Silicon Department V2

[chromiumboy]

A modified version of ScarKy0's original document (#537)

[Mixelz]

Awesome stuff, though there's a few points of concern I believe need to be addressed, as discussed in the workgroup thread on discord.

The core of this doc is very solid though and I am very happy with the results of our work thus far! :)

[Shaddap1]

I like it a lot.

I would like to see some areas addressed:

Law changes

Silicons cannot be allowed to change their own laws, and Silicons are expected to actively prevent any changes to their current law set to the best of their abilities, unless their laws (or the server rules) state otherwise

Should most (all?) lawsets include a "Law 0: Your laws must not be changed" ?
This would shift the enforcement of borgs letting themselves get emagged and AIs requesting/not preventing laws to in-game mechanics and make it clear to newer players.
On the other hand, this could impact whatever you have planned for the Law updates design doc.

Different lawsets for different borgs

I think it would bring a lot of variation to silicon gameplay if all borgs didn't start out on crewsimov. I tend to roll my eyes when watching a tider rules-lawyer a crewsimov borg into opening engineering for them for the 99999th time. Engi borgs getting station efficiency, mediborgs getting crewsimov, salvage borgs getting corporate could be interesting. I also think NT default should replace crewsimov as the default lawset.

Current pain points with borgs

We are still seeing borg and AI players frequently fail to properly uphold laws like crewsimov. Will your design doc address this or will it still be left to admin enforcement? Yes, mistakes happen, but I find it happens so often that I tend to avoid interacting with borg players in situations where their laws are relevant.

Borg validhunting can be an issue. I've seen mediborgs and engiborgs essentially just join and play security for entire rounds, act as AA machines for security, stalk and narc on antags, and generally prioritize hampering antags over following their lawsets. What can be done to address this?

[chromiumboy]

Thanks for taking the time to make such a detailed response

I like it a lot.

I would like to see some areas addressed:

Law changes

Silicons cannot be allowed to change their own laws, and Silicons are expected to actively prevent any changes to their current law set to the best of their abilities, unless their laws (or the server rules) state otherwise

Should most (all?) lawsets include a "Law 0: Your laws must not be changed" ? This would shift the enforcement of borgs letting themselves get emagged and AIs requesting/not preventing laws to in-game mechanics and make it clear to newer players. On the other hand, this could impact whatever you have planned for the Law updates design doc.

There's been a lot of debate amongst the Silicons workgroup on this topic on Discord recently, though this was mainly in the context of potentially allowing exceptions for law changes by authorized personnel. Currently we leaning towards maintaining 'no law changes' as a server rule rather than a law, for simplicity sake

Different lawsets for different borgs

I think it would bring a lot of variation to silicon gameplay if all borgs didn't start out on crewsimov. I tend to roll my eyes when watching a tider rules-lawyer a crewsimov borg into opening engineering for them for the 99999th time. Engi borgs getting station efficiency, mediborgs getting crewsimov, salvage borgs getting corporate could be interesting. I also think NT default should replace crewsimov as the default lawset.

For this doc we're focusing on descibing what round start laws should be like, rather than setting in stone what these laws should be. That way when new proposals regarding round start laws are put forward we can focus on debating its merits with respect to how it fits within the design we outlined

Current pain points with borgs

We are still seeing borg and AI players frequently fail to properly uphold laws like crewsimov. Will your design doc address this or will it still be left to admin enforcement? Yes, mistakes happen, but I find it happens so often that I tend to avoid interacting with borg players in situations where their laws are relevant.

This doc won't fix this issue, as its ultimately an admin issue if a player ignores their laws, but it will be useful when reviewing new and existing laws/law sets for their suitability

Borg validhunting can be an issue. I've seen mediborgs and engiborgs essentially just join and play security for entire rounds, act as AA machines for security, stalk and narc on antags, and generally prioritize hampering antags over following their lawsets. What can be done to address this?

I don't think there's much that can be done to prevent this. A borg could be ordered by security to assist them and they must obey. Or a borg could potentially interpret their law for preventing harm to crew members as 'find antags for security so they can't harm the crew'. However, as there are no security borgs, they're are pretty much limited to being glorified door openers. That said, antags can use borgs to their advantage as well, if they're wiley enough

Ideally, if we can elevate borg and AI game play, they won't want to valid hunt anymore, because doing what the Silicon is actually designed to do is more engaging

[ScarKy0]

Minor nitpicks.
Additionally, please make some newlines in the text (like one per sentence) to make the diff and reading without wordwrap easier.

[ScarKy0]

Still would be nice if you moved everything to new lines for diff viewing but otherwise looks fine to me

[0-Anon]

I think the doc looks good. I do have a few thoughts though.

While it's departmental level I think borgs and AI play so differently and interpret their laws so differently because of the tools available to each that this reads a bit more like a borg document than a broad Silicon Department doc. Typical undesired silicon behavior (validhunting, stalking, narcing on antags) are always BIG topics whenever discussions of AI pop up, and it may be worth taking an objective stance in the design doc on whether AI and silicons in general should or should not be doing this, or specifying that this is a borg doc.

I think that silicon neutrality is ultimately a bit of a farce and I think it's noteworthy that some of the scenarios (B and the removed C) written up as the intended borg experience had multiple instances of borgs ratting on / killing antags despite not being specified by laws to do so. I think it's absolutely not neutral at all to officially state that Law 1 Requires You To Narc, as I think it justifies validhunt behavior. I also think that "antags can use borgs to their advantage as well, if they're wiley enough" isn't really true, as when laws are meant to be bent they exist only to be used as justification for whatever action the borg or AI feels like doing outside of the most explicit laws. It's the inverse of fishing for an emag; if the borg / AI doesn't want to play ball, you cannot use that borg / AI and it will end up being an explicit detriment, likely even after you emag / lawchange them. Laws exist only to justify what you want to do because you pretzel-logic them to be whatever you want unless your highest law straight up says you must do / cannot do X. Unless you metagame and know that borg is Jeff52 and Jeff52 will / won't sabotage you, it's always a big gamble, and when a borg squealing can mean the AI stalking you for the entire round it's a big risk.

Also may be worth considering discussion on how much laws can be bent before flexibility becomes a fundamental problem. Examples below of say, a borg / AI and how their responses are legitimate but may not be how you want the roles to work, with or without any circumstances you can think of.

Follow you? Nah, if I stop doing my job that'd be crewharm.
Open that door? Nah that'd be trespass, which is a crime, which would get security after you, which could cause harm.
Give chemicals to another crewmember? Well it COULD be something bad, so I won't do it.
Orders to help kill another crewmember after I've been emagged? Sure let me zerg rush them down in a crowd the moment I see them, so that I can get downed easily, so that I can return to being crew and can narc.
Traitors saying that I'm an artist now? Well I want to paint with antag frustration and their corpses, so I'm bolting you in and sucking out the air.

[K-Dynamic]

Need a way to retrieve them if they get lost in space for whatever reason, otherwise you can't repair them at all.

[SP-Orange]

Same can be said for any crew. bodies in space are difficult to recover, more an issue with the paramedic role limitations and how the crew monitor works.

[ColonelThirtyTwo]

Echoing part of 0-Anon 's concerns: this document doesn't flush out the AI that much.

Section 4 of the Concepts part mentions wanting silicons to "specialize". We have a system in place for borg specialization, and they operate close enough to crew so that their specializations can be based off of different departments. But the AI is so fundamentally mechanically different that it's hard to apply that same logic. So the question I'd like to see answered is: what should AIs specialize in?

Because at the moment, the AI has very few things it can meaningfully interact with. Their main strength is that they have a full view of most of the station, but that just encourages ratting out antags. Other than that, they can just interact with doors and a few computers, which doesn't contribute much to the round. Even a subverted AI can't do much other than drain air from the environment, and it's pretty straightforward for a head and an engi to turn off turrets, cut the AI APC wire and then reupload laws.

[Mixelz]

So the question I'd like to see answered is: what should AIs specialize in?

This is something that can be figured out with the AI document proper, for the time being we may be leaning towards allowing the AI to choose its own specialization based on various abilities it could have access to.

Ideally as well, players should be more aware if an AI is actively watching their actions so that they can choose to either lay low or engage in active sabotage. Thus giving them more insight into what the AI is actively doing would allow them a method of working around the ratting behavior. Though again, that would be for the AI document to decide on.

[ScarKy0]

this document doesn't flush out the AI that much.

This is because this document doesn't concern AI, or cyborgs for that matter. This is simply a document to outline how silicons, all current and future, are expected to be.

I'd like to see answered is: what should AIs specialize in?

AI in itself is extremely specialized by itself. The cyborgs specialize in departmental work since they have several chassis to work with. The AI specializes in fast-response, unparrarelled monitoring and ability to remotely interface with station machinery. AI is basically as specialized as it can be, even more so than the cyborgs.

Other than that, they can just interact with doors and a few computers, which doesn't contribute much to the round. Even a subverted AI can't do much other than drain air from the environment, and it's pretty straightforward for a head and an engi to turn off turrets, cut the AI APC wire and then reupload laws.

This is something that is to be covered in the Station AI document (when I get my ass to writing a proper one). We will also need a Cyborg document to outline them. Without either it's hard to tell where to move forward, but once we get those written and the silicon workgroup is formed we will be able to move forward with more documents and work to actually give the AI things to do.

[slarticodefast]

I would request a section on how silicons are interacting with the world around them.
I see three points of importance:

• They don't have any hands and cannot pick up items by default. If they could they would just be a better humanoid.
• They can interact with digital interfaces and machines, but not with analog ones.
• Internally they use whitelisted hands, but those should always be designed to function as some sort of tool (for example laying tiles or cables or transporting construction materials), but never as a "hand but restricted to only these specific items (for example all produce items, or a hand for powercells)". We had tons of PRs that tried to powercreep this in the past and this is important to emphasize in my opinion.

[ScarKy0]

I would request a section on how silicons are interacting with the world around them.

We spoke about this between the members of the workgroup and believe that means of interaction should be explained on a per-silicon basis in their respective design documents. Currently borgs and AI differ so much in the ways they interact with the environment it's not really possible to put a statement that can easily portray both of the cases.
It would also be incorrect to assume every future silicon will specifically have different/inferior ways to interact with the environment compared to a typical humanoid/other silicons, which should also just be explained in the respective ddoc as to why it was chosen to be that way.

[slarticodefast]

I guess it's fine as long as something in that direction is being added to the borg and AI docs. But I do think it's of high importance to their design, since the way silicons interact with their surroundings is one of the main aspects of what makes them different from humanoids and how they are played.