I agree there is causation. But the alien doesn’t have enough evidence to conclude that there is causation. To the alien, it’s just “cars generally go when lights are green and stop when they are red, I wonder why?”. Any conclusions beyond that are unwarranted and merely based on speculation.
I would call it a hasty generalization. Green lights are not necessary for cars to go (for example, you could be at an intersection controlled by a stop sign, or traffic going in your direction is not controlled at all, or people could run a red light, or make a free turn on red, or move through the intersection against the red light when directed to by traffic police, and so on). Green lights are also not sufficient to go (you might have a green light but not move through the intersection to avoid causing an accident, or because you’re not paying attention, or because traffic police have indicated for you to stop, and so forth). So concluding that “green lights cause cars to go” is wrong, because it’s an overly broad generalization.
What the alien can fairly conclude is that “green lights often cause cars to go, but cars with green lights don’t go, and often cars go without a green light.” This should prompt the alien to wonder if there’s a separate factor beyond green lights which better explains why cars move when they do. It might conclude – and should! – that what actually causes cars to go is this separate factor which has, by definition, more explanatory power.
Am I the first person to mention Post Hoc, Ergo Propter Hoc (Latin: “After this, therefore because of this”)?
I thought it was the engines that made the cars go.
The green light causes the person to step on the accelerator. Stepping on the accelerator causes the car to go. 
Yeah, I’m not going down that rabbit hole mister!
You could say all the same things about cigarettes and lung cancer. After all, some people smoke for seventy years and never get cancer, and some people get lung cancer and never smoke at all. Nevertheless, cigarettes do cause cancer. And green lights do cause cars to go.
You’re still driving even when you’re not moving. Decision-making, including the decisions of when to move and when not to, is an inherent part of driving.
If you don’t believe me, try drinking a lot of alcohol, and then persuade the policeman that you shouldn’t get a ticket for DUI, just because the car wasn’t actually moving.
My car is sitting out in my driveway right now. It has an engine in it yet it’s not going anywhere. So an engine is another necessary but not sufficient factor in a car moving.
A green light causes a person to step on the accelerator? I’ve often encountered situations where the light turns green and I see a person not step on the accelerator. (In a few cases, I’ve been that person.)
There are also cases where people step on accelerators when the light is yellow or red. In many cases, people step on accelerators when there is no traffic light in the vicinity.
So while there is a correlation between a traffic light turning green and people stepping on the accelerator, it clearly is not an absolute one.
If an alien was to observe all of the events which occur in the vicinity of cars beginning to move, that alien would observe that the strongest correlation is between the actions of the person in the driver’s seat and the car moving.
I just read an xkcd that belongs in this thread.
No. Cigarettes can cause cancer, just as green lights can cause cars to go. But “can” is not “do.” You would have me believe that an explanation is right even though it is sometimes wrong. I disagree. An explanation that is sometimes right may be partially correct, but it is not completely correct unless it is never wrong. Given two explanations, the one which better matches observation is to be preferred, even if the less powerful explanation is sometimes or even frequently correct, which is why we no longer adhere to geocentric models of the solar system. This is as true of traffic lights and medicine as it is of astronomy.
Let me be clearer, because I think the part about the scientific method, while true, is a distraction.
The essence of my objection is this: on my understanding of English, “A causes B,” “A can cause B”, “B is caused by A,” and “B can be caused by A” are all similar but not entirely identical claims. If you disagree with this understanding, then you’ll also probably also disagree with my view that evidence for “A can cause B” does not permit one to conclude “A causes B.”
No, the fact that A does not fully explain B does not imply that there is no causal relationship. That’s not what causation means.
With cigarettes we have a statistical association and a well understood mechanism. Cancer usually requires about 6 key mutations, and smoking increases the mutation rate. A proven mechanism and a statistical relationship is the definition of causation.
You seem to be trying to draw a distinction between “causes” and “increases the statistical likelihood of”, but there is no such distinction - that’s what it means to cause something.
We have no complete models for anything, just models that explain less and better models that explain more. There is literally nothing for which B always follows A. Even a complete model can be probabilistic, incorporating random uncertainty. In fact a completely deterministic model for the entire universe would be such a probabilistic model.
True enough, but that has nothing to do with your attempt to define causation as an absolute relationship rather than a probabilistic one.
If we both agree to understand “to cause something” as “to increase its probability” then we agree: green lights cause traffic to move. But I don’t cede to you the right to define “to cause” as “to make more likely.”
If you claim that causal relationships are not probabilistic, please cite an absolute causal relationship where B always follows A.
I am forced to conclude that gravity works differently where you are.
What it means to cause something is a philosophical question. “To cause means to make more probable” is undeniably one theory of causation. It’s not the only one, and I don’t see why I am forced to conclude that because it’s the one you favor it must be correct.
Gravity is a force, not a causal relationship. (And not a force that’s not fully understood, for what it’s worth.)
You can easily refute my claim that causation is probabilisitic by citing some B that is always caused by some A, regardless of whatever other conditions prevail.
I see I was too opaque. If an object experiences a net force, it always moves. In what sense does that net force not cause the motion?
Look, you can go read up on theories of causation (I did) and you can insist that you have the right to decide which is correct. But I am no more obliged to agree with that choice than I am to disprove it.
When I go to the dictionary, I do not see “cause” defined as “make more likely.” When I go to the Stanford Encyclopedia of Philosophy, I see multiple possible theories of causation – I’ve linked you two – among which is one where “to cause” does loosely mean “to make more likely” and another where “to cause” does not mean “to make more likely.”
But here, you claim, there’s no need to worry because you’ve got it all figured out. I hope you’ll pardon my skepticism.
Actually, this conversation started when you were the one who claimed you had everything figured out, by excluding probabilistic relationships from the definition of causation.
If you want to go to canonical definitions, here’s the first two sentences of the Wikipedia article on causation:
This is not a counterexample. It’s similar to just saying: if you account for all the different causal factors, the effect is fully explained.
ETA: So you’d argue that the hundred million or so google hits on “smoking causes cancer” are all erroneous uses of the English word “cause”?
There clearly are “single-cause” counterexamples that you can come up with that it would be churlish to challenge, such as: standing 10 feet from a detonating nuclear bomb causes your heart to stop beating.
But the fact is:
(1) Virtually all practical situations that are remotely interesting and non-trivial have multiple causes, some of which are not fully understood. We don’t require a perfect model to assert that A causes B, and it does not necessarily mean that B always follows A.
(2) Ultimately all abstract physical laws are probabilistic, because they must account for the randomness inherent in QM. So no ultimate physical theory will ever be framed in the form “B always follows A”.
(3) Common English usage clearly includes probabilistic relationships in the definition of the word “cause”.
So, whether we’re talking about the underlying philosophical concept of causation, or common English semantics, it’s simply wrong to say that probabilistic relationships are excluded, in fact they form the bulk of the concept. It’s perfectly accurate to say that smoking causes cancer.