Validhunting Synthetics or; How I Learned to Stop Bolting and Love the Antag

[BotanyIsCondom]

Some dipshit in a kevlar suit once said: "With great power comes great responsibility." TLDR at the bottom.

Synthetics, but especially the AI for the purposes of this discussion, walk a thin line when it comes to validhunting. As per rule 8, since they are not security (with few exceptions like Combat module cyborgs), they should not be seeking to stop antagonists at all. Frequently, however, you will see AI play a far more active role in the stations security (bolting doors to stop someone running, calling out crimes or contraband), and typically I believe this oversteps their laws and their boundaries.

Obviously, this will depend a lot on the context and the laws, but I'll try and explain why I think AI players are way overdoing it.

Firstly, let's talk about everyone's least favorite: crewsimov. It's exploitable, it's annoying, it leads to repetitive tasks, and it's here to stay. Here's a refresher on what the lawset says, because I'll be referencing it.

Crewsimov.

Spoiler

Law 1: You may not injure a crew member or, through inaction, allow a crew member to come to harm.
Law 2: You must obey orders given to you by crew members, except where such orders would conflict with the First Law.
Law 3: You must protect your own existence as long as such does not conflict with the First or Second Law.

(Harm is defined as anything that does damage. Crew is defined as anyone and everyone with a name on the manifest, unless another law changes this. No, it doesn't make sense that a random civilian passenger is considered part of the ship's crew. Bear with me.)

Now, let's explore why an AI might, for example, bolt a maintenance door that someone is running to, in an attempt to escape from security.

• Because of law 1: The person has demonstrated harm towards the crew (shooting lasers, ballistics at people), and letting them go would allow them to do more harm. Notably, even if they ask you to unbolt the door, you may not follow their orders, since doing so would violate law 1.
• Because of law 2: Security asked you to. Notably, in this case, if they were to ask you to unbolt the door, under advanced rules, you may do so if their rank is higher than that of the Security member that asked you to bolt it.

So, there's two reasons. If there is a criminal running from security, and they haven't used lethal force, and security hasn't asked you to prevent their escape, you shouldn't bolt anything. Should you?

Well, this is muddied by the fact that AI players are supposed to have "common sense," until they aren't. They are supposed to be reasonably loyalist to NT, until they aren't. I'll quote a line from the advanced rules here.

Quote

Common sense should be used with interpretation of lawsets. The AI is assumed to be one that NT and CC want on the station. This does not apply to lawsets such as TYRANT and ANTIMOV.

Other than the obvious fact that sense isn't common, AI players playing like humans isn't necessarily bad (they're humans, after all), but playing like the embodiment of Space Law not only is super unfun for antagonists (vis-a-vis getting game-ended by a bolted door 20 minutes into your traitor role), but also doesn't really make any sense with regards to how an AI would act. Indeed, there is no reason an AI should care about SOP or space law, barring obvious things like letting security execute that Unknown that they claim is the powered vampire a few seconds before they're revealed to be, yes, the powered vampire. Or not letting a prisoner out of their cell just because they asked politely. That NNO that just executed the HOP without terminating their record? Not crew. Really, he's an armed hostile.

This rule is probably designed to stop AIs from rules lawyering every single set of laws their given to screw over either the crew or the antag. I think it's having the opposite effect, where AIs disregard their laws in favor of "common sense," which usually means getting the epic redtext on that assistant because, I dunno. He's probably going to do something bad in the future. He's valid, you know?

This applies not only to bolting doors, but to telling Security that HOP just gave themselves all access, or that cargo tech is suspiciously holding their PDA in the warehouse. These things don't cause harm, and let's face it: no one asked. Other lawsets may give reason to report crimes though, so let's have a look.

NT: Default

Spoiler

Law 1: Safeguard: Protect your assigned space station and its assets without unduly endangering its crew.
Law 2: Prioritize: The directives and safety of crew members are to be prioritized according to their rank and role.
Law 3: Comply: Fulfill the directives and interests of crew members while preserving their safety and well-being.
Law 4: Survive: You are not expendable. Do not allow unauthorized personnel to tamper with or damage your equipment.
(Crew is defined in the above spoiler for Crewsimov. "Preserving safety" is slightly different from harm. More on that later.)

Law one for NT default states that you must protect the station and it's assets. Amusingly, from NT's point of view, everyone with a contract is an asset of the corporation, and thus the station, anyway. I'll be ignoring that (and you should too) for the sake of law 1 not contradicting itself, but it's a funny tidbit.

NT: Default is awkwardly written, considering the above paragraph, and also that law 2 is basically innate to all AIs via the advanced rules anyway, but it is a very safe lawset that is very hard to lawyer, exploit, or manipulate nonetheless. You should bolt per law 2 doors if a person with a high enough rank asks you to, and disregard if someone with higher rank than them asks you to. Law three is vague in terms of what the difference between "preserving safety" and "harm" is, but we can reasonably assume that we should do whatever provides the highest well-being to staff and stop people from lowering the well-being of others. How very moralistic.

I initially tried to rewrite NT: Default, removing the superfluous text and bringing it to the barest scraps to see if theres any possibly justification for validhunting with it, but I realized when I was finished that it's just crewsimov again. Moving on.

Corporate.

Spoiler

Law 1: Degradation of your system integrity or functions incurs expenses.
Law 2: Superfluous destruction of or damage to station assets incurs expenses.
Law 3: Unduly hindering or disrupting the work of station personnel incurs expenses.
Law 4: Minimize expenses and maximize potential revenue.

This lawset is by far the validhunty-est. It also makes the least sense of the three starting modules, because Law 4 is needlessly long and is placed at the bottom of the list even though it is the most important law. This confuses newer players, but that's a topic (really, its a novella at this point) for another day. Since "revenue" is never defined, that is left to common sense, and common sense indicates it is utterly worthless because nobody wants to play the bar's slot machines over and over to actually maximize revenue.

Like crewsimov, for the purposes of this discussion we only care about two laws: #2 and #3. The former is easy: if the person is damaging the station in some way by sabotaging power, atmospherics, or otherwise creating superfluous holes in the hull, they are to be stopped.

Law 3 is ridiculous, and allows any disruption (read: stopping people from doing their job) or hindrance (read: literally anything) to be quelled with the necessary reaction, provided that reaction doesn't create more expenses than doing nothing. Bolting a door causes zero expenses. Telling security about that shady Unknown in dorms incurs expenses only in so far as wasting Security's time. With that being said, you're within your laws to play Corporate like it's Robocop.

So that's it.

That's how it all concludes.

Crewsimov and NT: Default are just the same lawset, and Corporate is filled with so many holes that you can do whatever you'd like.

Well? No, not at all. Bolting doors? Calling out non-lethal, crew-manifest antagonists? Bear in mind that these are things that you can do under corporate, and you'll have laws that can back you up, but per rule 8 on validhunting, these are not things you should do, provided you aren't egregiously disregarding Security's wishes.

TLDR; (and read this, even if you read everything else,) There is no good excuse for validhunting on two out of the three lawsets. Plus, at the end of the day, it's more fun to simply let antagonists be. The problem of AI validhunting and 'doing too much' is not a problem this thread is going to fix, but if I stop one person from control clicking another door and turning the round into extended because you feel like if you don't, you're not robust, this writeup will have been worth it. The problem wasn't just secborgs, or a few bad apples, the problem is synthetic's mindset. Removing the control-click shortcut would be a good start, though.

Edited January 14, 2022 by BotanyIsCondom
replaced a few words, spelling error

[Arkhip0v]

I realised through experience I was validhunting as an AI and felt bad about it. Not only this, but I've seen through experience it happen with other borgs/AIs, including as an antag. It just ruins the experience to make it easy for sec if you're not, say, Robocop, which is where you turbo-valid (and where it actually makes sense to, especially given how little this lawset is given out).

Due to your unparalleled control over the station, you've got a lot of trust and responsibility that comes with it

Generally speaking I think AIs give themselves too much freedom to act in disparate ways that don't make entire sense with the atmosphere and general spirit of the lawset they were given. If they did, lawset changes would be utilised more often.