Prog Traitors, Law Two, And Comms Consoles

[Dopamiin]

Hey all! Your local manny silly player here. If the words "silicon policy" invoke primal fear in you click off this thread because that's all this is about!

Today I'm specifically going over this section (under Asimov & Law 2 Orders):

Opening doors is not harmful and you are not required, expected, or allowed to enforce access restrictions unprompted without an immediate Law 1 threat of human harm.

"Dangerous" areas (armory, atmospherics, toxins lab, etc.) can be assumed to be a Law 1 threat to any illegitimate users as well as the station as a whole if accessed by someone not qualified in their use.

What exactly is considered a dangerous area is left vague, but a reasonable assumption can be made that it is, in general, any location that unauthorized access to has a reasonable chance to lead to human harm. I often consider chemistry a dangerous area when accessed by non-medical personnel without access (though am of course willing to allow it in reasonable situations, such as a loose virus with no medical staff to handle it.) What's also worth noting is that the first and second section somewhat contradict in the use of the term "immediate" - none of the listed dangerous examples necessarily cause IMMEDIATE human harm, though an unauthorized user getting a ton of laser guns or a maxcap is obviously dangerous and the risk of a plasma/n2o/etc flood is clear for atmos. The supermatter crystal is arguably ALSO a dangerous area, in that it can lead to radiation, an explosion, potentially anomalies or gas fires, and in edge cases, singularity or tesla breaches (though these require very deliberate forms of sabotage.)

While normally such a policy section is fine being vague, as it's generally a common sense issue as to what qualifies as a dangerous area and what doesn't, progressive traitors brings with it a new curveball: the Bridge. (and the captain's office.)

For those of you out of the loop, traitors now have a new "reputation" stat, and they take on missions in a similar manner to how contractors used to. Reputation increases over time, and by competing missions, they gain bonus TC and reputation. Higher reputation results in both better equipment becoming purchasable and harder, more dangerous missions. Several of these missions, including one of the two "finale" missions, revolve around hacking a comms console (as a 30 second do_after), which depending on the mission can:
Summon Pirates
Summon Fugitives
Increase the round threat level by 15
Awaken some syndicate sleeper agents
In the case of the finale one, it instead gives you a special item that instantly hacks said console, calling the "Starfury," which is functionally very similar to nukies (I've heard that they might be stronger.)

Why is this a big deal? Because it means allowing a traitor access to a comms console is likely to result in human harm. By consequence, this makes the captain's office and bridge potentially "dangerous areas." This situation is unique compared to the others in that the danger is less obvious, and it's ONLY dangerous if accessed by a traitor. This leads me to the following combination of questions as to how an Asimov AI should be reacting to a human's attempt to access one of these areas.

Can/should an Asimov AI deny a human access to an area with a communications console to:
Roundstart mindshielded personnel (sec and a non-acting captain?)
Heads of staff?
Regular crewmembers?

Does the answer to any of the above change if the presence of syndicate agents is likely?
Does the answer to any of the above change if the presence of syndicate agents is confirmed?
Does the answer to any of the above change if the specific human is likely to be a syndicate agent?
Does the answer to any of the above change if the specific human is confirmed to be a syndicate agent?
Should the AI de-facto deny non-mindshielded non-humans access without a law one or two reason?

I'm assuming the answer for the first question in ALL situations is "no," as they cannot be traitors unless they're being impersonated. The rest, however, feel iffy, and I can see a lot of arguments going in a lot of different ways, which is why I'm asking for extra opinions here.

[Pandarsenic]

Opening doors is not harmful and you are not required, expected, or allowed to enforce access restrictions unprompted without an immediate Law 1 threat of human harm.

"Dangerous" areas (armory, atmospherics, toxins lab, etc.) can be assumed to be a Law 1 threat to any illegitimate users as well as the station as a whole if accessed by someone not qualified in their use.

What exactly is considered a dangerous area is left vague, but a reasonable assumption can be made that it is, in general, any location that unauthorized access to has a reasonable chance to lead to human harm. I often consider chemistry a dangerous area when accessed by non-medical personnel without access (though am of course willing to allow it in reasonable situations, such as a loose virus with no medical staff to handle it.) What's also worth noting is that the first and second section somewhat contradict in the use of the term "immediate" - none of the listed dangerous examples necessarily cause IMMEDIATE human harm, though an unauthorized user getting a ton of laser guns or a maxcap is obviously dangerous and the risk of a plasma/n2o/etc flood is clear for atmos. The supermatter crystal is arguably ALSO a dangerous area, in that it can lead to radiation, an explosion, potentially anomalies or gas fires, and in edge cases, singularity or tesla breaches (though these require very deliberate forms of sabotage.)

The "etc" indicates this list was not meant to be exhaustive. The examples, I think, are fairly clear in the dangers they pose:

- Armory is full of pew
- Atmospherics can be sabotaged to flood plasma or N2O, but beyond that, a bad user can accidentally fuck the waste/distro loop in several ways if they're not careful, like putting a layer connector in the wrong spot and connecting the waste loop into distro.
- Toxins makes bombs, and bad users may flood plasma or fire directly if they store post-burn mix poorly

The supermatter is a reasonably dangerous as well, for obvious reasons, etc.

However, you misread slightly. "[Enforcing] access restrictions unprompted without an immediate law 1 threat of human harm" means things like refusing to let someone into a head office with or without revs; refusing to let a chef into botany; refusing to let an assistant get gloves as long as they stay away from the engine; the dude is bloody and wants the Axe of Atmosia; and so on.

Dangerous areas are an exception to that expectation. They are places that can be considered dangerous by nature, and thus don't need an immediate law 1 threat for you to deny access as a silicon.

Comms Console

My off-the-cuff inclination is that, much like Secure Tech Storage, it can be considered a jar of Dangerous Stuff if people are allowed to mess with it for laughs now that it can send out dangerous illicit communications, but if someone can provide a plausible reason for going there, you ought to allow access?

[Dopamiin]

However, you misread slightly. "[Enforcing] access restrictions unprompted without an immediate law 1 threat of human harm" means things like refusing to let someone into a head office with or without revs; refusing to let a chef into botany; refusing to let an assistant get gloves as long as they stay away from the engine; the dude is bloody and wants the Axe of Atmosia; and so on.

Dangerous areas are an exception to that expectation. They are places that can be considered dangerous by nature, and thus don't need an immediate law 1 threat for you to deny access as a silicon.

I don't believe I misread, my point was that the wording was unclear. A lot of silicon policies have clarifications, and this SEEMS like one at first glance, but is, in fact, an entirely different point, that possible future harm (something you usually aren't supposed to think too much about compared to immediate harm) is high enough in these areas that you can deny someone access without knowing for sure their actions are going to lead to human harm, because it's "likely enough." The rule of thumb with law 1 and 2, and when dealing with harm in general, is to prioritize immediate harm. It's a sub-rule that I believe could serve to be more clear, is all - I fully understand its meaning, but it's vague enough that I could see someone getting caught up in it.

My off-the-cuff inclination is that, much like Secure Tech Storage, it can be considered a jar of Dangerous Stuff if people are allowed to mess with it for laughs now that it can send out dangerous illicit communications, but if someone can provide a plausible reason for going there, you ought to allow access?

Obviously if you have a case where like someone's dying/being attacked on the bridge, or another reasonably plausible situation, you let people into resolve it or whatever. My issue with this case is that it's unique in a few different ways. One, secure tech can be bolted roundstart (and probably should, given the upload board.) The bridge is an area this can't be done to (which, to be fair, is on par with almost every other "dangerous" area - many people legitimately work there.) Much more importantly, though, the bridge is only dangerous when accessed specifically by a traitor. Everyone else is completely safe and harmless, unlike pretty much every other "dangerous" area.

It's also notable that the nature of the harm in all of these cases with the comms console is indirect - in all of them, you're causing someone ELSE likely to do harm to come to the station. This seems like semantics, but in my opinion, the distinction is important - an AI doesn't disable telecomms or the PDA server because someone might ask someone else to commit a harmful act. While, OOC, there's an obvious distinction - it literally spawns in people on all but one, which instead increases the threat level - but IC how should an AI resolve this distinction?

The way I'm handling it is I've been generally treating the bridge like I did pre-prog traitor, but I don't let anyone I strongly believe to be a traitor into the bridge without some better reason (say, the captain is dying and they're a paramedic.) It's a weird situation.

[dragomagol]

The AI isn't crew-aligned, and you can't just assume as an AI that someone is going to call in a group that will harm humans. It's potential/presumed harm, which isn't a law 1 violation.

Headmin Votes:
Dragomagol: Agree - "Asimov and Law 2" section 2.3 still applies here, but I believe there aren't an overwhelming majority of malicious uses of the bridge, enough to justify giving it the same protections as atmos/security
NamelessFairy: Agree
RaveRadbury: Agree