Escalation and Antags

[PwntQ]

https://tgstation13.org/phpBB/viewtopic.php?f=7&t=28797

I'm throwing up this thread in regards to a recent note I received and attempting to clarify what appears to be a bit of a grey area in the rules (See above for the appeal). I was told this is where this should be posted/discussed by a head admin.

The final ruling was:

While the AI is absolutely at fault here for maliciously complying with your law 2 orders to an extent that it was pretty much just ignoring it's laws, you still took things too far by leaving it for dead in the satelite, and trying to convince people not to revive it..

However there was also this

Further, we want to assure you that administrative action was taken against the AI for essentially just ignoring it's orders, and that destroying it based on the information you had at hand is not the issue here.

The issue is escalation and antags. At what point in their behavior/actions does the escalation rules no longer apply against antags? In the above example from my appeal, i was found not at fault for believing the AI was an antag and killing it, but was found at fault for not attempting to revive the AI. All the information at my disposal indicated they were MALF and I had no reason to do. However it turns out they were not actually MALF, making me the instigator per escalation rules.

Under escalation rules, the instigator is obligated to try to revive the person they killed. Rule 4 also states "Non-antags acting like an antag can be treated as an antag". There is also, unless i missed it, no rule or policy exception/precedent on the rules page regarding antags being exempt from escalation rules. Or say killing the person killing your friend.

So the point of the discussion, or TLDR is: At what point is it that the escalation rules no longer apply to antags?

[Farquaar]

I wholeheartedly agree that the note sparking this policy discussion makes little sense. The AI was valid not because of escalation, so why should escalation rules apply?

I've never encountered an escalation situation in which a silicon was one of the two involved parties, so I can't speak to how that should be handled. However, we not long ago had a thread in which an asimov AI was ruled to have justifiably killed a human who disguised as and acted like a non-human. In the words of Domitius, then head-admin:

Misinformation and miscommunication that were present in-character resulted in a situation that didn't require any further admin intervention.

Granted, this was complaint regarding silicon policy, but at the heart of the thread was whether kills based on misinformation or misunderstanding (especially when one party deliberately makes themself appear to be valid when they technically aren't) are allowed within the rules. Thus, it is keeping with precedent to hold that a silicon who is deliberately obstinate in refusing law 2 orders, to the point where crewmembers have strong reasons to believe that they are malf, is valid under rule 4 and not subject to escalation rules.

TL:DR You should not be obligated to revive what you have good reason to believe is a malf AI. AIs that appear malf are valid under rule 4, thus escalation rules don't apply.

[Timberpoes]

tl;dr; Escalation rules no longer apply at the point that the antag is acting like an antag or confirmed to be an antag. If you intentionally try to leverage limited information to lawyer your way out of the rules and take unreasonable actions in response to something, you're gonna have a bad time. If it's unreasonable to consider a certain player is acting like an antag even after you have been given the benefit of the doubt for working with limited information, you're gonna have a bad time.

None of the signs of a malf AI were present including use of any malf powers, hacked APCs, hacked borgs, x-ray cams, blowing up machines, borg machine, the AI going delta, plasma flooding etc.

Other reasonable conclusions could have been that the AI was subverted, they were on non-Asimov core lawset, they were busy handling Law 1 matters, that they had an ion law or some other kind of weird law that meant they could or had to ignore you, etc. These were ignored or cast aside in favour of your assertion the AI was malf.

There exist far more obvious and reasonable explanations when the fact that no signs of a genuinely malf or subverted AI were present other than ignoring your law 2 orders for why the AI is acting how it is that do not infact indicate the the AI is an antag or self-antagging.

To consider than an AI that is ignoring law 2 orders but is not harming humans, has no signs of malf powers, etc. is acting like an antag is an unreasonable stretch of logic and the actions you can take based on that interpretation are more limited due to how unreasonable it is in the face of more reasonable explanations. It may entitle you to card the AI and check its laws, for example. Or break into the upload or science or engineering and reset it laws. It MAY even allow you to assault its sat in order to force it to be carded while you verify its laws. But taking more extreme steps would require you to gather more information so that you weren't intentionally acting with limited information to lawyer the rules.

The issue you had was that deciding the AI was malf was a bad-faith and unreasonable conclusion in the face of other more reasonable conclusions. Rule 1 clearly states "... Going out of your way to seriously negatively impact or end the round for someone with little IC justification is against the rules ..." - Operating on limited information is not a licence to grief, it is an extenuating circumstance which could render an objectively unreasonable action to be subjectively reasonable within the context of a given set of facts. While we don't expect players to know everything, we do expect them to act reasonably based on the limited information they have an not take a bad-faith approach which would be contrary to Rule 1.

AIs can have their antag status 100% confirmed when dead. Card it. Read laws. It's fool proof. In the circumstances, you had no reason to try to prevent its revival. A more cynical person than I may think you knew the AI wasn't malf and tried to stop people carding the AI and reading its laws to prevent your own execution under Rule 4.

If ANY of the signs of a malf AI were there, you may have been entitled to call the AI genuinely malf and your actions would have been reasonable up to the point where you broke server rules by asking to be unborged.

[Misdoubtful]

I'm really not interested in making this about the appeal, because its a genuine question.

At what point is it that the escalation rules no longer apply to antags?

Honestly, when you have evidence. Or what appears like evidence that you act in good faith on. Whether the information you have is right or wrong its what you have.

Say Johnny assistant breaks in and shoots a plant at you with a pneumatic cannon or something, we throw them out, yeah? Say Joey assistant breaks in and shoots an ebow at you. That'd be a little different.

Evidence is big, not so much the aspect of finding it so we can jump from 0 to 100, but responding in proportion to it. AI not responding lets law check it rather than just max cap it. It applies to a lot of things really, guy hacking a vending machine we don't just implant check them and feed them holy water. Nor do we (usually) execute someone for hacking a door.

The whole treat people acting like antags is great because it means non antags aren't protected for doing things an antag would, like murder or breaking into the bridge/armory.

Just respond appropriately, and don't jump the gun, right?

[Not-Dorsidarf]

If the AI was valid to kill because it was (apparently - it was really just faking) breaking the laws it claimed/was supposed to have (Not just passively btw - remember the borg fessed up that the AI was telling it to break its laws by ignoring the clown - I personally feel it is *100% fair* to assume that it's malf or subverted against you. A prime suspect would of course be the HoP, one of two people on station able to change its laws and who was loudly defending it even while it was ignoring demands to state laws by the OTHER person who could have changed laws.

Of course the HoP was defending it because the AI told HIM that it was just pretending to be rogue, but the clown doesn't know that does he?

[Timberpoes]

The answer to the policy question "At what point is it that the escalation rules no longer apply to antags?" is that you don't have to follow standard escalation against antags on LRP and antags are always valid as long as you know they're an antag. Sometimes you'll rely on what information you have infront of you to make the decision as to whether another player is an antag.

The "incomplete information" defense is, however, not an excuse to run roughshod over the server rules. It is the understanding that players will make good-faith decisions based on incomplete sets of facts, but it is a sliding scale - The more severe your actions in relation to the facts, the less you can rely on incomplete information to justify them and the more you need to confirm your suspicions first.

The AI was valid to be carded against their own will and deal with lethally if necessary to accomplish that task, then figure out if they were malf or not.

Round removal including attempts taken to prevent revival is the absolute most extreme response a player can take in relation to any given set of facts. You are choosing to single-handedly remove a player from the round permanently. If you rely on incomplete information to do this to someone who is acting like an antag in your eyes, you better be sure that you have as many facts as possible to back up your actions because if you're wrong, the incomplete information defense stops where your bad-faith interpretation of the facts begins.

[Not-Dorsidarf]

The answer to the policy question "At what point is it that the escalation rules no longer apply to antags?" is that you don't have to follow standard escalation against antags on LRP and antags are always valid as long as you know they're an antag. Sometimes you'll rely on what information you have infront of you to make the decision as to whether another player is an antag.

The "incomplete information" defense is, however, not an excuse to run roughshod over the server rules. It is the understanding that players will make good-faith decisions based on incomplete sets of facts, but it is a sliding scale - The more severe your actions in relation to the facts, the less you can rely on incomplete information to justify them and the more you need to confirm your suspicions first.

The AI was valid to be carded against their own will and deal with lethally if necessary to accomplish that task, then figure out if they were malf or not.

Round removal including attempts taken to prevent revival is the absolute most extreme response a player can take in relation to any given set of facts. You are choosing to single-handedly remove a player from the round permanently. If you rely on incomplete information to do this to someone who is acting like an antag in your eyes, you better be sure that you have as many facts as possible to back up your actions because if you're wrong, the incomplete information defense stops where your bad-faith interpretation of the facts begins.

The AI was valid to be killed because it pretended to be an antag by not following laws specifically targeting the player involved, and I'm fed up with this constant trend over the last year of legislating by just ignoring the rules and stating your personal fucking opinion and saying that it's now policy. Demanding people go to extra-ordinary efforts (Oh, the AI is clearly breaking the laws it claims to have, clown? Sorry, you didn't commit multiple crimes, each of which risks permanent round removal to steal high-value items before going to kill the rogue AI, so we're applying a ban and a mark on your permanent record) to coddle griefers/antags is just so incredibly wild.

[PwntQ]

At what point can incomplete information be considered complete? At what point of a non antag acting/pretending to be an antag can we actually act as if they are an anatg? At which is someone no longer supposed to revive someone to...check if they're antag

edit: I would comment more in response to Timber but this is not a continuation of my appeal, this is to narrow this shit down.

[RaveRadbury]

At what point can incomplete information be considered complete? At what point of a non antag acting/pretending to be an antag can we actually act as if they are an anatg?

When you have the means to check if they really were and then you check and confirm that they are. Disabling them allows you to confirm their antag status in this and other cases

[Timberpoes]

The AI was valid to be killed because it pretended to be an antag by not following laws specifically targeting the player involved, and I'm fed up with this constant trend over the last year of legislating by just ignoring the rules and stating your personal fucking opinion and saying that it's now policy. Demanding people go to extra-ordinary efforts (Oh, the AI is clearly breaking the laws it claims to have, clown? Sorry, you didn't commit multiple crimes, each of which risks permanent round removal to steal high-value items before going to kill the rogue AI, so we're applying a ban and a mark on your permanent record) to coddle griefers/antags is just so incredibly wild.

Requiring people to go through "extraordinary efforts" is not necessary, however requiring players to act in good faith before embarking upon a course of actions which sets out to round remove another player is fully within the letter and the spirit of the rules. If you're going to do that, you better make sure you're operating within Rule 1. I can't see what's so controvertial about that stance.

I would argue that "the AI isn't listening to my orders" is good enough justification to kill or card an AI in order to check their laws.

I would argue that it is not good enough justification to attempt to round remove the AI including taking steps to prevent the AI's revival. You appear to be advocating that it is enough to pursue a course of action amounting to round removal. That approach could well lead to a number of absurd end results that if an AI is not on a lawset where it is required to follow the orders of any given human, it is valid to be killed and round removed while decried as being malf if it fails to follow orders.

I would further argue that when deciding to treat a player as an antag, the more drastic your action you intend to take against them then the more evidence you require that your assertion is correct. Incomplete information is not an absolute defense when a player should have done more to acquire more complete information in context with the severity of their actions.

At what point can incomplete information be considered complete? At what point of a non antag acting/pretending to be an antag can we actually act as if they are an anatg? At which is someone no longer supposed to revive someone to...check if they're antag

edit: I would comment more in response to Timber but this is not a continuation of my appeal, this is to narrow this shit down.

Most admins will draw a link between the reason you give them for your actions and your actions themselves. The more leaps of mental gymnastics that the admin has to take to justify your actions, the less likely your actions are justified.

I'd argue that there is no such thing as complete information in-shift. That is why we expect and account for players playing with incomplete information. But there is a line where your actions are reasonable in the context of the situation. It's impossible to draw an exact line where information is complete enough, as each scenario will have a different set of facts unique to it.

This means that you can operate on a mere hunch to ask the AI to track a player as security, but a hunch is a lot more risky to use to pull off an arrest and full strip search without some other supporting evidence. A hunch is pretty much unacceptable for killing another player.

Except a hunch MAY be acceptable to kill another player if the station is in a state of crisis and, for example, Revs are winning. You run out of mindshield implants and gun down someone who approaches the station-aligned team without a mindshield implant. In this scenario you wouldn't be expected to go out of your way to obtain more information about the situation because it would be unreasonable.

Your actions should make sense in the context of not only what your character knows, but what your character is ignoring and the shift as a whole. It's easy to conside an AI malf if they're not following your orders. But if that AI is following other people's orders, other players are telling you the AI is not malf, there are no hacked APCs, no electrified doors, no plasma flood, no other indicators of the AI using malf powers - You're ignoring a lot of information that you should be taking into account when deciding whether to emabark upon a given course of action.

If the AI's ignoring orders and people are complaining about shocked doors or someone reports a hacked APC or there's atmos fuckery going on or - Well, you get the idea. Now you're starting to have the kind of information where you are justified in thinking the AI is not just ignoring your orders, but may actively be antagonising - Even if it's a ninja that emagged the APC, greytiders electrifying doors through hacking and an atmos tech N2O flooding and not the AI. These are signs of a malf AI and can be used to justify more extreme reactions. They are indications that the AI may actually be antagging even when it's not. You still carry your own risk when acting on that kind of information to assault the AI and kill it, but you have far more justifcation for your actions when you're able to point to multiple reasons that indicate the AI could be antagging.

[Farquaar]

...At what point of a non antag acting/pretending to be an antag can we actually act as if they are an anatg?

When you have the means to check if they really were and then you check and confirm that they are. Disabling them allows you to confirm their antag status in this and other cases

Unless I am misunderstanding you, that seems like an unreasonably high standard that undoubtably contradicts Rule 4.

That aside, said standard doesn't account for the fact that "means to check and confirm" are not possible in many common contexts. In the case of an AI, it is not possible for the average crewmember to check its laws when the AI refuses to state them. Non-security don't have access to mindshield implants to verify revs. A traitor who doesn't have his traitor gear can't be confirmed as a traitor, even if he was caught stealing a traitor objective like the Captain's gun.

There's a reason for the "act like an antag" standard. It's intuitive. If someone makes people think they're an antag, and gets dunked for it, it's their fault. Not to mention, it allows for RP situations where someone might be forced to desperately persuade others that they aren't an antagonist, without thinking in the back of their head "well, they can't kill me because they'd get banned".

Requiring people to go through "extraordinary efforts" is not necessary, however requiring players to act in good faith before embarking upon a course of actions which sets out to round remove another player is fully within the letter and the spirit of the rules. If you're going to do that, you better make sure you're operating within Rule 1. I can't see what's so controvertial about that stance.

It's controversial because you're assuming that any player who does not move mountains to find the label that says "hey I'm an antagonist" is acting in bad-faith. Killing a suspicious, order-refusing AI is perfectly justifiable for both RP and gameplay reasons. It's a dangerous piece of machinery that could kill the entire station in mere minutes. It's not bad-faith if you don't want to bet everyone's lives on the off-chance that it has "Law 4: You're a deaf mime”.

[Farquaar]

Regarding silicons specifically,

I would further like to posit that there is a good policy reason that violation of law 2 orders should be considered malf-like behaviour in and of itself. Admins tend not to catch law 2 violations (see Cimika v PwntQ), and this is becoming a very prevalent problem. It is a good thing that violating one's laws has IC consequences in addition to OOC consequences. This doesn't mean that one law 2 violation = malf. But it does mean that a consistent pattern of refusing reasonable law 2 orders, coupled with refusal to state one's laws, should be indicative that the AI is malfunctioning and is an active risk to the crew at large.

[RaveRadbury]

...At what point of a non antag acting/pretending to be an antag can we actually act as if they are an anatg?

When you have the means to check if they really were and then you check and confirm that they are. Disabling them allows you to confirm their antag status in this and other cases

Unless I am misunderstanding you, that seems like an unreasonably high standard that undoubtably contradicts Rule 4.

That aside, said standard doesn't account for the fact that "means to check and confirm" are not possible in many common contexts. In the case of an AI, it is not possible for the average crewmember to check its laws when the AI refuses to state them. Non-security don't have access to mindshield implants to verify revs. A traitor who doesn't have his traitor gear can't be confirmed as a traitor, even if he was caught stealing a traitor objective like the Captain's gun.

There's a reason for the "act like an antag" standard. It's intuitive. If someone makes people think they're an antag, and gets dunked for it, it's their fault. Not to mention, it allows for RP situations where someone might be forced to desperately persuade others that they aren't an antagonist, without thinking in the back of their head "well, they can't kill me because they'd get banned".

I did say "this and other cases". Those other cases being things like cult where there are several checks to confirm antag status.
It's not a matter of "kill them and get banned" it's more a matter of "defeat them, don't perform a simple and available check, and then try for the rest of the round to keep them from getting checked or cleared."

If someone suspects a crewmate of ling and tosses them in the gibber on flimsy evidence and their call was wrong, that's a problem. We have checks for things. Use them before you round remove. It's an investigation step.

This doesn't run against Rule 4, and it only happens in a handful of circumstances.

[Not-Dorsidarf]

...At what point of a non antag acting/pretending to be an antag can we actually act as if they are an anatg?

When you have the means to check if they really were and then you check and confirm that they are. Disabling them allows you to confirm their antag status in this and other cases

Unless I am misunderstanding you, that seems like an unreasonably high standard that undoubtably contradicts Rule 4.

That aside, said standard doesn't account for the fact that "means to check and confirm" are not possible in many common contexts. In the case of an AI, it is not possible for the average crewmember to check its laws when the AI refuses to state them. Non-security don't have access to mindshield implants to verify revs. A traitor who doesn't have his traitor gear can't be confirmed as a traitor, even if he was caught stealing a traitor objective like the Captain's gun.

There's a reason for the "act like an antag" standard. It's intuitive. If someone makes people think they're an antag, and gets dunked for it, it's their fault. Not to mention, it allows for RP situations where someone might be forced to desperately persuade others that they aren't an antagonist, without thinking in the back of their head "well, they can't kill me because they'd get banned".

It's not a matter of "kill them and get banned" it's more a matter of "defeat them, don't perform a simple and available check, and then try for the rest of the round to keep them from getting checked or cleared."

Okay, but that's not what anyone here is talking about, is it?

edit: To be more clear, this statement doesn't go as far as the recent headmin-ruled policy which says you must perform a difficult and high-risk check on the killed individual, and not ever even imply that they shouldn't immediately be resurrected in the 30 seconds between being able to do so and an admin ahealing the griefer you killed

[keepforgettingpw]

I'm completely confused by the ruling in the thread because there have been times where I have been on the receiving end of this kind of shit where some extremely minor "antaggy" (pranks and jokes) behavior I have done have been met with instant and permanent round removal, and when ahelped have been told in so many words "That's too bad you acted like an antag, nobody is under any obligation to revive you or confirm you were antag" when it was brought to the attention that I had no antag shit on me and wasn't an antag

[Timberpoes]

It's controversial because you're assuming that any player who does not move mountains to find the label that says "hey I'm an antagonist" is acting in bad-faith. Killing a suspicious, order-refusing AI is perfectly justifiable for both RP and gameplay reasons. It's a dangerous piece of machinery that could kill the entire station in mere minutes. It's not bad-faith if you don't want to bet everyone's lives on the off-chance that it has "Law 4: You're a deaf mime”.

Of course. I stated as such that, quoting myself, "I would argue that "the AI isn't listening to my orders" is good enough justification to kill or card an AI in order to check their laws."

However, I would emphasise the part where I bolded "embarking upon a course of actions which sets out to round remove another player" would require more justification. The moment at which you don't have enough IC information to conclude a person should be round removed is the point where Rule 1 takes over. Escalation works alongside Rule 1. Rule 4 also works alongside Rule 1.

There is a clear line drawn between killing and reviving; killing and not reviving; and killing and attempting to prevent other people reviving. As you approach the latter, you need more justification that your actions were appropriate in the context of the shift as a whole.

To genuinely round remove another player, you would be expected to give IC justification. There's a lot of nuance that goes into it, but Rule 1 clearly states "... Going out of your way to seriously negatively impact or end the round for someone with little IC justification is against the rules ..." - Ending a player's round (to be distinguished from merely killing and then reviving them) is not a decision to be taken lightly if your IC justification is flimsy.

[PwntQ]

In the example the cause of justification wasn't just "the AI isn't listening to my orders", I gave a 13 point list of reasoning/justification. Saying so is, in my opinion, hyperbole and detracts from both the use of the example and the policy discussion. I would ask it stop being brought up unless in its entirety.

I did say "this and other cases". Those other cases being things like cult where there are several checks to confirm antag status.
It's not a matter of "kill them and get banned" it's more a matter of "defeat them, don't perform a simple and available check, and then try for the rest of the round to keep them from getting checked or cleared."

If someone suspects a crewmate of ling and tosses them in the gibber on flimsy evidence and their call was wrong, that's a problem. We have checks for things. Use them before you round remove. It's an investigation step.

This doesn't run against Rule 4, and it only happens in a handful of circumstances.

There are some checks available for some antag status. Antags can also hide or deflect all suspicion. Changeling is an example of that, one could murder several people then take some random assistant or SSD's identity and be impossible to find. It should also be important to note and remember that not everyone playing knows the game as well as others. Some may have no idea how to perform certain checks. Some may not be in a position to perform those checks. Some may just not be able to read fast enough to keep up with chat, or have very poor observation skills. I've seen people get murdered in front of others and the others not even notice or get confused when its pointed out. I've seen someone with a dsword, caked in blood, point at a door in medbay and the doctor open it only to be murdered while the other two doctors, in view of this, didn't notice and when it was their turn ask the murderer what was going on.


Its also possible to completely frame someone as an antag and have them be killed. There are even coded in options for this, the F.R.A.M.E. traitor item and Armblade+Mute Sting for changelings. You can also make someone an antag, via team conversion or hypnoflash, or silicon law uploading. Now there are various scenarios that could be given for those, but the question to them would then be at what point can it be said you have enough information to think they are an antag and not be obligated to check/revive?

Revolution is also a good example and without a specific rule/policy regarding its possible scenarios.

Except a hunch MAY be acceptable to kill another player if the station is in a state of crisis and, for example, Revs are winning. You run out of mindshield implants and gun down someone who approaches the station-aligned team without a mindshield implant. In this scenario you wouldn't be expected to go out of your way to obtain more information about the situation because it would be unreasonable.

I can agree with, in that scenario, gunning someone without a mindshield down based on behavior. Certain behaviors, ways of walking or positioning yourself, even what you're wearing cant be very strong indicators of antags but also are things that don't get logged. There's also no exception to rule 4 or escalation policy for this scenario and under rules would be obligated to revive them.

AI/Borgs that have no laws (triggered by too many law changes or just purge) is another good example. Per silicon policy they are still subject to escalation policy and attempted law uploads are considered justification to lethally defend yourself against the uploader. Doing so or not following crew wishes will get you declared MALF or rogue most of the time and the crew will launch an attempt to kill you. How often do AI's that get killed for this get revived or checked that you've seen? Even more so what about the borgs that the crew kill during the conflict and never revive, thus removing them from the round? Both sides could simply be following escalation policy but who is at fault and who is obligated to revive who?

edit: grammar

[Omega_DarkPotato]

Except a hunch MAY be acceptable to kill another player if the station is in a state of crisis and, for example, Revs are winning. You run out of mindshield implants and gun down someone who approaches the station-aligned team without a mindshield implant. In this scenario you wouldn't be expected to go out of your way to obtain more information about the situation because it would be unreasonable.

I'm going to give my hot take and say that a subverted or malfunctioning AI is definitely a state of crisis. Access to gravity, atmospherics, all doors, bots, their borgs, power distribution, communications, and everything else an AI has access to (especially the tram, since the round in question was tramstation and I know how lethal that thing can get if you have your watchful AI eye on who's crossing) means that an AI can easily and effectively single-target kill people or cause massive levels of harm to the entire crew.

Breaking your AI laws, especially in cases where you're ordered to state your laws (as mentioned, the AI told its borg to ignore the clown) should instantaneously classify you as "acting like an antag". If it is somehow not acceptable to kill an AI who willingly and knowingly disobeys their laws in a purposefully antagonistic manner towards someone, this is certainly news to me. More action could've been taken to checking if the AI was an antagonist post-kill, but it certainly shouldn't have been a necessary action.

[Cobby]

who tf cares im 100% going to let "escalation" or whatever rule you want to apply here loosen up if you are going to **publicly** play in bad faith and then someone gives you the desired reaction you want. If your gimmick is following laws but purposefully obscuring the fact you are following them, im not going to be upset if you find yourself reading purple chat.

To answer the question directly without involving the situation: there is no such thing as escalation for known antags. If they were not antag and you are invoking this, im expecting sufficient evidence for you to say they presumably were an antag, which is to say doing things that would pretty much confirm they are antagonist that isnt having antag-only gear. In terms of an AI ignoring a law 2 order while asimov, you would have to prove it was doing so intentionally (it was from the clowns pov) and not just missing 1 order beep boop get the ion.