But why isn't this an appropriate test for arising-under judistiction under § 1331? It's interpreted as less broad than the "arising under" category of Article III, even though there's evidence to suggest what Congress didn't intend anything narrower than Art. III.
So, this isn't a case about what the constitution would permit, it's a case about what the statute permits. Now why do we think those two differ? Osborne gave huge breadth, interpreting the constitution-- why do we want it narrower? Well, sheer load, for one thing. Still, why say the Constitution is broad, and the statute is narrow?
Well, constitutional interpretation is fixed (it's hard to change the constitution), whereas congress can keep making new statutes-- so if statutorily jurisdiction is found lacking, it can always be expanded. On the other hand, when interpreting the constitution, the court doesn't want to limit what Congress can do.
Still why worry about the constitution being too narrow, and the statute being too broad, when the language is the same? It's the role of Congress: they can save the court from the flood, or adjust the courts, etc. So we're setting out the canvas on which Congress can design the federal courts, and then we interpret the statute narrowly (and as a side benefit, we get out of having to overrule Osborne. So the real issue here is the scope of Congress's power: there's a vast pool of cases that could be assigned to the federal courts, but we assume that Congress wants their will to be interpreted modestly.
Anyway, the result is that text can mean different things depending on whether it occurs in a statute or the constitution. § 1331 lets us put the Constitution aside (except maybe for Shoshone, and deal with interpreting § 1331.
So what are federal courts for? They're for what Congress wants. But where congressional language is limp or vague, it requires an infusion of meaning from the courts.
Footnote: it turns out that Bendectin actually didn't cause birth defects, and it was pretty much the only good anti-morning-sickness drug, but it was driven off the market because of these suits.
So, as Congress, would we want drug label tort cases in federal court or state court? On the one hand, the tort causes of action arise under state law, so maybe they should be heard there. On the other, we'll get varying interpretations of the federal drug labeling statute from state to state. Many of them may be unscientific, and governed by runaway juries, compared to rational and expert federal courts. On yet a third hand, we certainly don't want to say that any time there's a federal law element in a state law cause of action, the federal courts have to take the case-- that will crush the federal courts.
Jurisdition is power, but also duty: the power to hear a certain kind of case is also the obligation to hear it. So § 1331 is what we rely on to keep the floodgates from opening.
Here, and at the end of Grabel (p. 286), the court talks about respect for federalism, but really it boils down to resource allocation, at least part of the time. Some kinds of cases (and personal injury is maybe one of them) are handled more efficiently in state courts. That's a little snobbish of the federal courts, but they want to conserve their resources for the kinds of cases that they are obligated to take.
Irony: the dissenter, Brennan, is one of the most liberal softies on the court, and he's trying to help the drug company get into federal court. Of course, he's motivated by the fear of anti-federalism: whatever state has the most stringent requirements will effecively be legislating for the whole country, because the drug company-- wanting its wares to be salable everywhere-- will comply with the highest standard.
We know that this is not a matter of federal interest, says the majority (Stevens), because if Congress had cared, they'd have supplied a federal cause of action. When we have a federal element, we look to the substantiality of the federal interest at stake. Of course, as Brennan points out, there's still a federal interest in controlling the interpretation of the law, even if we don't have private attorneys general: in fact, that indicates that congress wanted to control the application of the law.
Note that neither side wants to see a private cause of action found under the federal law: the plaintiffs want to stay in state court, and the defendants don't want a flood of lawsuits. The defendants cant remove on diversity grounds (they are headquartered in OH, and you can't remove in your own home state). Why doesn't a home-state defendant want to stay in state court? Usually it's the other way around. Maybe because the federal courts are a little better at getting this kind of law right and more economically realistic about awards.
The plaintiffs don't want to sue in their home countries because they don't want to see the suit dismissed on "forum non-conveniens" grounds. A suit in Scotland or Canada will fail on the merits, because it's hard to recover there. Suing in any other state will allow the the defendant to remove to federal court, and then move for dismissal forum nonconveniens to get the case shipped back to Scotland or Canada, where it will lose. They're so worried about this problem that they come to Merrell Dow's home state in order to sue; and that's why Merrell Dow wants to get into federal court, so they could get out of the US. The US is a litigation magnet for foreigners who want to find liability. Anyway, both sides are forum shopping.
This might be good jurisdictional theory, but Frankfurter says it's not good constitutional reasoning. These are scholarly debates.
This contrasts with Brennan's dissent in Merrell Dow (see p. 276).
Why, after all, did the court sacrifice the clarity of that nice black-and-white rule? Why introduce so much complexity? Why allow for a few rare cases where there will be federal jurisdiction despite a lack of a clear cause of action? What is so important about those cases, and what are we trying to do that's so good?
The majority in Merrell Dow COULD have just said that Holmes got it right (in American Well Works), and say that the absence of a federal cause of action meant no § 1331 jurisdiction, Why not just skip all the stuff about Congress being indifferent to it and the issue therefore being insubstantial?
The rule is stated in Grabel on p. 284 (and we have it laid out in a 4-point box on p. 287). Note that here, as opposed to Merrell Dow, there's also not a federal cause of action, but the federal interest is deemed to be sufficiently substantial. So can we recognize substantial issues when we see them? They're rare, but when they come up, the court really wants to get them into federal court.
So an isolated case in state court, with a huge federal interest, might get us into federal court. On the other hand, a basic personal injury case won't make it, even though it has a minor federal element (Merrell Dow). This is that little extra bit of jurisdiction that you can get under § 1331, that differentiates it from the Holmes rule.
No, they didn't forget it: Shoshone was a federal cause of action, and the box on p. 287 says "state cause of action."
The cause-of-action test is slightly under-inclusive of § 1331 jurisdiction (see Smith and Grabel). But the cause-of-action test is also slightly over-inclusive, in the sense that it would sweep in Shoshone.
So Shoshone tells us that you can have a federal cause of action, and still not have federal jurisdiction. Maybe this isn't a constitutional problem (the issue could be forced, and we could learn for sure, if there were an explicit grant of jurisdiction bundled with the cause of action), but rather a workload problem: the court may not want these cases (i.e., "we don't want to be swamped with a bunch of local mining disputes").
It's the nature of the federal interest, maybe. The federal government isn't a party to the case, by the way.
But there are some federal interests so substantial that we're distrustful of courts handling them at all. See the discussion of the welcome mat on p. 286. Merrell Dow set up a pattern that would exemplify future cases, whereas this case just rules on a procedure by the IRS-- the question is discrete and won't keep being re-asked. And it has a strong federal interest.
The case is important because, as in Smith, the court is protecting US financial interest-- the Holmes test would be simpler, but it would also exclude a bunch of cases that we really would want federal courts to hear.
Do we have this doctrine because we don't trust these issues to state courts? Maybe so, but it's congress mistrusting them, ostensibly. It's also pretty limited: we want these three very important acts protected, but so far it's really only these three. Basically the complete preemption doctrine is the court clearing away any lingering state causes of action, especially in areas where the claims could have been brought in federal court anyway. Those state claims don't exist any more, in effect.
We talked about this a bit during the discussion of standing. So this is limited to cases of actual controversy, but anyway we do have the declaratory judgment act.
In Skelly, the court says that the declaratory judgement act doesn't expand jurisdiction. Basically you just visualize the lawsuits that would have occurred if there were no declaratory judgment.