti-AI rule, or is it there while a more permanent solution is figured out?

We are still discussing how to act directly with regards to AI proposals, which may or may not materialise in further changes at some point!

[Imperium Anglorum]

I'd note that rules have historically capitalised the opening letters of each word for emphasis: "House of Cards" is not called "House of cards" and "Game Mechanics" is not called "Game mechanics," for instance.

Ugh "title" case. I suppose it should be in title case for consistency.

[Simone Republic]

I guess in terms of the name, either "formatting" or "boilerplate" is fine. Maybe also specify that no code is allowed to exploit a bug, such as that list code bug.

The rules are burdensome for new authors to follow anyway. If your preferred game is not a simulation of a Congressional committee trying to make changes to the US Legal Code (or something similar), I gather Helldivers 2 or The Last of Us look quite attractive. Or Animal Crossing.

[Haymarket Riot]

I guess in terms of the name, either "formatting" or "boilerplate" is fine. Maybe also specify that no code is allowed to exploit a bug, such as that list code bug.

The rules are burdensome for new authors to follow anyway. If your preferred game is not a simulation of a Congressional committee trying to make changes to the US Legal Code (or something similar), I gather Helldivers 2 or The Last of Us look quite attractive. Or Animal Crossing.

The “list code bug” is not an exploit (it’s most often in a resolution by mistake) and not something I am personally interested in regulating, given it’s a tech issue the admins have said is on their list of things to fix, and is frankly also not a hallmark of the thing we’re mostly trying to somewhat regulate here (poorly written proposals written without engaging with the community and its standards, whether AI- or human- generated)

[Desmosthenes and Burke]

Just echoing Hay, I have no interest in punishing authors for errors in the site's code.

[Bisofeyr]

If there is a superfluous [list] BBCode marker or similar in a proposal, would that run afoul of this proposed rule? Under the current reading, I do not see how it would not.

[Haymarket Riot]

If there is a superfluous [list] BBCode marker or similar in a proposal, would that run afoul of this proposed rule? Under the current reading, I do not see how it would not.

That's a correct reading to me. This would encourage authors to be fastidious about a formatting a proposal before submission.

[Bisofeyr]

Thank you for the note!

I have to say, I'm not a fan of the proposal. I understand it, and also get frustrated with AI proposals and think measures ought to be put in place to limit them. However, I feel that the restricted actions here should not warrant an illegal proposal. I think that BBCode typos should not be punished: it is not GenSec's responsibility to police which typos are okay and which are not. Markdown, while officially a markup language (and the one used by ChatGPT, notably), is often used in day-to-day typing, such as to give *emphasis* on a certain word. Not something I personally would like to be in WA legislation, but not something that I think should disqualify an otherwise-excellent proposal from passing.

On the "boilerplate" part of this, I don't really see the harm in users doing this, especially when it is often done as an honest mistake on the part of newcomers. Again, it should be something shot down at vote. I find the justification given in the OP ("[it is] a practice which would not occur under the presumption in-universe that the GA would have professional drafters or counsel") to be unconvincing, as holding authors to this standard makes the WA less accessible (even if it's just an in-universe justification and not indicative of further rules being introduced to reduce this accessibility).

I also have a question on example [5] given on the justification for the boilerplate justification, as that is not a one-to-one recreation of gameside text, and implies to me that this extends into author-generated "long titles" (which is an unconventional standard, certainly, but based on the text of the rule is not a natural conclusion, to me). I guess my question is if some clarification can be given on what the line is, here? If a proposal with category "Regulation" has a preambulatory clause that states "The World Assembly... seeking to enact uniform standards that protect workers, consumers, and the general public, therefore...", is that illegal under the new rule? It should be, if my reading is correct, even if it's not in the same formatting of the text given.

---

Unrelated to my opposition, though, if this goes forward I would agree that these should become separate rules under the "Format" section of the rules, as they are largely unrelated to each other.

[Haymarket Riot]

Thank you for the note!

I have to say, I'm not a fan of the proposal. I understand it, and also get frustrated with AI proposals and think measures ought to be put in place to limit them. However, I feel that the restricted actions here should not warrant an illegal proposal. I think that BBCode typos should not be punished: it is not GenSec's responsibility to police which typos are okay and which are not. Markdown, while officially a markup language (and the one used by ChatGPT, notably), is often used in day-to-day typing, such as to give *emphasis* on a certain word. Not something I personally would like to be in WA legislation, but not something that I think should disqualify an otherwise-excellent proposal from passing.

On the "boilerplate" part of this, I don't really see the harm in users doing this, especially when it is often done as an honest mistake on the part of newcomers. Again, it should be something shot down at vote. I find the justification given in the OP ("[it is] a practice which would not occur under the presumption in-universe that the GA would have professional drafters or counsel") to be unconvincing, as holding authors to this standard makes the WA less accessible (even if it's just an in-universe justification and not indicative of further rules being introduced to reduce this accessibility).

I also have a question on example [5] given on the justification for the boilerplate justification, as that is not a one-to-one recreation of gameside text, and implies to me that this extends into author-generated "long titles" (which is an unconventional standard, certainly, but based on the text of the rule is not a natural conclusion, to me). I guess my question is if some clarification can be given on what the line is, here? If a proposal with category "Regulation" has a preambulatory clause that states "The World Assembly... seeking to enact uniform standards that protect workers, consumers, and the general public, therefore...", is that illegal under the new rule? It should be, if my reading is correct, even if it's not in the same formatting of the text given.

---

Unrelated to my opposition, though, if this goes forward I would agree that these should become separate rules under the "Format" section of the rules, as they are largely unrelated to each other.

People should be putting time into spellchecking their work and I’m not going to call it gatekeeping to ask that they do that. Authors can have resolutions spellchecked on the forums and use NSBBCode for emphasis. I don’t see why this is an issue as you contend.

[Elyreia]

People should be putting time into spellchecking their work and I’m not going to call it gatekeeping to ask that they do that. Authors can have resolutions spellchecked on the forums and use NSBBCode for emphasis. I don’t see why this is an issue as you contend.

I wouldn't place the faith of spell checking on the forum given its proclivity for overlooking things. I'm also curious how strict that will be. I misspell the word decommission all the time and I literally "decommission" assets multiple times a day in my employment. Will a single honest spelling mistake rending otherwise sound legislation illegal, especially if for some reason it has passed the forum's scrutiny to the point of submission, quorum, or vote? And in that case whose fault is it that the spelling mistake was retained, the forum failing to provide correction or the author failing to notice?

Edit: I'm leaving it in, but I accidentally typed "rending" rather than "render". Would this sort of mistake also cause an illegality challenge? That's discouraging a lot of non-native English speakers for the possibility of incorrect words, or worst case, non-standard English for American or British English (e.g. the usage of "do the needful" in some nations where English is a common second language which is grammatically correct but in Western English is reworded as "do what is needed")

The purpose of this rule change appears to me to be aimed at making it easier to shoot down legislation that is obviously written with generative language models, by targeting common styles and formats attributes to them. The simplest (and correct) solution would be to just outlaw generative AI proposals, and provide metrics for measurement. No need to burn the whole forest down when you're just removing a single diseased tree, especially when this starts getting into some dangerous weeds (ADA accessibility, non-native speakers, dyslexia, dysgraphia, non-standard phrasing and notation, the list goes on).

Besides, we also start opening cans of worms. Which English do we use? Valor, honor, color vs valour, honour, colour? I didn't put the period after "vs", is this now illegal? The word "ain't" is an older American and Georgian-era British word comparable to "am not" that's fallen out of favor and considered slang now when it was in the same ballpark as "can't" and "shan't", of which the latter has also undergone a change to "shant" or "shalln't". Casual reminder the American Constitution uses the word "chuse" multiple times in its writing instead of "choose" and this in no way seems to have affected the legality of it, nor required an amendment changing the words (as required by the constitution itself).

Language is a method of conveyance; either spoken, written, or legalese. If the purpose gets across loud and clear, despite whether something was "comissioned" or if "their are honest mistakes", that's just a "whoopsiedoodle", not an earth shattering, world ending, "the law will never work!" situation.

The barrier for entry must be lowered for all, and the barrier for success must be dependent upon the community. If we want to target Generative LLMs making proposals, do that. Don't catch every person trying to use supportive tools or stock designs to ease their learning and engagement.

[Haymarket Riot]

People should be putting time into spellchecking their work and I’m not going to call it gatekeeping to ask that they do that. Authors can have resolutions spellchecked on the forums and use NSBBCode for emphasis. I don’t see why this is an issue as you contend.

I wouldn't place the faith of spell checking on the forum given its proclivity for overlooking things. I'm also curious how strict that will be. I misspell the word decommission all the time and I literally "decommission" assets multiple times a day in my employment. Will a single honest spelling mistake rending otherwise sound legislation illegal, especially if for some reason it has passed the forum's scrutiny to the point of submission, quorum, or vote? And in that case whose fault is it that the spelling mistake was retained, the forum failing to provide correction or the author failing to notice?

Edit: I'm leaving it in, but I accidentally typed "rending" rather than "render". Would this sort of mistake also cause an illegality challenge? That's discouraging a lot of non-native English speakers for the possibility of incorrect words, or worst case, non-standard English for American or British English (e.g. the usage of "do the needful" in some nations where English is a common second language which is grammatically correct but in Western English is reworded as "do what is needed")

The purpose of this rule change appears to me to be aimed at making it easier to shoot down legislation that is obviously written with generative language models, by targeting common styles and formats attributes to them. The simplest (and correct) solution would be to just outlaw generative AI proposals, and provide metrics for measurement. No need to burn the whole forest down when you're just removing a single diseased tree, especially when this starts getting into some dangerous weeds (ADA accessibility, non-native speakers, dyslexia, dysgraphia, non-standard phrasing and notation, the list goes on).

Besides, we also start opening cans of worms. Which English do we use? Valor, honor, color vs valour, honour, colour? I didn't put the period after "vs", is this now illegal? The word "ain't" is an older American and Georgian-era British word comparable to "am not" that's fallen out of favor and considered slang now when it was in the same ballpark as "can't" and "shan't", of which the latter has also undergone a change to "shant" or "shalln't". Casual reminder the American Constitution uses the word "chuse" multiple times in its writing instead of "choose" and this in no way seems to have affected the legality of it, nor required an amendment changing the words (as required by the constitution itself).

Language is a method of conveyance; either spoken, written, or legalese. If the purpose gets across loud and clear, despite whether something was "comissioned" or if "their are honest mistakes", that's just a "whoopsiedoodle", not an earth shattering, world ending, "the law will never work!" situation.

The barrier for entry must be lowered for all, and the barrier for success must be dependent upon the community. If we want to target Generative LLMs making proposals, do that. Don't catch every person trying to use supportive tools or stock designs to ease their learning and engagement.

Misspellings are not what is being addressed here and not what we are focused on. There's a difference between scrutinizing words and using the preview feature to ensure that you don't have any errant tags.

[Elyreia]

I wouldn't place the faith of spell checking on the forum given its proclivity for overlooking things. I'm also curious how strict that will be. I misspell the word decommission all the time and I literally "decommission" assets multiple times a day in my employment. Will a single honest spelling mistake rending otherwise sound legislation illegal, especially if for some reason it has passed the forum's scrutiny to the point of submission, quorum, or vote? And in that case whose fault is it that the spelling mistake was retained, the forum failing to provide correction or the author failing to notice?

Edit: I'm leaving it in, but I accidentally typed "rending" rather than "render". Would this sort of mistake also cause an illegality challenge? That's discouraging a lot of non-native English speakers for the possibility of incorrect words, or worst case, non-standard English for American or British English (e.g. the usage of "do the needful" in some nations where English is a common second language which is grammatically correct but in Western English is reworded as "do what is needed")

The purpose of this rule change appears to me to be aimed at making it easier to shoot down legislation that is obviously written with generative language models, by targeting common styles and formats attributes to them. The simplest (and correct) solution would be to just outlaw generative AI proposals, and provide metrics for measurement. No need to burn the whole forest down when you're just removing a single diseased tree, especially when this starts getting into some dangerous weeds (ADA accessibility, non-native speakers, dyslexia, dysgraphia, non-standard phrasing and notation, the list goes on).

Besides, we also start opening cans of worms. Which English do we use? Valor, honor, color vs valour, honour, colour? I didn't put the period after "vs", is this now illegal? The word "ain't" is an older American and Georgian-era British word comparable to "am not" that's fallen out of favor and considered slang now when it was in the same ballpark as "can't" and "shan't", of which the latter has also undergone a change to "shant" or "shalln't". Casual reminder the American Constitution uses the word "chuse" multiple times in its writing instead of "choose" and this in no way seems to have affected the legality of it, nor required an amendment changing the words (as required by the constitution itself).

Language is a method of conveyance; either spoken, written, or legalese. If the purpose gets across loud and clear, despite whether something was "comissioned" or if "their are honest mistakes", that's just a "whoopsiedoodle", not an earth shattering, world ending, "the law will never work!" situation.

The barrier for entry must be lowered for all, and the barrier for success must be dependent upon the community. If we want to target Generative LLMs making proposals, do that. Don't catch every person trying to use supportive tools or stock designs to ease their learning and engagement.

Misspellings are not what is being addressed here and not what we are focused on. There's a difference between scrutinizing words and using the preview feature to ensure that you don't have any errant tags.

As you'll see from paragraphs 3, 5(partly), and 6, I also address the proposed rule change directly. I'll direct you there.

[The Ice States]

Misspellings are not what is being addressed here and not what we are focused on. There's a difference between scrutinizing words and using the preview feature to ensure that you don't have any errant tags.

As you'll see from paragraphs 3, 5(partly), and 6, I also address the proposed rule change directly. I'll direct you there.

As mentioned earlier, we are still discussing how to proceed with LLMs (a discussion which has been going on, albeit slowly, since December of last year!). The main issue is that it is often difficult to determine whether a proposal was AI-generated, both in terms of false negatives and false positives. A few of us even experimented with AI models and managed to generate multiple proposals which passed both a cursory glance and most AI detectors; there is also a due process issue in that it would often be very difficult for an author to defend against a claim of using AI, even when they honestly did not do so. Prohibiting the use of Markdown in proposals is a much easier standard to enforce, even if it does not eliminate all AI proposals.

I would also add that there is no reason for someone to use Markdown, or repeat boilerplate text, in a proposal; when players do this it is almost always an error, and a fairly easy one to notice at that. Therefore I don't think it's onerous to require authors to be more careful and avoid these sorts of (again, generally obvious) errors.

[Haymarket Riot]

Misspellings are not what is being addressed here and not what we are focused on. There's a difference between scrutinizing words and using the preview feature to ensure that you don't have any errant tags.

As you'll see from paragraphs 3, 5(partly), and 6, I also address the proposed rule change directly. I'll direct you there.

Noted and I understand your concerns. As Mage mentioned, I don't feel GenSec is at a place right now where we're able to find a way to "ban AI proposals" that wouldn't be entirely self-defeating. Point taken on whether this rule change would be burdensome.

[Elyreia]

I apologize if I come off strong in this topic. User and entry accessibility is something my life is focused on. All of my projects are developed to ensure ADA compliance and accessibility; it is a strong focus of my career.

[Haymarket Riot]

I apologize if I come off strong in this topic. User and entry accessibility is something my life is focused on. All of my projects are developed to ensure ADA compliance and accessibility; it is a strong focus of my career.

As someone who strives to be an advocate for disability justice and making NationStates in general more accessible, I appreciate your perspective. It's a hermeneutic blindspot I am not accustomed to questioning. I apologize if my comment came off as dismissive by not responding to much of your post, I don't like to comment where I feel I can't provide a useful reply at this time.

[Attempted Socialism]

I can support the change, but would suggest some rephrasing. First, to improve clarity it should read "... bBCode or markup languages, such as Markdown ..." since the ban is on markup languages generally, and Markdown is itself an example. Next, the example of "**Article 1**" was unclear to me initially, not recalling Markdown notation. A clarification that it's the usage of asterisks, pound-symbol, etc., that is meant as the example given would be good. Such as "(e.g. asterisks to indicate emphasis, such as "**Article 1**")".

I also maintain that the best way to currently deal with AI generation broadly is to ban it as plagiarism, but make an admission of using AI generator the criterion for punishment. AI generated proposals are usually so poor that they would not pass (or even get to a vote) in most cases and in doubt the community can vote down a proposal. That way only those who self-admit to plagiarism are punished (no false positives) and the community has so far been a capable safeguard against suspected instances of AI. Until (/if) some reliable way of AI detection comes along that we as community can agree on that is probably the best way to deal with AI generation as plagiarism.

[Desmosthenes and Burke]

*snipping just the relevant portion*
I also maintain that the best way to currently deal with AI generation broadly is to ban it as plagiarism[...]

This is not on the table. It is beyond GenSec's power and moderation has already made clear their position on the matter.

[Osvarnia]

I can support the change, but would suggest some rephrasing. First, to improve clarity it should read "... bBCode or markup languages, such as Markdown ..." since the ban is on markup languages generally, and Markdown is itself an example. Next, the example of "**Article 1**" was unclear to me initially, not recalling Markdown notation. A clarification that it's the usage of asterisks, pound-symbol, etc., that is meant as the example given would be good. Such as "(e.g. asterisks to indicate emphasis, such as "**Article 1**")".

I tend to agree that this rule could benefit from a few adjustments, although it seems to me that everyone could make the effort to check the terms they have problems with. However, I would be in favour of a certain tolerance for this rule, particularly given the rapid democratisation of AI tools. In my opinion, it would be preferable to start with a clear warning to inform the player that they have broken a WA rule, and to give them a chance to correct themselves. In the event of a repeat offence, stronger measures could then be considered (although this is not my remit).

PS: Having recently joined the WA, I haven't yet mastered all the aspects of how it works. Please bear with me if any of my comments lack precision.

[Attempted Socialism]

*snipping just the relevant portion*
I also maintain that the best way to currently deal with AI generation broadly is to ban it as plagiarism[...]

This is not on the table. It is beyond GenSec's power and moderation has already made clear their position on the matter.

I know. I also find Moderation's stance baffling, and so, not entirely unlike Cato the Elder, I can end all posts on the topic with a suggestion for how to delenda AI.

[Wallenburg]

I am vehemently opposed to style guides as rules, and this leans rather strongly in that direction. That it would be a somewhat effective net for computer-generated text makes it intriguing, but I don't want this serving as a basis for a future GenSec to ban certain writing styles because generative models like to use them.

[The Overmind]

I will say that the community has been pretty effective, so far, at regulating AI use, so I'm not entirely sure this is necessary at this point. If AI becomes a serious problem, with clearly detrimental effects on the community and its engagement, or perhaps just before it reaches that point, maybe more concrete actions, like this one, should be considered.

I say this as someone who I am sure, by now, has developed a reputation for my disdain for undisclosed AI use. But it doesn't seem to be as big of a problem as it could otherwise have been. Generally speaking, the proposal gets posted, the AI use is pointed out, and it drowns in ignominy before sinking into the recesses of the forum forever.

[The Ice States]

*** Markup and Boilerplate rule proposal ***

The Secretariat has decided not to finalise this change in light of community input being largely opposed. This decision was supported by Barfleur, Desmosthenes and Burke, Imperium Anglorum, Separatist Peoples and The Ice States. Haymarket Riot did not express a view.

Our thanks to everyone who gave their input throughout this process!