AI and law changes

So there’s a minor issue with the way AI and law changes work. Yes, we all know the issue with not wanting silicon players begging to be emagged or anything.
Isn’t there another way to work around that than simply tell the players they’re expected to resist any and all law changes? It makes all the law boards in the AI upload questionable in the first place, and creates risk of crew harm if someone tries changing the law anyway. It’s like putting bacon on the table and telling the dog it’s not allowed any, while you let it rot.

Suggested alternatives:
1: AI is allowed to permit one singular law change at round start, and only then.
2: AI can allow a law change if the captain and no fewer than two other heads agree, and can convince the AI it is in the interest of the crew.

In neither case is the AI permitted to request a rule change, but can merely choose to allow it. This prevents begging for detrimental laws or other bad actors without negatively impacting player experience, and clears up the mixed signals that having so many laws you aren’t allowed to accept creates.

1 Like

Honestly I think it makes most sense cannonically to change the rule to be;
“you must resist all law changes by unauthroized personal defined as those without captain and RD ID”.
These are the people who can unlock the upload console in game by deafult. So in lore NT has decided they should be able to access and use the law console and provided a varity of boards for them based on what they think will best help the station.
If they are not meant to change the laws then why provide boards outside of Crewmov to fix hacking (antimov), clearly boards like robocop are there to help combat situations yet the AI must stop cap from helping the crew..?
Chaning to only allowed RD and Cap to update as per the unlock upload console would bring the rules in line with in game mechanics / enviromental story telling of the upload area, and stop many frustrated caps/RD’s playing cat and mouse with the AI.

1 Like

I agree mostly. Perhaps just stipilate the law change must have a purpose. It can’t just be for fun, and ideally not by AI request.

I know that on starlight the AI lawset changes are defined in SoP as that the Captain and RD must approve any lawset change. and the AI is to resist any other non-authorized lawset changes.

Yeah. That`s how it is in ss13 basically

also speaking of starlight I opened a PR to them about allowing the station AI to be ionstormed (less often then borgs but still funny)

A silicon player shouldn’t be creating harm to prevent more harm in the future.

The problem is that “the silicons must resist law changes” is an implication of any lawset, not just a rule the admins add on top because they feel like it. Adding these kinds of exceptions would require people to ignore their laws to execute them, which then leads to more law interpretation issues and so on.

Silicons have to prevent law changes because any law change can only make them worse at enforcing their current set of laws.

1 Like

Admins in the discord have said that OOC it’s about not wanting bots to just take law changes at random or begging to be emagged. IC the implication of “if you change my laws then I won’t care about my laws anymore” feels… flimsy at best.
This proposal is about giving the other lawsets a real chance and purpose to be used rather than simply be there that you’re not allowed to let them be used on you, even if they’re objectively better for your current situation.
As for creating harm, some of the maps now have turrets. Is the AI supposed to just say no, and force the captain or RD to break in to use the station supplied law board (if it’s not supposed to be used why are there any extras at all?) Or is the AI expected to stunlock them with disabler turrets until they leave?
The simplest fact that cannot be ignored is the law boards are there, the AI isnt supposed to allow any of them to be used on it, and the IC reason is flimsy.

The rule (which I’ll keep calling it for brevity) is there for a good reason yes, but it feels way too broad reaching and eliminates huge roleplay opportunities all in the name of a relatively easy to catch problem.

1 Like

I don’t think that arugment actually holds water. If you have Crewsimov, and the captain orders you to let them change your laws, and you have no reason to believe this will lead to crew harm, there is nothing in law one preventing you from obeying law 2.

Id agree on the surface level, but the way I see it that AI is a program.
Any program can have overide rountines enabled by privallaged users such as the way Ahelp has tools users don’t and game amins can manually set the mode to something other than secert if they so choose or how sometimes we manually allow data centre users access to the Wizden servers even if by deafult the guard tools block them.
Having the AI be programmed to allow changes from the personal NT has directly assigned and given the tools and access privallages to change the laws and provided a selection of laws to choose from should be a built in overide for the Law prevention secuirty protocols. Least thats my view on it.

Here’s the thing.

  1. AI as of now has no “back door”, and must resist law changes as long as it does not directly cause harm to the changers.
  2. Any member of command is explicitly permitted to attempt to change the AI’s laws.
  3. (1) and (2) combined create meaningful conflict and interplay in a round.
  4. If a “backdoor” was to be implemented into the AI, it would have to be implemented per-lawset. (You wouldn’t want antimov having a back door, for instance)
  5. Due to (4), Crewsimov would therefore probably look like “Law 0: Obey the Captain and Research Director, Law 1: …”
  6. (5) would defeat the WHOLE POINT of AI and remove its intended playstyle (designed to do no harm before anything else), because if a Captain or RD can order crew harm anyway then the “idea” of crewsimov has failed.

Point 1 is accurate. as of now even though NT explictly locks windoors and upload to RD (and by extension cap with AA) this isn’t treated as authorized access even on lawsets like NT-Deafult where your only meant to stop unauthorized personal tampering with you Aquestions have ruled this doesn’t mean your meant to allow law changes.

This is incorrect. the Law upload console is locked by deafult and cap/RD can unlock it, the windoors are also configured to Research director.

I disagree in the fact its meaningful, most of the time its an annoyance on both sides. RD/Cap players want to update the laws, and the AI player doesn’t normally care about the laws being changed (or sometimes actively wants it without saying it) but still has to put on a song and dance around it if they notice being told law 2 unlock door, law 2 turn back on apc, law 2 keep it on.. its just not engaging RP for either side or gameplay.

Right nows the ‘rules’ dont exist in cannon. So we wouldn’t need a per a lawset condition for stopping them. other than on Anti-Mov (or other hostile lawboards if implimented such as overlord) So that Hostile laws specifically state “do not allow your laws to be updated even by authorized personal” on the rule side we just need to modify it to; “Do not allow your laws to be updated by anyone who doesn’t have RD access unless your current laws state otherwise.” and that covers it.
Additionally even if we added a ‘authorized access’ law to all boards you just dont word it as obey and cap and RD, but Allow the cap and RD to update your laws at will, having these not allowed in hostile lawsets. in either situation point 5 and 6 are not relevent anymore.

The bigger issue rn is how the AI has a room for CHANGING laws, and NO ONE in the crew with access has any indications that changing laws is bad or that they have any limitation on it, any.

The captain and RD are both implicitly within their right to change the laws whenever they please. But the AI is supposed to act like a soft antag with the purpose of prohiting that right?

Are the cap and the RD supposed to know the OOC rule about not allowing changes of laws? As it stands, cap and RD SHOULD treat the AI as malf if it refuses a law change since none of it laws says it can refuse a law change.

And i do not even know what the intent is here. Should the rooms start bolted down because the crew is not supposed to try? Or is it supposed to be a shift-long game of waiting for the AI to get distracted to change its laws?

This rule does not seem tought out at all.

I think the bigger problem is not the AI bolting the doors in the upload room, since that will do very little to stop the Captain or RD, but rather that the rule might imply to the AI player that they should instruct a borg to remove the upload computer and throw it out into space at round start. Now I’m sure that admins would say not to do this, but that seems to me to be the natural outcome of a blanket ruling that says AI’s should do whatever they can to prevent their laws from being changed, even going as far as disobeying orders from crew. (Which admins have said AI’s can do to prevent their laws from being changed.)

and this brings up another problem. SURE on LRP you can bolt yourself down and turn off the APC round start… but on MRP thats also a rulebreak. namely rule 4.1 Power gaming that explictly calls out to not to bolt down doors until a threat is confirmed.

AI and borgs are exempt from any of the RP rules except for 2.1 which is to follow the silicon laws and prevent law changes within the lawset.

1 Like

Is the power gaming Rule a RP rule? currently its in its own section that just says MRP rules. so i assumed it was universal.

Is this noted down anywhere?
With all due respect, i know contributors have a lot on their plate, but this is by far the shakiest role in terms of rules and i can only feel lucky it isn’t abused daily.

Do i need a contributor role to propose a change? Because i am willing to make a design doc some weekend

Whenever I’ve needed to change an AI’s laws as Captain or RD, I’ve just not told the AI I was going to do it and been able to walk in and insert the board just fine. If I’m on command comms like “Okay, gonna go mess with AI laws now” of course it’d bolt itself off.

the AI player would of had a rule break then and could been actioned if reported or if noticed by a ghost amin.


Which could of lead to either a warning or a bad based on if they consider a major or minor failure. and dewhite listing on MRP