NEWS

Why are cars such terrible drivers?

  • 16 Replies
  • 5818 Views

Top Dog

  • *
  • Ace Runner
  • ****
  • Posts: 1219
« on: <05-19-15/0540:14> »
So this has been bothering me for some time, but it never really affected me. But why are cars such terrible drivers?

Consider: in the real world, right now, (self-driving) cars are generally better drivers than humans. They're not in general use for several reasons - bugs, politics, they still have trouble in some situations - but Gridlink and 60 years should solve those easily. Cars are especially good at reacting - a self-driving car won't have any problem with say someone walking in front of it, as long as it's physically capable of stopping.

Humans, on the other hand, are generally terrible drivers. Even trained ones, but for now, let's just consider the everyman. Especially sudden things appearing are a problem, but they're basically worse at every aspect except stuff like detours and road works.

Now, in game terms, humans have an average reaction around 3, so let's take that. People that drive - today, at least - also have spend a few days training in driving a car before they got a license - let's be conservative and say they have Pilot Ground 1. That means the average human has 4 dice. That's enough to do most of the easy stuff (merging, sudden stops) etc in stressful situations (everyday situations are automatic) most of the times. Sometimes they fail, and they're not good stunt drivers. So far, that seems about fine.

Pilots, however, are terrible. Most have 1 die. Some have 2, and very rarely you have one with 3 dice. That means that, if someone were to suddenly appear in your car (which isn't an everyday situation, so it should require a roll), 4 rolls out of 6 that person is dead. Slightly better odds if people have a better car. But that's the sort of thing self-driving cars are made for - they're supposed to be safer exactly in situations like that! Even a completely untrained driver will out-perform a self-driving car in Shadowrun.

Now this won't come up that often - there's the whole "don't roll in everyday circumstances" clause, and most GM's won't throw random traffic accidents your way while driving to work - but it might come up occasionally, and it makes no sense. Plus, even if it doesn't come up often, it does come up often enough - I remember one of the earlier season 5 mission that called for a driving test for bad (but not terrible) weather. Pilots simply didn't have enough dice to even attempt that - and I can't believe all traffic everywhere stops if there's a tiny bit of snow. There's ways to mitigate it. Maneuvering autosofts are cheap (for the average runner) and add a lot of dice. But those don't come standard with cars. So most cars on the road are still accidents waiting to happen.

So, what do people think about this problem? Am I overreacting? Is it true, but irrelevant? How do you handle car performance in an actual game?



(Not sure if this is Rules or Gear, but it's mostly about cars, so I thought Gear was the better forum)

Lucean

  • *
  • Ace Runner
  • ****
  • Posts: 1159
« Reply #1 on: <05-19-15/0557:23> »
I think it's because the runners should have options to be better than the machines and the rigger being even better.
If cars and especially drones were too good right out of the box, it would require more effort to make a difference in taking manual control.

But to be honest, your points are interesting and didn't occur to me before. I think there should be ways to reflect it from a mechanical side without making drones too powerful.

Herr Brackhaus

  • *
  • Prime Runner
  • *****
  • Posts: 3041
« Reply #2 on: <05-19-15/0656:19> »
To be fair, computers are (supposedly) excellent drivers because they can process much more data and react much faster than almost any human. This is represented in game by the fact that a car driving itself using GridGuide will probably never even have to roll a Pilot Groundcraft test, as most of the vehicle tests are for actions a vehicle would be unlikely to take.

That's just my view of it, anyway; they get around town because of GridGuide, which transcends game mechanic. If you're having your autopilot do a hairpin turn in crowded city streets, you might be doing it wrong.

Ursus Maior

  • *
  • Omae
  • ***
  • Posts: 280
« Reply #3 on: <05-19-15/0744:11> »
I would also argue that modern cars are doing a fine job driving safely from point A to point B, while observing all the rules. But that is not a difficult task. In SR5 this would mean roling vs. a threshold of 2 or 3 maximum. So the dice pool should be 6-9. Considering that these cars will always enjoy a wifi bonus - which I think is missing from the core rulebook, but let's hope the rigger book will arrive soon - they might add 2 dice more for GridGuide and raise the limit by 2 as well.

But looking at today's autonomously controlled cars, they cannot handle emergency situations all that well, when the opposing party e.g. is not behaving rationally (driving recklessly etc.) or tries to hit the autonomously controlled car on purpose. The reaction of Google's car is great for situations they were programmed for. But recognizing what a person standing on the sidewalk will do next - cross the road, don't cross the road, talk to his buddy and not care for the cars on the road etc. - is hard to anticipate for computers.

So I would argue, as long as autonomously controlled cars can manage to safely bring you to your destination in normal traffic, everything is fine. And SR5 reproduces that. Driving Formula 1 or a Car Wars duell, not so much...
Liber et infractus

Top Dog

  • *
  • Ace Runner
  • ****
  • Posts: 1219
« Reply #4 on: <05-19-15/0822:00> »
Well driving without issues shouldn't even be a 2-3; most everyday things (merging, passing, stopping) are a 1. Maybe a 2 to avoid a sudden obstacle (but as you say, cars don't have to autosucceed that one as they're not that. Before modifiers, of course - heavy traffic (which should be the most common situation for most cars in a city!) is a +3.

Often times they won't have to roll, but there are times, even in "ordinary" traffic, that people (or Missions) call for vehicle tests, even when you're driving normally. And as is, a pilot program simply has no chance in most of those situations (and a small one in others).

Frankly, I'm not a big fan of 'just handwave it off because GridGuide handles it' rules - it makes the rules on what exactly is an uncommon situation far to arbitrary and GM-dependant. If your GM handwaves it all away, your shiny Autosoft is a waste of money; if he's ever slightly harsher, the un-autosofted cars in the game will explode as soon as something weird happens.

One way I see to houserule it, is to say:
  • GridGuide lowers thresholds by one, if following it precisely (IE, normal autopilot operation).
  • When automatically piloting with normal parameters, you get the effect of a free +6 Maneuvring autosoft.
That makes every car behave well in traffic, and even in harsher conditions most cars will do fine most of the time. Driving from A to B - assuming no-one is ambushing you - is practically automatic (you now have the dice pool to simply buy hits, for one). Cars should have sensors to deal with adverse light conditions and such as well. However, as soon as you do something different (like command it in a firefight to help you fight better, or exceed speed limits, or make sudden turns), you suddenly drop down to base piloting levels - and you better have a dedicated Autosoft bought seperately (or a good driver) because the autopilot isn't going to be able to handle it otherwise.

Herr Brackhaus

  • *
  • Prime Runner
  • *****
  • Posts: 3041
« Reply #5 on: <05-19-15/1143:56> »
That seems fair as a houserule.

I still say there are a lot of things that SHOULD get houseruled, simply because it isn't worth the hassle. If your team doesn't have a dedicated rigger, you don't penalize them and throw them into a car chase they have exceedingly small odds of succeeding at, just like you wouldn't put a team without a magician up against a metric crapton of magical security to which they have very little ability to counter.

Personally, I don't think there's a need for a houserule here, but your suggestion seems fair if you were to implement it.

Malevolence

  • *
  • Ace Runner
  • ****
  • Posts: 1552
  • Matrix Addict
« Reply #6 on: <05-19-15/1230:21> »
Personally, I feel that drones should come with a Maneuvering Autosoft equal to their Pilot rating so that most driving activities would be performed with a dice pool of Pilotx2. This more closely matches the behavior of things like medkits that can do what they were designed for at ratingx2. For pilot 1 (and even pilot 2) vehicles, the dice pool is still abysmal, but at least more in line with what I think the designers intended - that a rating 3-4 pilot is good (hence its inclusion on higher end vehicles) while 2 is slightly better than average (average driver with minimal training) and 1 is lackluster (i.e. equal to an untrained human driver). They still would be deficient in non maneuvering roles such as stealth, but at least they would be decent at avoiding accidents.


The mitigating factor that would keep them from replacing human drivers is the fact that when the pilot is in control, it obeys its programming and by extension, the law. So it would not exceed the speed limit or maneuver unsafely, making it mostly useless in a car chase. Likewise flying drones with stock maneuvering autosofts would not enter no-fly zones without proper authorization, etc. At the GM's option, buying an aftermarket Maneuvering autosoft could enable such options - obviously police and military vehicles would need to be able to participate effectively in a chase or evade engagement where the only way to be effective would be to drive outside legal parameters.


Autosofts like EW, stealth, evasion and targeting should probably have an R availability anyway, so making aftermarket Maneuvering autosofts "military grade" would leave only Clearsight as completely legal. But again, houserule, but seems like a more balanced way to do it.
Speech Thought Matrix/Text Astral

Vibral

  • *
  • Newb
  • *
  • Posts: 70
  • Deaf Mute
« Reply #7 on: <05-19-15/1332:24> »
I think the basic concept of GridGuide needs a little clarification. Here is how I view it:

GridGuide is not only the car talking to a GPS map, it also has information from ALL the other cars on the grid. So if there is a traffic problem it is alerting all the cars that might be taking that route. It should also be transmitting information as crashes or erratic behavior is detected. I would imagine it is like a giant teamwork test between all the cars that allows it to function as it does in game terms.

Of course I could be totally off base.

Herr Brackhaus

  • *
  • Prime Runner
  • *****
  • Posts: 3041
« Reply #8 on: <05-19-15/1407:01> »
I like that suggestion, Malevolence. Good call.

Mirikon

  • *
  • Prime Runner
  • *****
  • Posts: 8986
  • "Everybody lies." --House
« Reply #9 on: <05-19-15/1759:13> »
Top Dog, the problem with your question is simple. Self-driving cars today drive better than people WITHIN SET PARAMETERS. The programs, while good, cannot react to things outside their programming as fluidly as a person can. This is the same deal with Pilot programs. Not only do they have a programming range that is only so wide, but they've got to rely on the vehicle stats for the other part of driving tests, instead of having a metahuman's reflexes. In normal traffic, or dealing with normal situations, the pilot program is just fine. But it isn't programmed for, say, vehicle chases.
Greataxe - Apply directly to source of problem, repeat as needed.

My Characters

BetaCAV

  • *
  • Omae
  • ***
  • Posts: 303
  • 2-legged devil rat
« Reply #10 on: <05-19-15/2020:58> »
Doesn't using GridGuide (at least) imply inviting marks on your vehicle by the GG system?
I do like the concept of a teamwork test, but it feels like it ought to be between the GG and the vehicle's pilot, rather than between pairs of vehicles (for simplicity's sake, if nothing else). GG may even have riggers on standby to jump in and take over, if the system encounters issues it clearly can't resolve/handle -- like a Force 12 Earth spirit ripping up the roadway, or a vehicle with improved invisibility on it (best advice -- don't do it).

The Wyrm Ouroboros

  • *
  • Prime Runner
  • *****
  • Posts: 4471
  • I Have Taken All Shadowrun To Be My Province
« Reply #11 on: <05-21-15/0425:28> »
Always throw your PCs into vehicle chases, whether or not they have a rigger.  This is the one area (as compared to a major magical or matrix issue) they can actually extract themselves from on their own, afterwards appreciating what a rigger could do for them.  There is little better than demolishing a step-van along with the samurai's rigger pretensions, and forcing the characters to think and shoot their way clear.
Pananagutan & End/Line

Old As McBean, Twice As Mean
"Oh, gee - it's Go-Frag-Yourself-O'Clock."
New Wyrm!! Now with Twice the Bastard!!

Laés is ... I forget. -PiXeL01
Play the game. Don't try to win it.

Rooks

  • *
  • Omae
  • ***
  • Posts: 992
« Reply #12 on: <05-25-15/2326:20> »
Ya I never quite understood why taxi drivers are still a thing

Mirikon

  • *
  • Prime Runner
  • *****
  • Posts: 8986
  • "Everybody lies." --House
« Reply #13 on: <05-26-15/0104:56> »
Ya I never quite understood why taxi drivers are still a thing
Because manual override means hackers can't just get free rides all over the sprawl.
Greataxe - Apply directly to source of problem, repeat as needed.

My Characters

Top Dog

  • *
  • Ace Runner
  • ****
  • Posts: 1219
« Reply #14 on: <05-26-15/0140:52> »
Ya I never quite understood why taxi drivers are still a thing
Because manual override means hackers can't just get free rides all over the sprawl.
You have to ask if the losses from hackers getting free rides are more than the losses from paying all the taxi drivers though.