[Vehicles] How big should a self-destruct be?

[DataPacRat]

"The Osterhagen Key is to be used if the suffering of the human race is so great, so without hope, that this becomes the final option."

A particular space-borne cybershell has around 100 hp (and around DR 50 of armour, which won't really affect a self-destruct charge). It also has four computers, each with component armour of DR 100 (and maybe 1 hp each). If things get bad enough that the AI doesn't want any trace of itself left, it needs to be able to reduce not just the whole cybershell, but all four of those computers to -10 hp. IIRC, DR is squared against explosives, so a self-destruct charge would need to have a near-certainty of doing 10,010 points of damage.

According to Vehicles, 100 micrograms of antimatter should do around 0.00125 kilotons of boomage, or 15d*2000 damage, an average of 105,000 points. (10 micrograms would only do an average of 10,500, which is even odds of not being enough.) And at TL10, a storage bay, with -20 bonus to safety rolls, holding that much AM, plus that much AM, is 0.42 lbs, 0.0042 cf, and $5.20. (Not counting the minor detail that it's around LC0, if not LC-1.)

One concern is long-term reliability; the AI doesn't want that AM to be let loose unless it's absolutely sure it wants to self-destruct. So I'm leery of including more than one AM charge, such as one inside each computer-casing, as that multiplies the chances that any one might have something go wrong. Just how safe is that -20? Vp186 says an AM fuel bay will blow up on a 16 or less if damaged; but it doesn't say that's a skill roll, so I'm not sure if the usual "always crit on extremest roll regardless of modifiers" principle applies. If the roll is reduced to effectively "never", then I could significantly reduce the size of the charge(s), and avoid the AI having to deal with uncomfortable problem of deciding whether collateral damage nearby is worth keeping itself out of hostile hands. (Plus, it's much easier to get permission & licensing for boomage that barely escapes the cybershell's armor, compared to boomage that can level a city block or three.)


(I haven't yet decided whether or not to steal a bit of technobabble from Schlock Mercenary: "fullerened antimatter", with antiprotons (or heavier anti-atoms) stored inside carbon-buckyballs. I'm open to other suggestions for "effectively permanently stable, but explodable on demand" AM storage.)

[Rupert]

It depends on whether you're using 3e or 4e. In 4e DR is not multiplied vs explosions. Also, in 4e internal explosions count as vitals hits for triple damage.

Also, the charges should probably include ones inside the computers' armour.

[DataPacRat]

Originally Posted by Rupert

It depends on whether you're using 3e or 4e. In 4e DR is not multiplied vs explosions. Also, in 4e internal explosions count as vitals hits for triple damage.

I'm taking flufftext, building approximations with 3e's Vehicles/Robots/Mecha/etc system, and then translating any stats I need to 4e. (I tried starting with 4e Spaceships, but prefer the extra detail from 3e.)

Also, the charges should probably include ones inside the computers' armour.

Probably. I might go for chemical explosives for those, since they only need 11 or so points of damage; maybe some 4e TL10 10mm HEMP limpet-grenades would do.

[Anthony]

There isn't really a method of leaving no trace -- any explosion enough to leave no traces of the shell will be big enough to be a 'obviously something unusual was here' trace by itself.

[DataPacRat]

Originally Posted by Anthony

There isn't really a method of leaving no trace -- any explosion enough to leave no traces of the shell will be big enough to be a 'obviously something unusual was here' trace by itself.

Sorry, I phrased myself badly; I don't necessarily mean no trace of the probe ever having existed, just no trace of the AI's programming or data.

[Ulzgoroth]

Originally Posted by DataPacRat

Sorry, I phrased myself badly; I don't necessarily mean no trace of the probe ever having existed, just no trace of the AI's programming or data.

That should be achievable with very small devices inside the computer units, directly destroying the processing and storage hardware.

[Anaraxes]

Originally Posted by DataPacRat

no trace of the AI's programming or data.

Real-world tech includes some curious anti-tamper memory devices that can erase themselves if they get exposed to light (say, the case is opened). There are also encrypted memories that rely upon the processor to supply the correct key to be read, either with conventional encryption or by scrambling the address and data bus to confound snoopers. It'd be pretty easy to make the key with some types of one-time-programmable memory that you could just blow to all zeroes when you wanted to push the big red button.

Technical details aside, you could have components that destroy the data and programming without actually destroying the components. If you're willing to physically destroy stuff (or prefer to do that, so the enemy can't reverse-engineer your tech), then it wouldn't take much of an explosion to render the important ICs unusable. And if you're really paranoid, you might spend those few ms to erase the memory before blowing them into little bits, just in case of bad luck on the explosion or to prevent anyone from gleaning bits from recovered pieces.

[Anthony]

Originally Posted by DataPacRat

Sorry, I phrased myself badly; I don't necessarily mean no trace of the probe ever having existed, just no trace of the AI's programming or data.

Shorting the power system through the brain will likely do the job, but if you want something more reliable I think thermit blocks are a tradition.

Of course, if it's antimatter fueled it might blow up anyway, but that's just a function of how much fuel it carries. Antimatter isn't really that good at destroying evidence, the byproducts are mostly penetrating radiation so it spreads the energy release across millions of cubic meters. It will efficiently wipe electronics, millions of rads does that, but physical structure will remain intact unless there's a lot.

[RyanW]

Originally Posted by DataPacRat

Sorry, I phrased myself badly; I don't necessarily mean no trace of the probe ever having existed, just no trace of the AI's programming or data.

Would securely wiping the storage media, leaving an inert but intact shell, suffice?

[Plane]

Originally Posted by DataPacRat

A particular space-borne cybershell has around 100 hp
..
According to Vehicles, 100 micrograms of antimatter should do around 0.00125 kilotons of boomage, or 15d*2000 damage, an average of 105,000 points.

One idea that came to mind to model a self-destruct would be taking B136-7's Fragile: Explosive as a "Controllable Disadvantage" perk.

6DxHP/10 would be 6Dx10 for something with 100 HP though so that would fall way short of what you want.

Makes me wonder if it could be bought as a leveled perk though...

Like if you take Self Destruct [100] you get 100x the damage from your explosion?

6Dx1000 is only an average of 21,000 so I think you'd actually have to take 500 levels of it.

I wonder if a cheaper approach might be to design a one-use-ever innate attack (this is -80% or 1/5 cost depending on where you look) which is also powered by character points (may as well: you'll have no use for them!)

You could design this as an "Internal Advantage" to cut costs further, I think? 1/5 there

I don't think there's anything stopping you from defining an internal advantage as AE and having that AE damage things outside you, but your own body HP would probably count as "cover DR" for anything outside of you, so that would limit your ability to damage external foes without destroying yourself...

Except of course if you took "Injury Tolerance" because unlike buying DR that won't add to your cover DR...

Kind of a neat trick there, potentially. Hard to exploit for low-cost characters due to the high cost of injury tolerance, of course.