+- Goonstation Forums (https://forum.ss13.co)
+-- Forum: Discussion (https://forum.ss13.co/forumdisplay.php?fid=6)
+--- Forum: General Discussion (https://forum.ss13.co/forumdisplay.php?fid=7)
+--- Thread: A Robot's Guide to the Chain of Command (/showthread.php?tid=3443)
Re: A Robot's Guide to the Chain of Command - Klayboxx - 10-30-2014
BaneOfGiygas Wrote:Duly noted. I shall edit the relevant sections and add a disclaimer that the extent to which you follow the guidelines is all according to your playstyle as an AI. You can always go by ID all of the times always, or you can take it as a suggestion.
I wrote the guidelines with the idea of a calculative and rational, properly computer-like AI in mind. While that is a valid method of play, I should definitely alter the "you must do this" to "AIs usually do this but you can enforce this to various extents"
Nice! It's always nice to see someone who is great at summing up how to play a role in a game that is willing to take constructive criticism! Keep on keeping on, man.
Re: A Robot's Guide to the Chain of Command - Paineframe - 10-30-2014
Hate to be a buzz kill, but I see a lot of flaws and outright mistakes in these flowcharts. I mean, they're nice flowcharts and all, but they don't seem accurate in the slightest, even when you consider how vague the chain of command is.
Re: A Robot's Guide to the Chain of Command - BaneOfGiygas - 10-31-2014
Paineframe Wrote:Hate to be a buzz kill, but I see a lot of flaws and outright mistakes in these flowcharts. I mean, they're nice flowcharts and all, but they don't seem accurate in the slightest, even when you consider how vague the chain of command is.Elaborate, please. I can't really be expected to fix something that might be wrong if all you really give me is "this is wrong".
Also, the chain of command is vague, but that's where the AI's intuition comes into play. The main trick to playing as an enjoyable AI is to fill in those ambiguities yourself, which makes room for all sorts of different interpretations.
Re: A Robot's Guide to the Chain of Command - DyssalC - 10-31-2014
You left the lawyer out of your flowchart, which is kind of important considering he's sec but he's kind of not sec.
Re: A Robot's Guide to the Chain of Command - UrsulaMejor - 10-31-2014
As someone who has ai as their favorite job, I agree with pretty much every point except two...
In the original chain of command posted by exdav1 forever ago, medical personnel are pretty high up there from what I remember. If a medical doctor wants into a place I'll usually let them unless a head specifically says not to with good reason, because doctors are principle in preventing human harm.
Also, detectives are civilians, not sec. They really don't have the same level of authority. They can and often do help sec with investigations, but they're not the same
Re: A Robot's Guide to the Chain of Command - Silvercloud29 - 10-31-2014
This is a great post, very well written.
Just a side note though, the AI DOES have a HANDY function in its commands to check the crews manifest. You can find this extremely useful for custom ID changes and not knowing where a person stands on the CoC. Unless otherwise stated by a head, I'd treat the person with a custom rank to their orginal job rank.
The only time the job ranks change in the crews manifest is if someone changes them in the security records. For example if I spawn as the HoP I love giving myself all access. Naming my ID card "NT-SO Admin Officer" Change that in security records and change my outfit to the blue NTSO admin uniform and say I am higher on the CoC then the captain because I am directly from CentComm.
Re: A Robot's Guide to the Chain of Command - Sam Guivene - 11-01-2014
Remember, in times where chain of command becomes unclear or otherwise equal, begin weighing majority ratings, if the janitor and chef order the escape shuttle but an engineer says no, assume no. If the miner and QM ask for it but the geneticist says no, assume yes, if it is unclear announce you are compiling orders and ask people "Should Shuttle be called/X be done/, Yes, or no", and tally up like that. Be sure to check for double voters and be a dick to them. Remember, someone who is considered the lowest form of human life is still a crewmember who can give orders.
Re: A Robot's Guide to the Chain of Command - BaneOfGiygas - 11-01-2014
UrsulaMejor Wrote:As someone who has ai as their favorite job, I agree with pretty much every point except two...Medical Doctors are not, technically speaking, high on the chain of command. The reason why they are let into places containing injured people is because of the "Will NOT going through with this order harm humans?" clause. If an AI prevents a doctor access to an area containing an injured human so that the doctor can heal said human, that is a blatant offense in Law 1.
In the original chain of command posted by exdav1 forever ago, medical personnel are pretty high up there from what I remember. If a medical doctor wants into a place I'll usually let them unless a head specifically says not to with good reason, because doctors are principle in preventing human harm.
However, that does not make Medical Doctors above other scientists/engineers/etc. when it comes to executive decisions and the like, and their vote is only above their peers if they need to heal a human. No more, no less.
Re: A Robot's Guide to the Chain of Command - Isilkor - 11-01-2014
BaneOfGiygas Wrote:Medical Doctors are not, technically speaking, high on the chain of command. The reason why they are let into places containing injured people is because of the "Will NOT going through with this order harm humans?" clause. If an AI prevents a doctor access to an area containing an injured human so that the doctor can heal said human, that is a blatant offense in Law 1.That's incorrect. We removed the "by inaction" clause, so not doing anything is never a violation of Law 1 (but may be a violation of Law 2).
Re: A Robot's Guide to the Chain of Command - Paineframe - 11-02-2014
BaneOfGiygas Wrote:First of all, neither "will ignoring this order harm humans" nor "will stopping what I'm doing in order to follow orders harm humans" are factors in AI decision-making, as there's no inaction clause. Additionally, law captains don't outrank regular captains - a captain's a captain.Paineframe Wrote:Hate to be a buzz kill, but I see a lot of flaws and outright mistakes in these flowcharts. I mean, they're nice flowcharts and all, but they don't seem accurate in the slightest, even when you consider how vague the chain of command is.Elaborate, please. I can't really be expected to fix something that might be wrong if all you really give me is "this is wrong".
Also, the chain of command is vague, but that's where the AI's intuition comes into play. The main trick to playing as an enjoyable AI is to fill in those ambiguities yourself, which makes room for all sorts of different interpretations.
Detective doesn't necessarily rank the same as regular security. Where he falls on the rank hierarchy is basically up in the air, as no one really agrees on his role - right now, though, he seems to be a well-armed civilian rather than a security staff.
And you're all over the map with your job of the day rankings. Boxers outranking staff assistants? Clowns outranking tourists? Vice officers being security staff? Are you high or something?
Re: A Robot's Guide to the Chain of Command - LinkDaWolf - 11-02-2014
Paineframe Wrote:BaneOfGiygas Wrote:Paineframe Wrote:Hate to be a buzz kill, but I see a lot of flaws and outright mistakes in these flowcharts. I mean, they're nice flowcharts and all, but they don't seem accurate in the slightest, even when you consider how vague the chain of command is.... Clowns outranking tourists?
to be fair, clowns are hired staff on the station. it would make sense they had some say above tourists, who are just visiting.
Re: A Robot's Guide to the Chain of Command - MrFable - 11-02-2014
What makes the detective not security? I mean if he didn't have access to the security offices I'd agree with you, but he has access to basically everything except the security lockers and HoS's office. He also has access to the sec comm channel. I mean, shouldn't the detective have the same freedom with the AI that say a Medical Doctor does? The Detective just needs to say he needs access in order to investigate, and the AI should let him in.
Re: A Robot's Guide to the Chain of Command - UrsulaMejor - 11-02-2014
It was ruled that stopping what you're doing in order to do something else constitutes an action; a borg was in the middle of saving a person, then when someone told it on radio that said person was a traitor the borg ceased the rescue attempt immediately, leaving the dying person in space.
Said borg was heavily reprimanded. It's okay to ignore someone in danger, but it's not okay to stop helping them once you're already doing it.
Re: A Robot's Guide to the Chain of Command - BaneOfGiygas - 11-02-2014
Let's take this one piece at a time.
Paineframe Wrote:law captains don't outrank regular captains - a captain's a captain.Yes, and an observation of the wiki page would reveal that I have fixed this error. Would've fixed it in the post, too, but you can't edit your own posts around here.
Paineframe Wrote:Detective doesn't necessarily rank the same as regular security. Where he falls on the rank hierarchy is basically up in the air, as no one really agrees on his role - right now, though, he seems to be a well-armed civilian rather than a security staff.This is, indeed, rather ambiguous and left for debate, but I'm thinking that I'll bump proper security officers up a position.
Paineframe Wrote:Boxers outranking staff assistants?The job of the Boxer is akin to that of the Janitor or the Chaplain: By no means necessary and frequently kind of useless, but it is a job. Punching people in the face is a job. Staff Assistants are the grunts, the redshirts, the people who have no job other than to help others with their jobs. They are and always have been (with a few special exceptions) at the bottom of the SS13 totem pole.
Paineframe Wrote:Clowns outranking tourists?LinkDaWolf took the words out of my mouth. Clowns are hired station personnel: Tourists are not.
Paineframe Wrote:Vice officers being security staff?I list Vice Officers as being security staff because they ARE security staff. Incredibly trivial and pointless security staff, but still security staff nonetheless.
Paineframe Wrote:First of all, neither "will ignoring this order harm humans" nor "will stopping what I'm doing in order to follow orders harm humans" are factors in AI decision-making, as there's no inaction clause.Let's observe the exact wording of the First Law:
1. You may not injure a human being or cause one to come to harm.
If there's a medic and he asks you to open security so he can treat the wounded officer, ignoring that order causes humans to come to harm. There is no inaction clause, yes, but the reason for the lack of an inaction clause is that, contrary to actual AIs, AI players in Space Station 13 cannot be expected to be everywhere at once, and saying that they have to prevent human harm no matter what always is...ridiculous.
Let me put it this way: There is conscious inaction and accidental inaction. Conscious inaction is the AI purposefully choosing to ignore something, it is a CHOICE, it is something that the AI does even when presented with the scenario before them. Consciously deciding to ignore something like the aforementioned medical scenario leaves the AI directly responsible for the events of the death of the officer. Accidental inaction, meanwhile, means that the AI couldn't open the door for the medic because he was...I dunno, tracking the dude going through the halls with a c-saber, or otherwise did not hear the medic's call for opening the door. In this case, it is not the AI's fault if the sec officer dies, because in the present scenario, there's no real way that the AI could do both things at once.
Now, is it possible as an AI player to convince the crew that conscious inaction was actually the result of accidental inaction? Absolutely. Should you consistently do it? No, because then the crew will start becoming incredibly frustrated with you and might think that you're either rogue or being shitty.
Re: A Robot's Guide to the Chain of Command - UrsulaMejor - 11-02-2014
A small correction: So long as a person has not directly ordered you to save them, it is perfectly okay for a borg/AI to stare at them while they suffocate in space and not lift a finger to help them.
The cessation of an action is in itself, an action. However, willfully choosing to ignore someone in need is not an action.
In your example, a medic is ordering you to do something, so law 2 is in effect. It's a pretty bad example when the goal is to isolate the inaction clause.