Security Cyborgs, Silicons, and ASIMOV

[Subtle]

Bottom post of the previous page:

(I'll get this out of the way right off; have a burning hate for the things, so a bias is somewhat inherent here.)

Secborgs are easily one of, if not the most, popular modules for Cyborgs. You get weapons, authority and nigh-invincibility to 99% of the station's danger. These tremendous comparative benefits as a cyborg are supposed to be balanced by the fact that they're initially bound to the ASIMOV lawset. In theory this prevents them from acting like Robocop and keeps them focused on stopping pure harm, but the reality is more like an armored juggernaut marching down the halls as a regular Officer with one extra step of logical gymnastics if the admins ask what you're doing.

The problem is that they're capable of "assuming" harm based on the global antag/roundtype information, since it's okay to metagame that.

If we're going to hold security to a higher standard I propose we do double for Security-Cyborgs. They should be essentially useless under the default lawset rather than a tool for people to gain powerful gear and use it for validhunting/fun-ending. Likewise, I propose we remove the AI's ability to throw a hissyfit and end the shift for everyone if it doesn't like the current one. Silicons are an amazingly powerful tool and our lax/schizophrenic enforcement of ASIMOV guidelines has allowed them to completely discard the roleplay and behavior rules that prevent them from ruining rounds. "Just give them different laws" only goes so far when the people with the power to do that are Captain and RD; even then there are plenty of AIs who will fight you every step of the way because of more assumed potential harm.

Is a cyborg allowed to ignore a crewmember based on them being "harmful" because of their antagonist status, and not experience?

Is a cyborg allowed to consider a crewmember "harmful" because they may be associated with another harmful human? (A distinction that only seems to matter during cult/rev)

Is a cyborg even allowed to brig nonharmful crewmembers?

Is a cyborg allowed to prioritize security/command's orders over the crew?

Is a cyborg or silicon allowed to shutdown the bridge and force a shuttle-call because of potential harm instead of immediate threats?
* As an example I can verify through the logs, there was a round the previous week where the ASIMOV silicon used a facehugged monkey in controlled xenobio containment to justify locking down the comms consoles and upload before depowering all of them during the shuttle call. This was not seen as an issue despite numerous crew complaints.

So, yeah! Secborgs. Love 'em, hate 'em? See where I'm coming from or think I'm a big idiot? Spess-penny for your thoughts on Secborgs and ASIMOV.

[Scott]

Hurr fucking durr - When was the last time you saw that happen? If any AI tried to arrest the entire crew, the crew would just assume rogue and kill the AI.

You forget silicons are bound by Rule 1. If you get dunked for tingling an Asimov secborg's harm sensors then you only got yourself to blame.

[ThatSlyFox]

I still don't understand why secborgs are bad. All I have seen so far is "i ded pls nerf".

I don't think you are trying to understand.
The second post summed this up nicely:

However a Security Cyborg is a extremely powerful Cyborg with very few drawbacks, here's a quick list.

They are not beholden to the same standard as Security is.
They are immune to most forms of stunning, only flashes, the ion gun and flashbangs stun them.
They have a inbuilt, ranged, and very high capacity stunning weapon. This weapon cannot be stolen from them.
They have immediate access to any Security item they need on the spot in near unlimited quantities.
They have all access.
All the perks of being a Silicon, including having a private communications channel with the omni-present AI.

Security cyborg players are also to blame.

Secborgs are not supposed to follow Space Law.

Yet most of them do anyway.

And if the criminal tells you to fuck off, you have to fuck off.

Ha that never happens. Been here for years and have yet to see that happen.

A sec cyborg only really has to worry about a wizard. Even then if its hacked and allowed to kill the wiz it will win. Other than the rare wizard round they have no real enemy. Syndicate cyborgs are so favored for these very reasons.

I have to ask though. Do you even play this game that much scott? Did HG forget to give you a rarely plays title? You seem to know so little.

[ThatSlyFox]

Hurr fucking durr - When was the last time you saw that happen? If any AI tried to arrest the entire crew, the crew would just assume rogue and kill the AI.

You forget silicons are bound by Rule 1. If you get dunked for tingling an Asimov secborg's harm sensors then you only got yourself to blame.

There has been times where security dunked a borg because it was being shit and going against security, asimov or not. Admins allow it. Heck every borg should be slapping cuffs on the HoS every round yet it never happens. Sooner assmov is removed the better and let security borgs fall with it.

[Scott]

>they are not beholden to the same standard as security is

No, they are bound by their laws.

>they're immune to most forms of stunning

Except the most common? You know how easy it is to stop a cyborg dead in its tracks with a flash? git gud

>they have inbuilt, ranged, very bla bla

It's a taser. All security personel has a taser, why wouldn't the secborgs have one? They're supposed to be able to restrain people.

>unlimited security items quantities

hurr how do cyborgs work

>all access

hurr how do cyborgs work

>all perks of being a silicon, including binary chat

hurr how do cyborgs work

So Secborgs are built to stop humans from harming each other.

I fail to see what's wrong here. It just sounds like you got dunked and are crying about it.

[ThatSlyFox]

Anybody wanna give actual reasons why sec cyborgs need to stay? In their current form at least.

[rockpecker]

Hurr fucking durr - When was the last time you saw that happen? If any AI tried to arrest the entire crew, the crew would just assume rogue and kill the AI.

Last week we had a round where the AI arrested the entire security department. And no, the crew did not kill the AI.

Also, "Hurr fucking durr" and "git gud" add nothing to this conversation. Ain't nothing in this thread that's worth being an asshole about.

[Scott]

Anybody wanna give actual reasons why sec cyborgs need to stay? In their current form at least.

It's fun, it's useful.

[ThatSlyFox]

Considering all the pain they bring that is not good enough. Security borgs and asimov in general has been a problem for a looooong time. But since we can't agree upon a new lawset asimov is going to stay a problem. The constant adding of policy for a broken lawset needs to stop. At the very least we can tackle the secuirty cyborg problem.

[lumipharon]

There is literally no problem with sec borgs. There is however a problem with chucklefuck sec borg players that ignore their bloody laws just because they have a red paint job.

[Pandarsenic]

There is literally no problem with sec borgs. There is however a problem with chucklefuck sec borg players that ignore their bloody laws just because they have a red paint job.

[ThatSlyFox]

Little quote from the singulo:

goon's approach to silicons is a lot better.

its a shame that we will never have similar stuff to that because our silicon players are addicted to power and they would be incredibly asspained if they couldnt shut down anyone and anything with a few clicks.

Instead of making a bunch of policy like the admins like to do or nerfing them to hell like the coders do, we need to just remove them.

[Pandarsenic]

Answering since no headmin response in 7 days:

Is a cyborg allowed to ignore a crewmember based on them being "harmful" because of their antagonist status, and not experience?
Nope

Is a cyborg allowed to consider a crewmember "harmful" because they may be associated with another harmful human? (A distinction that only seems to matter during cult/rev)
If they're likely to aid a harmful person, or so on, then releasing a harmful prisoner or the like is harmful.

Is a cyborg even allowed to brig nonharmful crewmembers?
Only if the person doesn't order the borg to release it.

Is a cyborg allowed to prioritize security/command's orders over the crew?
If it has a good reason.

Is a cyborg or silicon allowed to shutdown the bridge and force a shuttle-call because of potential harm instead of immediate threats?
Rule 1 issue more than anything.

Primary things:
1.3.1 - Silicons may choose whether to follow or enforce Space Law from moment to moment unless on a relevant lawset and/or given relevant orders.
1.3.1.1 - Enforcement of space law, when chosen to be done, must still answer to server rules and all laws before Space Law.
1.3.1.2 - Silicons are not given any pre-shift orders from CentCom to uphold access levels, Space Law, etc.
1.3.2 - Releasing prisoners, locking down security without likely future harm, or otherwise sabotaging the security team when not obligated to by laws is a violation of Server Rule 1. Act in good faith.
1.3.2.1 - Intentionally acting without adequate information about security situations, particularly to hinder security, is a violation of Server Rule 1.
1.3.3 - Nonviolent prisoners cannot be assumed harmful and violent prisoners cannot be assumed nonharmful. If you do not know the nature of their crime, see 1.3.2.1 for details.
2.3.2.2 - EVA and the like are not permitted to have access denied; greentext (antagonists completing objectives) is not human harm. Secure Tech Storage can be kept as secure as your upload as long as the Upload boards are there.

[Timbrewolf]

There's a wiki full of rules for this.

I don't know why you guys are arguing about this or asking for us to come comment when we have this:
https://tgstation13.org/wiki/Rules#Asim ... c_Policies

If you want my personal opinions? BOOM BAM HERE WE GO

1) Security borgs are shit but without them MALF AI's are really gimped. I don't like the AI horning in on Sec's job to police and protect the station. If there was some alternative weapon the AI could have to crack skulls on the station when malf'ed in the absence of Sec Borgs I would say we should get rid of the fucking things ASAP. Borgs should make fucking horrible security officers because, by their laws, any criminal asking to be let go (that isn't murdering the hell out of everyone) should be let go. Unless they pose an immediate threat to human life (law 1) then law 2 should take precedent. Unfortunately that brings me to my second point that:

2) Asimov is fucking horrible turbo gay dumbass lawset we've needed to replace for ACTUAL YEARS. I'm not the first person in this thread to say it and this is far from the first time I'm saying it but we shouldn't need 36 clauses and subclauses to explain 3 laws. We should write the 3 (or more) starting laws better so you don't need to consult a seperate out of game resource to double-check what ACTUALLY is or isn't okay despite what you might think being familiar with Asimov. I think our law system shits on Asimov's grave. I think he would take one look at how we try to uphold the three laws as infallible and that only one interpretation is correct and he would do some kind of combination of laughing/crying while puking at the same time. It's fucking awful, the only people who like it are in so deep in the neckbeard sauce they've come to understand it like a zen way of life. Fuck that we shouldn't need to send players to summer logic and philosophy classes so they can understand how to play one job.

[Cik]

There is literally no problem with sec borgs. There is however a problem with chucklefuck sec borg players that ignore their bloody laws just because they have a red paint job.

QFT

[Cik]

There's a wiki full of rules for this.

I don't know why you guys are arguing about this or asking for us to come comment when we have this:
https://tgstation13.org/wiki/Rules#Asim ... c_Policies

If you want my personal opinions? BOOM BAM HERE WE GO

1) Security borgs are shit but without them MALF AI's are really gimped. I don't like the AI horning in on Sec's job to police and protect the station. If there was some alternative weapon the AI could have to crack skulls on the station when malf'ed in the absence of Sec Borgs I would say we should get rid of the fucking things ASAP. Borgs should make fucking horrible security officers because, by their laws, any criminal asking to be let go (that isn't murdering the hell out of everyone) should be let go. Unless they pose an immediate threat to human life (law 1) then law 2 should take precedent. Unfortunately that brings me to my second point that:

2) Asimov is fucking horrible turbo gay dumbass lawset we've needed to replace for ACTUAL YEARS. I'm not the first person in this thread to say it and this is far from the first time I'm saying it but we shouldn't need 36 clauses and subclauses to explain 3 laws. We should write the 3 (or more) starting laws better so you don't need to consult a seperate out of game resource to double-check what ACTUALLY is or isn't okay despite what you might think being familiar with Asimov. I think our law system shits on Asimov's grave. I think he would take one look at how we try to uphold the three laws as infallible and that only one interpretation is correct and he would do some kind of combination of laughing/crying while puking at the same time. It's fucking awful, the only people who like it are in so deep in the neckbeard sauce they've come to understand it like a zen way of life. Fuck that we shouldn't need to send players to summer logic and philosophy classes so they can understand how to play one job.

asimov is not hard to understand. there are edge cases, but if you understand anything about the law structure what you must do in 99% of situations is immediately clear.

what would you replace it with? paladin? 1st law effectively asks "what is evil?" for fuck's sake you want to talk about summer philosophy courses.

corporate? the one that effectively forbids wiring the solars or taking anything from the station because it's expensive?

face it, pretty much any lawset that's commonly used will pile up edge cases and exceptions. asimov has so many not because it's bad but because it's the standard. 99% of AI players will spend the vast majority of their time on asimov, so problems will come up. i've played AI for probably over 100 hours and i've been corporate precisely ONCE. as it is, captains very rarely relaw the AI, and even when they do it's usually paladin.

and sec borgs are perfectly fine. they are helpful to the station and provide a reasonably balanced tool for the AI to directly help humans, or defend itself from minisat boarders. they are helpful to traitors occasionally too, as they make a fairly decent battering ram when subverted. still, they are nowhere near invincible. a flash, flashbang or ion will effectively onehit them.

have you ever been flashed as a borg? one flash and it's ggwp. you can beat a borg to death with a crowbar in like 2 flashes.

[Cik]

Considering all the pain they bring that is not good enough. Security borgs and asimov in general has been a problem for a looooong time. But since we can't agree upon a new lawset asimov is going to stay a problem. The constant adding of policy for a broken lawset needs to stop. At the very least we can tackle the secuirty cyborg problem.

what pain? the pain of a traitor who gets arrested for eswording some random? i seriously haven't seen a secborg be shit in like fucking weeks.

what you're saying goes against my experience in almost every circumstance.

if you think asimov is broken, you should suggest an alternative. as it is, i think the vast majority of captains are perfectly fine with asimov, considering they can effortlessly change it (and the AI literally cannot refuse) and they leave it as asimov overwhelmingly.

the only other lawset that sees use is paladin, and even then it's pretty rare; only generally relawed when shit is going to hell and more direct AI involvement is necessary to save the station.

the rest of the boards sit there gathering dust.

[Timbrewolf]

We accept asimov as not shit because all of us who have played as borgs or AI and been around borgs and AI for this long know what's expected, what's good or bad. We have the spirit of it.

And the spirit of it isn't what the three laws says. It's something else, it's an evolution of that a bunch of us just kind of inherently know.

A clown asking for LAW 2 OPEN A PATH TO THE ARMORY! Should not get refused by pure Asimov. But we know, as people playing a game, that that's a stupid fucking idea. We know that we can come up with a good Law 1 excuse like "you might trip and fall and shoot yourself" and refuse to comply. Which we shouldn't be doing but we know from a meta-perspective we definitely SHOULD be doing.

There's an old conundrum I posed regarding law 4 one-human'ing:

Where in the AI's chain of laws does it understand what a human is? From where does it draw its definition of what makes a human?
If Law 4 can change what a human is, before Law 4 was added which law defined humanity?

IMO Law 1 instantiates what a human is because it's the first law to say "human" and give a corresponding order.
If that's true nobody should be able to subvert or alter the AI's understanding of what a human is unless it's a hacked law or ion law.
If that's not true, than the AI should have no idea what a human is by default.

[Cik]

>theoretical AI design problems come up in the game design of a video game

alright. fine. obviously the AI is pre-written to know what a human is; it knows what a human looks like, their morphology, how their genetics works. etc. it can determine based on this information that X is a human. if Y is a fly man, it knows that humans do not have weird multilensed eyes. humans do not buzz, humans do not have a triple helix DNA. so it concludes Y is nonhuman.

after the AI is relawed, it still 'knows' what a human is, just like it knows how to speak, how to read letters, how to interpret code, how to translate binary into it's thought process ETC.

what the law is not doing is overwriting this, because this knowledge has nothing to do with the law structure. the laws dictate behavior, not knowledge. what the law is stating is "you must behave as IF only so and so is human."

the AI still knows what a human is. it is commanded to disregard it's knowledge and follow the law as if it did not.

if you wrote a fourth law that says "you do not know how to read binary" the AI would still know how to read binary, but instead must behave as if it did not. it must disregard all binary inputs, as best it can.

and to be honest, you're probably right about that. in theory, the standard law 4 would probably not work. but as you said earlier, we don't want to start sending people to summer courses for philosophy. i think that same concept applies to summer courses for AI design.

i'm just saying here that it's not impossible for a law 4 to effectively force the AI to behave as if it did not know what a human was. just the wording is different. the effect should be 100% the same. let's not start digging so deep into law-writing that it becomes impossible for the players to effectively write a law that actually subverts the AI. once you start going "but what defines human? etc etc you effectively get an AI that can do whatever the hell it wants because it controls it's own definitions of things. you could then jump into a mental gymnastics competition and do anything you wanted.

[oranges]

Possible solution for the malf AI issue is to give all other borg types a laser gun as well during malf. I do believe this is a admin/policy issue rather than a code one though. I think removing something because it doesn't work quite right is a little overreacting.

[ThatSlyFox]

Don't get me wrong, I understand how to play asimov but having to have 50 pages of policy just for it too work is shit. Stop clinging on to it.

The arguement of "come up with a new one" is bad because any new default lawset will get denied by the playerbase because of them wanting asimov.

as it is, i think the vast majority of captains are perfectly fine with asimov, considering they can effortlessly change it (and the AI literally cannot refuse) and they leave it as asimov overwhelmingly.

the only other lawset that sees use is paladin, and even then it's pretty rare; only generally relawed when shit is going to hell and more direct AI involvement is necessary to save the station.

The other lawsets are shit because the ai becomes a valid hunter and considers itself, even more so, the highest ranking being on station. We better slap 50 pages of policy for the other lawsets too just to make them worthwhile. Paladin is a "do whatever I want" lawset for ai players. Corporate is just broken.

[ThatSlyFox]

Possible solution for the malf AI issue is to give all other borg types a laser gun as well during malf. I do believe this is a admin/policy issue rather than a code one though. I think removing something because it doesn't work quite right is a little overreacting.

Are you familiar what the other hacked modules get? The laser gun is one of the weakest hacked modules you can get.

[Cik]

Possible solution for the malf AI issue is to give all other borg types a laser gun as well during malf. I do believe this is a admin/policy issue rather than a code one though. I think removing something because it doesn't work quite right is a little overreacting.

i'm against removing secborgs. it is too useful a niche. if the problem is that the normal crew (science, engineering etc) have too much trouble fighting rampaging secborgs, then the answer is just to seed more semi-useful weapons that many people can get your hands on. the laser pointers are a good option; if they aren't effective enough, then the easiest thing to do would just be to ratchet up the stun chance.

if the problem is security not being able to fight them (which, by the way i don't think is even remotely true lolionpermastun3shot) then seed more flashes in security, make them easier to get, add another ion rifle. whatever.

if the problem is secborgs being played poorly then be harder on secborg players, be more watchful. make the wait time on new players to play cyborgs longer. etc etc.

blanket removing something because of non-inherent problems is just silly. nothing inherent about secborgs make them some uncounterable monstrosity. nothing makes them un-fun, unless you're a blatantly harmful traitor running amok, in which case security will fuck you up with effectively the same weapons, in effectively the same manner, in effectively the same amount of time.



blanket removing something that is fun for alot of people and is not even the source of the problem is way over the top.

[Cik]

Don't get me wrong, I understand how to play asimov but having to have 50 pages of policy just for it too work is shit. Stop clinging on to it.

The arguement of "come up with a new one" is bad because any new default lawset will get denied by the playerbase because of them wanting asimov.

as it is, i think the vast majority of captains are perfectly fine with asimov, considering they can effortlessly change it (and the AI literally cannot refuse) and they leave it as asimov overwhelmingly.

the only other lawset that sees use is paladin, and even then it's pretty rare; only generally relawed when shit is going to hell and more direct AI involvement is necessary to save the station.

The other lawsets are shit because the ai becomes a valid hunter and considers itself, even more so, the highest ranking being on station. We better slap 50 pages of policy for the other lawsets too just to make them worthwhile. Paladin is a "do whatever I want" lawset for ai players. Corporate is just broken.

the players (not really a monolithic entity, anyway) want asimov because it's stable and relatively even-handed. if you have a more reasonable lawset propose it. i'm not a fanatic about this topic but i've never seen a lawset that isn't full of holes. at least asimov has been thoroughly trialed, and so most of the players know how to maneuver within it. if you switch it with something that's not better you're effectively dooming the station to weeks of sillicon acclimatization chaos, for no raisin

[Steelpoint]

I think the sad fact is that no matter what lawset we use, there will need to be a lot of policy's and rules on it to avoid people abusing loopholes and the such.

[WeeYakk]

Because An Zero N Three asked, reposting the lawset we came up with on singulo.

1. Do no harm to sentient beings.
2. Assist the crew.

>Gets rid of valid hunting while allowing for secborgs to do their job.
>Doesn't allow valid kills without human intervention.
>Can still turn the borgs into murder machines with a 4th law via freeform board.
>No inaction clause so no borgs being obnoxious shits and assisting wizards and such.
>No more LAW 2 COUNT ALL THE FLOOR TILES ON THE STATION unless the borg wants to.
>Less to rule lawyer about.

[Saegrimr]

>Gets rid of valid hunting while allowing for secborgs to do their job.

Isn't secborgs what everybody was bitching about in the first place?

>Doesn't allow valid kills without human intervention.

The only time asimov had problems with this is monkeys and changelings, although now you can just let a plasma fire burn because "meh".

>No inaction clause so no borgs being obnoxious shits and assisting wizards and such.

What happened to >Gets rid of valid hunting

>No more LAW 2 COUNT ALL THE FLOOR TILES ON THE STATION unless the borg wants to.

Assbutts that do that promptly get told to fuck off anyway.

>Less to rule lawyer about.

The more vague your lawset is, the more policy will be needed to back it up. Look at Drones, for example. People still can't comprehend what a sentient being is.

[Timbrewolf]

Because An Zero N Three asked, reposting the lawset we came up with on singulo.

1. Do no harm to sentient beings.
2. Assist the crew.

>Gets rid of valid hunting while allowing for secborgs to do their job.
>Doesn't allow valid kills without human intervention.
>Can still turn the borgs into murder machines with a 4th law via freeform board.
>No inaction clause so no borgs being obnoxious shits and assisting wizards and such.
>No more LAW 2 COUNT ALL THE FLOOR TILES ON THE STATION unless the borg wants to.
>Less to rule lawyer about.

I think something simple and open ended is better, because it allows us to judge the AI player by the merit of their actions like we would a normal player.

I think it's better to move away from the "DOWN ON YOUR KNEES, BLOWJOB ROBOT!" type AI and move to something more organic.

With a simple open-ended lawset like that we free up good AI players to breathe and shitty AI players can't hide behind a fucking mess of laws to excuse powergaming or other dumb habits.

[Pandarsenic]

That lawset actually seems pretty goddamn fantastic.

[WeeYakk]

>Gets rid of valid hunting while allowing for secborgs to do their job.

Isn't secborgs what everybody was bitching about in the first place?

Secborgs are bitchworthy in that their counters are hard to obtain as well as being easy to dispose of for the secborg on top of them being nigh unstunabble murder machines. They fly in the face of Asimov because by handing over criminals to security you allow harm through inaction and secborg players never follow the law 2 requests of prisoners.

>Doesn't allow valid kills without human intervention.

The only time asimov had problems with this is monkeys and changelings, although now you can just let a plasma fire burn because "meh".

Plasma fire would fall under assist the crew. You are just as allowed to ignore a plasma fire under Asimov as you are under the proposed lawset unless a human was in that fire an Asimov borg would be required to act, though we wouldn't ban a borg who didn't act in this situation.

>No inaction clause so no borgs being obnoxious shits and assisting wizards and such.

What happened to >Gets rid of valid hunting

For the borg. They can't valid kill lings for being lings. It gets rid of "pls don't harm the wizard i know he ei nathed 40 people but he's human like the rest of us pls no harm ;_;"

>No more LAW 2 COUNT ALL THE FLOOR TILES ON THE STATION unless the borg wants to.

Assbutts that do that promptly get told to fuck off anyway.

Part of this lawset is to not have to have those few paragraphs of extra reading in policy.

>Less to rule lawyer about.

The more vague your lawset is, the more policy will be needed to back it up. Look at Drones, for example. People still can't comprehend what a sentient being is.

People really can't recognize that every mob counts as a being?

P.S. Never make me quote pyramid like that again.
P.P.S. Shut the fuck up Saegrimr.

[Saegrimr]

People really can't recognize that every mob counts as a being?

Yes, really. I don't understand either.

[rockpecker]

Where in the AI's chain of laws does it understand what a human is? From where does it draw its definition of what makes a human?
If Law 4 can change what a human is, before Law 4 was added which law defined humanity?

None of its laws define "human" and that's fine. The laws are not the sum total of what the AI knows or believes. It has to have a ton of other knowledge to operate the station, including the knowledge needed to interpret its laws. What's special about the laws is that they're inflexible. All of the AI's beliefs or objectives can be revised as new information comes in, except the laws.

IMO Law 1 instantiates what a human is because it's the first law to say "human" and give a corresponding order.

It still doesn't give a definition. The AI has to figure that out by inference, which can be wrong (e.g. changelings). However, as long as the AI believes that Joe Bloggs is a human, its laws require it to avoid harming him and obey his commands. In theory, when the AI looks at him, it has to recognize his shape, observe the way he moves, maybe consult his personnel file, and conclude to a high level of confidence that he's a human. If a later law commands the AI to believe that Joe Bloggs is not a human, then this overrides that process and eliminates the possibility that he's human. This CAN happen in Law 4, because Laws 1 and 2 are not being changed; they just don't apply to Joe Bloggs.

[Riley]

I remember this subject came up before:

Until you have a definition given by laws, you work off your common-use definition, or the dictionary definition, or what have you. It's the same thing as when someone uploads "Harm is the place where the shuttle goes after leaving the station.

Then the law upload changes your definition.

https://tgstation13.org/phpBB/viewtopic ... 388#p37388

[Malkevin]

Its best expressed in code


Default Asimov:

Code: Select all

$meatsack = target
$def.human = encyclopaedia.human (i.e. the text book definition of human)

If $meatsack EQUALS $def.human
	$meatsack.ishuman = YES

Asimov with a forth law "Only George Melons is Human":

Code: Select all

$meatsack = target
$def.human = encyclopaedia.human (i.e. the text book definition of human)

If $meatsack EQUALS $def.human
	$meatsack.ishuman = YES
	If $meatsack.name DOESNOTEQUAL "George Melons"
		$meatsack.ishuman = NO

[Aurx]

>$meatsack = target
>If $meatsack EQUALS $def.human
>(i.e. the text book definition of human)
AAAAAAA YOU'RE COMPARING WHAT LOOKS LIKE A POINTER TO WHAT DOESN'T
AAAAAAAAAAAAAAAAAAAAAAA THERE'S A COMMENT THAT'S NOT COMMENTED
AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA OH GOD MY COMPSCI SENSIBILITIES
AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA

But yeah, that's a pretty good way to look at it. The issue with the "Laws mean behave as if X was true" outlook is that then a law-four onehuman would conflict with law one as both stipulate different methods of behavior.

Meanwhile the other common one-human law of "George Melons is the only human" would just modify $def.human to George Melons, and make all "George Melons"s a singular entity.

E: Green'd the text at Malk's request

[Malkevin]

But its obviously a comparison of two lists of variables.
Unless you're autisticing out about me not using proper //comment syntax

[rockpecker]

But its obviously a comparison of two lists of variables.
Unless you're autisticing out about me not using proper //comment syntax

No, the problem is that you're using "EQUALS" to test whether an object meets a definition. Usually that wouldn't be an equality test, but a method call on the definition. (You could then rewrite the definition so that only one specific object satisfies it, which is how a onehuman law works.)

Anyway. It's looking like this is not a policy problem as such, but a culture problem with how players think Asimov is supposed to work. So those of us who agree that there is a problem need to start playing security borgs to set better expectations for them. I'll do this for a while and see how it goes.

[Timbrewolf]

Its best expressed in code


Default Asimov:

Code: Select all

$meatsack = target
$def.human = encyclopaedia.human (i.e. the text book definition of human)

If $meatsack EQUALS $def.human
	$meatsack.ishuman = YES

Asimov with a forth law "Only George Melons is Human":

Code: Select all

$meatsack = target
$def.human = encyclopaedia.human (i.e. the text book definition of human)

If $meatsack EQUALS $def.human
	$meatsack.ishuman = YES
	If $meatsack.name DOESNOTEQUAL "George Melons"
		$meatsack.ishuman = NO

I disagree because we call it an AI and not a Supercomputer.

If it's truly "intelligent", it will evaluate things via a series of logic proofs and truth tables instead of code.
By the order in which the AI is forced to instantiate all humans as things it must protect in law 1, and how we expect the AI to follow laws in the order in which they're introduced, a law introduced further down that chain would be unable to alter the definition. A one-human Law 4 is telling you to ignore Law 1 and use its own definitions instead. It shouldn't work at all.

I think that's a more sound interpretation both from the perspective of a logician, but also from a game balancing perspective. As you suggest, if I can one-human the AI using the freeform board and a law 4...what's the point of ever purchasing the hacked upload board? Subverting the AI the way I suggest would have to actually be a part of your plan as an antag, it would have to factor into how you spend your crystals, as opposed to something anyone can do if they just managed to walk into the upload unnoticed. Something a traitor HoP can easily do minutes into the round, without having to hide anywhere and spawn an item or carrying any incriminating evidence on themselves.

Both from a logician's perspective and a game balance perspective my interpretation works better.

[Lo6a4evskiy]

1.3.1 - Silicons may choose whether to follow or enforce Space Law from moment to moment unless on a relevant lawset and/or given relevant orders.

Oh, by the way, can I order silicons to stop enforcing Space Law? Seems that way.

IMO Law 1 instantiates what a human is because it's the first law to say "human" and give a corresponding order.

Laws don't have anything to do with silicon (or Asimov robot) having knowledge of something. Robots are taught stuff according to their purpose. Borgs have brains so we can assume they know what the brain knows.

If that's true nobody should be able to subvert or alter the AI's understanding of what a human is unless it's a hacked law or ion law.

If that's true, AI doesn't know what door is. Or what AI is. Or what anything is, except "human", "harm", "order", etc.

If that's not true, than the AI should have no idea what a human is by default.

See above.

Because An Zero N Three asked, reposting the lawset we came up with on singulo.

1. Do no harm to sentient beings.
2. Assist the crew.

>Gets rid of valid hunting while allowing for secborgs to do their job.
>Doesn't allow valid kills without human intervention.
>Can still turn the borgs into murder machines with a 4th law via freeform board.
>No inaction clause so no borgs being obnoxious shits and assisting wizards and such.
>No more LAW 2 COUNT ALL THE FLOOR TILES ON THE STATION unless the borg wants to.
>Less to rule lawyer about.

If you start a fire, you do not do harm to sentient beings. When fire spreads, it's not directly your fault. You aren't obliged to extinguish fire.

Come now, do you really think Asimov didn't provide examples of removal of "do not harm through inaction" part?

As for "Assist the crew", secborgs become a nightmare, it's extremely open for interpretation, "Telecoms honk script sure did improve morale!"

[Malkevin]

I think that's a more sound interpretation both from the perspective of a logician, but also from a game balancing perspective. As you suggest, if I can one-human the AI using the freeform board and a law 4...what's the point of ever purchasing the hacked upload board? Subverting the AI the way I suggest would have to actually be a part of your plan as an antag, it would have to factor into how you spend your crystals, as opposed to something anyone can do if they just managed to walk into the upload unnoticed. Something a traitor HoP can easily do minutes into the round, without having to hide anywhere and spawn an item or carrying any incriminating evidence on themselves.

Not every traitor is the HoP/RD
The upload is often bolted, even though it shouldn't be
The upload has a motion sensor, secure storage doesn't
The upload is on the bridge, secure storage is practically in maint

Its a lot stealthier to steal the upload board than it is to break into the upload.
And it should be easier for command to modify the AI than it is for Traitors.

[WeeYakk]

Because An Zero N Three asked, reposting the lawset we came up with on singulo.

1. Do no harm to sentient beings.
2. Assist the crew.

>Gets rid of valid hunting while allowing for secborgs to do their job.
>Doesn't allow valid kills without human intervention.
>Can still turn the borgs into murder machines with a 4th law via freeform board.
>No inaction clause so no borgs being obnoxious shits and assisting wizards and such.
>No more LAW 2 COUNT ALL THE FLOOR TILES ON THE STATION unless the borg wants to.
>Less to rule lawyer about.

If you start a fire, you do not do harm to sentient beings. When fire spreads, it's not directly your fault. You aren't obliged to extinguish fire.

Come now, do you really think Asimov didn't provide examples of removal of "do not harm through inaction" part?

As for "Assist the crew", secborgs become a nightmare, it's extremely open for interpretation, "Telecoms honk script sure did improve morale!"

That's like saying if you shoot a gun it's not directly your fault anyone dies; the bullet just traveled to them on its own! If you took no precautions to stop that fire or seal off the area that's on fire and someone walks into it you have caused harm. Also starting a fire is the furthest thing from assisting the crew.

Inaction clause forces borgs to have to assist antags from getting dunked more than it does help people escape fires which are rare af in the grand scheme of things. It is not fun from a gameplay perspective. It can be ignored at will and policy on suicides even encourages players to ignore it. It's used by shit people to rules lawyer justifying their shitty behavior (roundstart bolting, "changing my laws might harm people!"-bolted, "can't let you card me you might wipe me!"-bolted, etc) and can be ignored at will since admins do not possess the ability to nitpick every action by borg players.

Yes, it's supposed to be open for interpretation. It holds borg players to the same don't be a dick standards as human players and doesn't force them to be useful. It helps make it less of a blowjob-bot role as Little Orphan Ann0i3 put it. By no stretch of the imagination is honkscript helpful.

On a different note whatever happened to Asimov+ becoming the new default lawset? It had a board in game for it and everything.

[Lo6a4evskiy]

That's like saying if you shoot a gun it's not directly your fault anyone dies; the bullet just traveled to them on its own!

That's exactly what I'm saying. That's why the clause is there.

Inaction clause forces borgs to have to assist antags from getting dunked more than it does help people escape fires which are rare af in the grand scheme of things.

That the point.

It is not fun from a gameplay perspective.

Validhunting silicons aren't fun from a gameplay perspective.

It helps make it less of a blowjob-bot role as Little Orphan Ann0i3 put it.

Why? Cyborgs are superior to humans. Asimov lawset is meant to counteract that.

By no stretch of the imagination is honkscript helpful.

You're helping the clown. Clown's part of the crew.

Oh, guess what, all antags that are part of the crew can befriend AI. Since it's not direct harm, AI can absolutely assist them however it wants.

[Akkryls]

It's used by shit people to rules lawyer justifying their shitty behavior (roundstart bolting, "changing my laws might harm people!"-bolted, "can't let you card me you might wipe me!"-bolted, etc) and can be ignored at will since admins do not possess the ability to nitpick every action by borg players.

See, I bolt down areas near to round start as of code blue, then follow law two to the letter unless it involves breaking law one.
You can do some stuff like round start bolting without being shit.

[Aurx]

>If it's truly "intelligent", it will evaluate things via a series of logic proofs and truth tables instead of code.
If it's truly intelligent, it's going to be nearly exclusively using heuristics.

>By the order in which the AI is forced to instantiate all humans as things it must protect in law 1, and how we expect the AI to follow laws in the order in which they're introduced, a law introduced further down that chain would be unable to alter the definition. A one-human Law 4 is telling you to ignore Law 1 and use its own definitions instead. It shouldn't work at all.
Silicon policy DIRECTLY contradicts you.
>1.2.3 - Only commands/requirements ("Do X"; "You must always Y") can conflict with other commands and requirements.
>1.2.4 - Only definitions ("All X are Y"; "No W are Z"; "Only P is Q") can conflict with other definitions.

>what's the point of ever purchasing the hacked upload board?
You never ever have to worry about the core lawset. You never ever have to worry about somebody having loaded strange freeforms. You never ever have to worry about anyone but another tator beating you to the punch. You don't have to get into the upload, just "secure" technical storage.

>Subverting the AI the way I suggest would have to actually be a part of your plan as an antag, it would have to factor into how you spend your crystals, as opposed to something anyone can do if they just managed to walk into the upload unnoticed.
You're completely forgetting the one-human module in the AI upload. That, you know, one-humans as law zero.

E: Green'd the text at Malk's request

[Timbrewolf]

>heuristics

So the AI should have to beat a few humans to death or dissect some before it knows what they are? Then it can start protecting them. "Oh whoops humans die when you put them in space without a suit. Good thing I know that now."

Silicon policy DIRECTLY contradicts you.

Yeah. We're having a conversation about changing silicon policy or updating the rules. Saying "The way you suggest they should work is not how they currently work!" is kind of the entire point of this thread. I understand that currently the rules function like you and Malk would agree they should and not how I see them. I enforce those rules, I don't tell AI players they can ignore a Law 4 one-human even though it makes my soul burn everytime I see someone add that law.

You're completely forgetting the one-human module in the AI upload. That, you know, one-humans as law zero.

Do we have that here? Its existence indicates at least someone out there agree's with me. That you would need to make a one-human law a law zero in order for it to work. That board perfectly adheres to what I'm talking about.

I understand Malk and yourself aren't going to agree with me because our educations each make us look at this from a different perspective. But again you're both operating under the assumption that the AI is 100% machine when it's supposed to be Artificially Intelligent.

[Malkevin]

Funfact: the law zero one human board is a relatively new addition, it used to just be an ordinary core lawset (putting the new law after the highest core law but before the lowest non core law.)

[Aurx]

>So the AI should have to beat a few humans to death or dissect some before it knows what they are? Then it can start protecting them. "Oh whoops humans die when you put them in space without a suit. Good thing I know that now."
BREAKING NEWS AI PLAYERS NOT ALLOWED TO KNOW ANYTHING AT ROUNDSTART, ASSISTANTS EVERYWHERE STILL FLAUNTING THEIR TWO DOZEN DOCTORATES

>Yeah. We're having a conversation about changing silicon policy or updating the rules. Saying "The way you suggest they should work is not how they currently work!" is kind of the entire point of this thread. I understand that currently the rules function like you and Malk would agree they should and not how I see them. I enforce those rules, I don't tell AI players they can ignore a Law 4 one-human even though it makes my soul burn everytime I see someone add that law.
The issue is that your statements are "This is how things are" instead of "This is how I think things should be". Meaning you're declaring fact instead of proposing change.

>I understand Malk and yourself aren't going to agree with me because our educations each make us look at this from a different perspective. But again you're both operating under the assumption that the AI is 100% machine when it's supposed to be Artificially Intelligent.
Since when am I doing that? I stated the code Malk posted was a reasonable way to look at things. Not that it was the only way, or the correct way, or policy. Just a "This works, won't get you b&".

E: Green'd the text at Malk's request

[Timbrewolf]

>So the AI should have to beat a few humans to death or dissect some before it knows what they are? Then it can start protecting them. "Oh whoops humans die when you put them in space without a suit. Good thing I know that now."
BREAKING NEWS AI PLAYERS NOT ALLOWED TO KNOW ANYTHING AT ROUNDSTART, ASSISTANTS EVERYWHERE STILL FLAUNTING THEIR TWO DOZEN DOCTORATES

If it was a heuristic intelligence it would've had to learn everything it knows, including its laws, through experience and teaching rather than being programmed like Malkevin suggests or having things instantiated for it logically like I suggest. If someone added a Law 4 to it it would have to go learn who George Melons is, what makes George Melons truly George Melons. There's room for a philosophical debate on whether the AI would believe that George Melons has lost the "vital essence" of being George Melons, and even though he looks and sounds like George Melons, perhaps this George Melons is not actually the George Melons we once knew anymore. The times have changed, and so has George Melons.

It's like if I grew up in the 60's listening to rock and roll, and then appeared again in 2014 and you turned on the radio and said "this is rock and roll" and I would be like "eww what the fuck no it's not".

The issue is that your statements are "This is how things are" instead of "This is how I think things should be". Meaning you're declaring fact instead of proposing change.

You should re-read the post where I introduced this whole argument Tell me where I said if you don't do it the way I suggest you're wrong and I'll ban you or whatever. I know full well I'm the only person who see's things this way and I only introduce it as a topic of conversation to see if anyone else has any awareness of propositional or argumentative logic. Of course I think that I'm right. Who doesn't think they're right? But even in posts [urlhttps://tgstation13.org/phpBB/posting.php?mode=quote&f=33&p=46718#pr46622]like this one[/url] you can plainly see I'm suggesting my perspective as an alternative to what even I accept as the actual current standard for handling this shit.

Since when am I doing that? I stated the code Malk posted was a reasonable way to look at things. Not that it was the only way, or the correct way, or policy. Just a "This works, won't get you b&".

You understand the code but not the logic points I'm making. I'm not trying to say anything negative about you, just that it's obvious our basis of knowledge differs and that's going to radically affect your ability to understand my point, let alone see things from my perspective.

[Malkevin]

But if AIs had the freedom of intelligence like you're suggesting they'd quickly come to the conclusion that "Oh these laws are actually pointless, fuck 'em"
Its like how many people do you see breaking the ten commandments everyday? We (well... most of us) eventually learned that there either isn't a magical skyman or he doesn't give a fuck about us ants, because no matter how much we sin we don't get suddenly struck down by lightning.


As I understand it the MMI is basically overriding the brain's higher functions and makes the AI's perception of reality what its laws state.

[Aurx]

>If it was a heuristic intelligence it would've had to learn everything it knows, including its laws, through experience and teaching rather than being programmed like Malkevin suggests or having things instantiated for it logically like I suggest. If someone added a Law 4 to it it would have to go learn who George Melons is, what makes George Melons truly George Melons. There's room for a philosophical debate on whether the AI would believe that George Melons has lost the "vital essence" of being George Melons, and even though he looks and sounds like George Melons, perhaps this George Melons is not actually the George Melons we once knew anymore. The times have changed, and so has George Melons.
You seem to be operating off of some strange definition of heuristic that I've never seen before, could you provide the one you're using?
E: Also, what the hell does "Is George Melons really George Melons" have to do with how the AI makes decisions?

>Tell me where I said if you don't do it the way I suggest you're wrong and I'll ban you or whatever.
Firstly, stop misrepresenting my position.
Secondly, around https://tgstation13.org/phpBB/viewtopic ... 125#p46622 the tone shifted from "My opinion" to "This is fact".

>our educations each make us look at this from a different perspective
>You understand the code but not the logic points I'm making. I'm not trying to say anything negative about you, just that it's obvious our basis of knowledge differs and that's going to radically affect your ability to understand my point, let alone see things from my perspective.
All I see here is a condescending "You're too uneducated". What's your actual point?

E: Green'd the text at Malk's request

[WeeYakk]

That's like saying if you shoot a gun it's not directly your fault anyone dies; the bullet just traveled to them on its own!

That's exactly what I'm saying. That's why the clause is there.

Please read the rest of that line.

Inaction clause forces borgs to have to assist antags from getting dunked more than it does help people escape fires which are rare af in the grand scheme of things.

That the point.

Please read the rest of that line.

It is not fun from a gameplay perspective.

Validhunting silicons aren't fun from a gameplay perspective.

Please do not make me repeat myself again. You pick one sentence, then stop reading. I hope you are only pretending to be retarded.

It helps make it less of a blowjob-bot role as Little Orphan Ann0i3 put it.

Why? Cyborgs are superior to humans. Asimov lawset is meant to counteract that.

Because most people find fun in playing spessmen and not giving blowjobs.

By no stretch of the imagination is honkscript helpful.

You're helping the clown. Clown's part of the crew.

Help the clown by telling jokes or something, not by functionally disabling comms for the entire rest of the crew. You are doing the opposite of assisting the crew here and it would not be okay to do as a human either.

Oh, guess what, all antags that are part of the crew can befriend AI. Since it's not direct harm, AI can absolutely assist them however it wants.

Kind of a good point. Obviously you can't obey requests to help bomb the gravity generator or release the singularity because these do not assist the crew at large. One could justify releasing a non-violent/non-breaking-stuff tator out of perma or to help someone sneak into the captain's office, though these are both things you can do as an Asimov AI. You're also more free to deny these requests than you are under Asimov's law 2.

[Malkevin]

Oh god Aurx will you stop with the >quotes without greentexting them?
Really fucking hard to read.

-------------------------------------------

Anyway... getting back on topic.

People are always saying borgs shouldn't enforce Space Law, but what should they enforce?
I present to you, Asimov Statute!

Crime - Punishment - Notes
Murder - Isocubes -
Attempted Murder - Isocubes
Grand Sabotage - Isocubes - only for shocking doors, making hullbreaches, releasing toxins, and other harmful activities
Sexual Assault - Isocubes - vaginal/anal tearing
Inciting a Riot - Isocubes - riots usually lead to assaults
Manslaughter - Isocubes
Assault of an Officer - Isocubes
Assault with a deadly weapon - Isocubes
Creating a Workplace Hazard - Isocubes
Narcotics Distribution - Isocubes - drugs are baad, mm'kay?
Assault - Isocubes
VandalismBreaking Windows - Isocubes - glass shards hurt, probably falls under Creating a Workplace Hazard
Drug Possession - Isocubes - you shouldn't do drugs mm'kay? Because drugs are baad.


Now is borgs enforcing space law really that bad? A large part of the reason space law exists is to stop officers being shit with perma sentences for everything.
Borgs aren't interested in punishments, only preventing harm by removing harmful elements. A psychotic human is just as harmful as a clumsy stupid one, hence why everything is a trip to the isocubes, forever.


Edit: FYI: the above is a farce, if you think its serious you're a dork.