I have a problem: When designing security as the GM, I often forget just how CHEAP drones are now. Basically, for every human guard with just a modicum of training and equipment you can get two drones, which then cost basically nothing much, while the guard keeps eating wages.

Then there is the issue of how to control them: A drone on pilot is pretty piss poor at everything, unless you invest in a good pilot program, autosofts, etc. So, a drone wiht a DP of 6-8 isn't much good at actually killing stuff, and they also won't be able to make many "smart" decisions. On the other hand, they should be able to spam a lot of lead, which generally takes care of hitting issues.

I still feel that probably i'm not using them nearly enough. I use them for colour, and basically as mobile camera systems, where I should be spamming then. Then there is the issue of how to control them. As I said, on pilot they aren't that dangerous. With a jumped in rigger this can change pretty quickly (if their sensor packages are good enough). Then there is the problem of how secure to make the networks that run them. It seems to me that if security is too light, then you basically just gave your runners a present of a multitude more guns and eyes on their side. But realistically, what numbers should a drone network have?

So, my questions:
How do you avoid this issue? Is using drones second nature to you? What is your ratio of human guards to drones? How many of them are flying about in your everyday city sky, or rolling around in your everyday city streets? How scared are your NPCs of drone theft, or rather, how scared are you that your PCs will just see ubiquitous drones as an invitation to lining their pockets? How secure are your police or corporate drone networks?

And of course, how do you configure opposign drones? Are they customized or do you deploy them basically straight from the book?

Only those with good matrix security uses them with me (And a few nutjobs who take wireless off and invest in fuzzy-logic on their robots).

The everlasting fear they could be turned (easier and for longer than a mage could turn people) by hacker and *gasp* TECHNOMANCER people and mow down the staff is too high.

But some fixed turrets with less-than-lethal weapons. Neurostun and Tasers... sure, why not.

Also: yeah, Drones are INSANELY cheap at the low end.. implying that there are no low-level jobs anymore and a 50% unemployment rate... not good for the game. I think crnking the costs up by 5 times or so would be better. (This would impact maintenance etc. as well.) The cyborgs and stuff... they are priced right. (Robot capable of functioning as a good human etc. for 200.000+ is fine)

Actually the cost of a cheap sensor or two (and not to open that failure of RAW again) is a great way to improve a drones performance. The pilot+best applicable sensor is the way to go. So if doing gunnery you could add a pilot of 4 + a rating 6 thermographic camera + 4 targeting auto soft and a +2 dice for smartlink as well as another +2 dice for fuzzylogic (gunnery is always a complex action) and then the topper +2 dice for a tacnet for a total of 20 dice.

It is not that hard, but it depends if you allow drones to benefit from the smartlink and how you handle upgraded sensor packages (read arsenal-the rules are screwy). As to hack ability, make them slaved to the riggers commlink--where the rigger is running and a decker--ahem systems security specialist is camped and hacking their drone network should be difficult.

I'd say they are used more often for security than not with a typical office have at least a few (mostly-aka all but one) unarmed sensor platforms. The few that are armed are armed with non-lethal weapons such as tasers or gel rounds. For your typical top secret research lab (oxymoreon)-probably at least half a dozen that are armed for bear, but that would vary based on the value of the reasearch. Drones are cheap compared to hiring a security guard. The cost of a low level security guard is 12,000 nuyen per year + medical benifits, while the initial cost of the drone is about that much, after a year on the job it pays for itself, and is not bribable.

Ok... correct me if I'm wrong: The way I read it is that when on autonomous mode and using pilot the sensor isn't even included. It's Skill+sensor of rigged in, or command+ skill for command, but just pilot + autosoft (+smart) for autonomous. Which means that basically you are stuck with a hard limit of 14 dice for conventional device ratings.

I think this one of the areas where the consistency of the rules fails again.

Usually we have fixed guns and cameras or drones on tracks, all hard wired in medium level stuff. High end places often have drones as a 2nd layer of defense holding up intruders until the flesh and chrome guards arrive.

Read the rules on gunnery. AFB-but as I recall the gunnery test is gunnery skill (AKA Pilot)+sensor rating+autosoft + other-I don't think it made an exception to the rule for drone pilot programs.

Pilot replaces attribute and autosoft replaces skill. But drones don't use gunnery as an autosoft, they get an autosoft Targetting(weapon) which means it works as described in the "Controlling Drones" section in the core book. And that is pretty precise on the fact that you don't add sensor unless you are rigged in.

Edit:
Actually, when using active targetting the drone should also be able to add the net hits from the sensor test to the shooting test.

So, for instance, when trying to shoot a metahuman the drone could first use active targetting, making a sensor test with Sensor rating+clearsight+signature mod (usually negative) vs stealth or no stealth, and it can try this twice per IP. Even the best non-military drone will have only 7 dice to do that sensor test against a metahuman.

So my group just came into possession of a large compound, with a large security fence around it's large parameter. So, we had some contractors come in and build another one 30M inside the first one. We ran a track along the top of the inside fence, and bought 5 of the tracked drones. Upgraded the sensor and pilot, gave them LMG's and put them to work securing the 4800 feet of perimeter. The fence cost more than the drones did.

I use drones like modern-day guard dogs: an alert, patrolling response force that can provide a minimal level of individual attack ability, a moderate amount of pack attack ability, and - this is the part that separates drones from dogs - and extremely lethal attack platform when the controlling rigger jumps in. So while a drone on Pilot [without lots of expensive upgrades] isn't intended to wipe out the runner team by itself, it is meant to notice they're there, harass them [drawing in other members of the "pack," but without pulling so many they leave the perimeter unobserved: any attack could be a feint], and either suppress them, or keep them busy until the rigger shows up and starts bringing buckets of dice to the show.


One nice thing about drones is that they scale. A base-model drone is good for keeping the riff-raff away, for presenting the appearance of security without necessarily a lot of its substance. Think a modern-day security camera: they reduce crime drastically, but it's hardly like they're going to stop a professional criminal: you scout the target, learn the system, and then cut the power, jam the transmission, hack the server, or just wear a hat, as appropriate. The lock on your front door stops 99 percent of people from taking your stuff, but most locks take seconds to open, a few take minutes. The bank vault takes a little longer.

Drones are the same way. On the low level, they provide the appearance of security, and provide observation and a mobile platform for the rigger [if there is one]. As you scale them, they can get more lethal, have better decision-making capability, more autonomy, and be better at making those initial observations. They can scale to the expected opposition. [If you live in the suburbs, you don't lock your doors; if you live in the inner city, you bar your windows.]

[edit: I should note a lot of what I said is rooted in earlier editions. I don't think drones scale quite as well in SR4.]


If you're the security system designer, you weigh the cost of the system versus the expected loss without it. [Locks are cheap, and they stop most crime, so they're no-brainers. 24/7 monitored security cameras for your empty farmhouse in the country costs a lot more than the loss you're likely to take.] But for GM's who aren't looking for simulations, but appropriate challenges, you really want to calibrate that number to what your team can just barely succeed at. That number is going to be different for every situation, but in SR4, it's often as simple as comparing die pool sizes.


The ratio of guards to drones varies, depending on the facility, expected opposition, and corporation. Some places will have nothing but drones, and a rigger a continent away to attend them. Some places will eschew them entirely in favor of "unhackable" human guards, ha ha ha.

I definitely like a drone-rich setting, in which unmanned autonomous vehicles are common and widespread, for everything from weather observation to long-haul trucking. Because of that, I also use highly secure drone networks, which the players are unlikely to be able to hack...but in exchange, the players also use highly secure networks that the NPCs are unlikely to be able to [quickly] hack [although if they have a mind to, you're typically doomed]. In cases where the mechanics don't support this, we ignore or alter the mechanics [which are our servants, not our masters].


I spend a lot of time just playing with the vehicle creation rules, making new things, so I always have a ready supply of custom drones lying around, but I use drones from the book, or modified drones from the book, as necessary, or if I get caught out by the players [like, always].

Well, the thing is that even the best network security doesn't seem to help much against a determined hacker - especially against a slow hack.

If the drones are running without wireless, you can still always confuse them as in the example text (game, set, match, or whatsitcalled) (as convoluted as that may be) and they will call home.

Then there is the question of how good the drones really are at even detecting an intrustion.

Drones have to use the Signature table modififers, which means that to detect a humanoid they roll a maximum of 7 dice (without tacnet). And then if the human is using any kind of stealth gear or spirit powers this willl quickly be reduced to nothing (much). On higher alert you might give them the benefit of actively look (+3 dice) - but for that you need some sort of indication what to look for. So that's not good at all.

And to some degree, if you have hackers in the group [we almost never do], and you want the group to be able to succeed, you want them to be able to compromise networks, particularly given time and planning. We tend to run [particularly pre-SR4] a pretty heavily encrypted game, so basically everyone gets their own networks and only the most sophisticated hacks can compromise them, but that style of game doesn't work for all groups.


Yeah, in SR4, these are definitely issues. The idea of wireless in SR4 was to give hackers something to do, so you have to make stuff hackable, but that means an unrealistic setting in which just about anybody with a decent wad of nuyen can just hack the world. And you definitely need to beef up a drone's ability to perceive - high level Sensors, careful selection of individual sensors and modifications thereof, Autosofts - if you want it to be able to usefully detect an intruder with a decent level of skill or equipment.


Can you spell that out for me? I'm not sure where you're getting the maximum number of dice from, is all.


If they're not doing anything else, wouldn't they always be considered actively looking?

As 3278 mentions in our game Hackers and Drone Riggers aren't super common choices by the players-a combination of complexity and a desire to have different types of fun. However in every scenario I write I do write up a section on security, and whether the opposition has access to a drone network.

Now if it's an opposition that reasonably can expect to have access to drones then I write it up in an internally consistent manner-considering their budget, what they're protecting and why. In theory it's pretty easy for me as a GM to design an adventure that could not reasonably be negotiated by a team of player characters.

Sensor 6, Activesoft 4, -3 for detecting humans

I use drones a lot on corporate runs - they are cheap, and if a drone gets shot/stolen, you do not need to train a new one. Most of the drones are utility drones though. Also, all my (corporate/law-enforcement) security teams have armed drones, mostly with tasers, SWAT level forces with LMGs and/or revolver grenade launchers.

In tactical combat, the opposing dogbrain drones are mostly used for supressing fire or spamming the targets with taser darts - 10 Ferret RPD-1X with Defiance EX-Shockers will take down most shadowrunners.

@Signature table: my group uses a different signature table:

Oversized Vehicles: +6 (Trains, MBTs, Construction Vehicles, Zeppelins, Airliners)
Standard Vehicles: +3
Drones, Metahumans, Critters (>= cat): 0
Minidrones, Rodents: -3
Microdrones, Insects: -6

Right, but that's not the maximum number of possible Sensor dice, though, right? What about things like Vision Enhancement? Doesn't Radar use some different rules, as well?

Ok correction, was AFB. You're right it is Pilot (6)+clearsight(4)+smartlink(2)+tacsoft(4)+Fuzzy Logic(2)+bonus dice from sensor test (we'll say 2). Total 20 dice....cost of course is an issue. Just to get there is about 30,000 nuyen give or take before the cost of the drone and weapon.

Worth noting that ferrets can't actually take weapon mounts. There really SHOULD be a 'tiny gun' mod option for smaller drones, but there really isn't.



If your drone has a camera, and is using it to spot people, then it gets vision enhancement bonus.


Worth noting: Smartlink does not apply to perception tests. That looks more like an attack test, though, so I think you are mixing up Clearsight and the Targeting autosoft.
I know of at least two bonus dice you could be using there instead. You may also have better luck using a combination of Active and Passive sensor targeting for the shot, instead of active targeting supporting a normal attack roll.

well, you always have the option of a special machinery tiny weapon mount (like in the dragonfly drone)

Which requires GM approval, because there is literally no cost or modslot usage for it, so most won't even bother to come with one.

well, just take a weapon mount, and limit it to light pistols and tasers.

A weapon mount just so happens to be an illegal installation on minidrone class drones, sadly.

Other players have told me that this kind of thing existed in previous editions, but its utterly lacking in 4th.

..... well, just take a weapon mount, and limit it to light pistols and tasers and call it special machinery mod (with identical mod slots and cost as the normal weapon mount)

Hmm... ok, I had Pilot(6)+Targetting(4)+Tacsoft(which I don't really want to be 4 unless it's a high-level installation) + smartlink(2), and then the result of the sensor test. I don't see how fuzzy logic is supposed to help, I don't think it's supposed to apply to shooting. (But I haven't looked it up lately.)

Going by RAW, that is:

Sensor (6) + Clearsight (4) + tacnet. And maybe you could add vision enhancement, but then you have to work against vision. I like to use sensors as a compounded rating without worrying too much about the stealth tech and type of sensor employed. Then the signature table gets into that. Vision mods get into that when using vision. Stuff like Ruthenium on the PCs loses more dice, and anyone worth their money will have concealment with at least 6 up when going into a highly secure facility - especially since that strangely works on the astral, too. So really there are really not many dice left. And that is on the BEST available equipment, which off the line drones just don't have. (Pilot 3, sensor 3, etc.)

Now I know I did a sensor perimeter once and somehow it had a lot of dice, but I probably did something wrong.

The obvious solution is to change the signature table.

Why is drone perception so different from metahuman perception? Really, the two should use the same mechanics, and a Drone of 3s should be just about as good as [but more dogbrained than] a Dude of 3s.

Well, that's what I think, too. Except the game as it is disagrees. Well... one more thing to house rule.

So, considering we get these perception issues out of the way, how would you go about securing a corporate drone network?

Let me put down my loose thoughts, and then if anyone feels I'm doing things wrong please point out my mistakes:

Obviously communication of all subscribed drones will be encrypted. They should be running a firewall and analyze on their own nodes.

Non-subscribed drones should run with wireless off by default, unless they meet a situation they can't handle. They will then connect to the network and alert the rigger/spider. Surveillance on important facilities is run mostly with wired/rail drones so those can be connected all the time without too much risk of intrusion. Less important facilities will probably still run with wireless, even when just using cameras, becuase setting that up is so cheap.A few wired cams will cover the most important sections, and drones can be additional eyes and ears.

The network itself will obviously not be switched off, but it will run in hidden mode.


Now for the next problem: Police drone networks. I'm thinking the cops have two problems: They are notoriously spread thin, but they still need to be able to run up a profit.
So... they have to cover entire cities with drone networks or security cams. Cams will be wired - the trouble is they will be prone to destruction in troubled neighbourhoods. A high-flying drone network is harder to destroy, but has the usual problems of being sensitve to hacking. I would simply give it a straight 6s system, but when that fails, tough luck.

Oh I was talking about the gunnery. For spotting yeah it kind of goes like this:
Sensors(6)+Clearsight (4)+Fuzzy Logic (2)+Tacnet (4)+ Actively Searching(3??)-3 looking for a metahuman=16 dice before visabilty modifiers. The cost is about 12,000 nuyen depending on the sensors and options selected.

But Fuzzy logic gives a +2 bonus to anything involving a complex action, and a +3 for anything taking more than a minute-where it has time to run additional analysis..

As to drone networks--best thing to do is to slave them. THen for all practicality they can only be spoofed (basically sending false commands). I would expect the drone to a have a decent firewall (4 or better) and some Agents with an attack program, trace (especially KE or Lone Star), and analyze.

A possible solution for surveillance drone would be to have them communicate with their 'base' through tight-beam radio and/or laser setups, with the non-directional wireless used only as a backup if the normal communication fails. Add some ruthenium coating and radar-absorbent materials to make the drone hard to spot. For extra fun, have several transmitting stations in different locations (say one on each precinct) and jump between them at random intervals or as soon as there's a hint of someone is messing with the signal.

It means that to hack the drone you'll have to spot it, figure out where it is transmitting and intercept the directional transmission without tripping the alarm and having to intercept in another direction.

that's not as good as a wired control but it comes close.

Great ideas. Problem, though: It seems if all those things were common, then common off-the-shelf drones would have at least some of them, right? And then suddenly they aren't so cheap anymore. If most police drones had Ruthenium, laser comm and whatever else, then these would eventually turn up on the black market.

What you're basically suggesting is that regular off the shelf drones don't cut it. That was sort of my thought, too, but...

Honestly, you guys are forgetting the basics.

Police already use non-standard radio frequencies, and people still buy police band radios and things for their cars every now and then.

Basically, any non-consumer drone doing essential tasks is *probably* going to be carrying a Nonstandard Wireless Link instead of its usual Signal. Its a really simple precaution, and pretty much makes sense. Real world does it already.

Threshold 5 Detect Hidden Node tests pretty much keep all the riffraff out. Your average hacker is just probably not going to be able to FIND the drone's network in order to hack it. Even, or especially, while its shooting him.

Also laser links, microwave links, and directional antennae are all excellent, and sometimes even cheap.

Not very.
We usually don't have anything that can so much as scratch vehicle/hardened armor . .
And we don't use them ourselves, beause we don't have a dedicated rigger . .
much less someone who really understands the shadowrun third edition drone rules . .

I'll grant you in SR3 there weren't that many drones in the world, and when there were, they were usually EXPENSIVE already. The drone rigger had the problem that he was paying a metric crapton of cash for a super-specialised piece of equipment, which then got basically binary damage: Either it wasn't scratched, or it blew up. Taking over a hostile drone network was a really, really complicated affair, we did it a few times, and it took forever.

I think SR4 isn't as bad, in SR4 there is the problem of what you can get for how much money, and you could literally get tons of cheap drones. Here the problem is the network: Even with 100 cheap drones, all you need is hack the network and you get ALL of them - for a while.

@Udoshi:
You're probably right. I keep forgetting it's actually hard (in a general sense) to detect hidden nodes (because we don't use the iteratively depleting DP rules from 4A at our table for extended tests.) So basically that takes care of a lot of basic problems, and doesn't cost much at all.

Detect Hidden Node isn't an extended test anyway, unless you are doing the REALLY LONG one.

If you assume that most security drones are running Basic Cost-Effective Security packages, you can basically just throw Analyze+Encrypt 6, optimized 3, ergonomic on most drones. With a good Firewall, its gives them decent odds to spot a hacker, and call their remote rigger's e-support for antiviral backup.

If I was adjusting that for The Full Unwired Experience, while still keeping it mostly sane, I would probably throw Purge 6 in there as well as the Homeground autosoft, which gives the pilot actually decent analyze test dice. (remember, automatic non-action Analyze tests on page 228 are everyone's friend). I'd add a Reality Filter as well, because it gives +1 response, and running a bunch of programs gives you -1 response, so its a good way to not lose dice (drones use response for a lot of things) if you are running a lot of Autosofts, or need to boot up an IC.

If i was feeling generous, maybe purge 4 or 5, since some corps don't always have the best software due to budget constraints, licensing issues, or management doing cost-benefit analysis.

That should pretty much be enough to cockblock casual hackers and virus spam from 'secure' drone networks while still being fairly simple.

If you compare the price of one such drone to the cost of equiping says ten city blocks with fixed cameras and their own links, it doesn't sounds as high. Such a drone won't be very common : most corp watching one or a few buildings won't have use ofr it. Only the cops or extensive installations (say military base or industrial zone sized) will use it.

But directional links are rather cheap and you can use them on fixed cams too - since they're usually placed on high spot, odds are you can draw a line of sight to several transmiters.