Cert the 9th Circuit Before It Kills Again

by publius

The Ninth Circuit made a potentially big decision yesterday – Fair Housing Council v. Roommates.com -- that could significantly increase website companies’ liability for content posted by their users (I’ve posted the decision below the fold as an iPaper). Eugene Volokh and Susan Crawford both have interesting things to say about the case. Volokh thinks it’s both correct and fairly inconsequential. I’m not sure I agree. To me, this case illustrates the stark divide between law-in-theory and law-in-practice. As a purely theoretical matter, I tend to agree with Volokh that the decision isn’t a big deal. But when you consider the practical implications, I think it becomes quite harmful.

First, some background. Let’s say that, in the comments section here, Gary Farber accuses “Cleek” of having poor indie-rock sensibilities. Outraged, Cleek sues Typepad (the host of our blog) for defamation. Typepad, however, didn’t actually say anything about Cleek's music tastes, it just made blog space available for others to provide comments.

In this case, Cleek would be out of luck because of 47 U.S.C. § 230. Section 230 grants immunity to websites and service providers for content posted by other parties that use their sites. Thus, you can’t sue Yahoo for offensive statements made on chat boards, nor can you sue Google for content that its search engine pulls up. It’s a good law, and it’s been interpreted quite broadly over the years to avoid “chilling” Internet activity. For instance, Blogger wouldn’t exist if the company could have been sued for every random comment on blogspot sites.

With that in mind, the next relevant law is the Fair Housing Act. To be grossly general, it prevents housing brokers from discriminating on the basis of, among other things, sex and family status. Because brokers can’t refuse to sell to families, they generally can’t even ask whether you have children.

Enter Roommates.com. As the name suggests, this site allows people to find and provide housing. To get started, the site's users must enter information about themselves in various prompts (they were drop-down windows as I understand). Some of these questions involved children and gender (e.g., “Children present” or “Children not present”) that would normally be illegal.

Long story short, Roommates got sued under the Fair Housing Act. Roommates responded that they are immune under Section 230 because its users were the ones who actually entered the information. The court disagreed, finding – and here’s the key – that the structure of the question prompts was itself illegal content creation. In other words, the question prompts themselves were illegal because they required people to answer illegal questions. Critically, the court went on to find that the site’s search engine also lacked immunity because its results were based on these illegally-structured question-and-answers. (This is a very brief description, so read the opinion below the fold if you want more detail).

As a purely theoretical matter, the decision seems harmless enough. As Volokh explains, Roommates was “channel[ing] the speech in likely illegal directions.” Also, as Crawford notes, the logic of the case could be limited to the more egregious facts of this case. If it is so limited, the decision won’t be that big a'deal.

The problem, though, is that the case will create big problems in the real world. Specifically, it will be impossible to cabin the case’s logic to these specific facts. Vague holdings, like children, tend to grow more expensive through time. If the Supreme Court lets the decision stand, I predict that it will significantly increase litigation and chill Internet activity (e.g., sites like Roommates will be much less efficient as prompts become bulletin boards).

The reason I’m skeptical is that litigation if often done in bad faith. As any real litigator will tell you, the point of litigation isn’t necessarily to vindicate a right, but to harass an opponent with discovery, document productions, and other expensive tactics. So long as a claim is plausible, you can inflict real damage (and maybe get a favorable settlement) even if you think you will ultimately lose.

The beauty of broad, bright-line immunity under Section 230 is that it allows parties to end this type of litigation quickly. If Google or Craigslist gets sued for content its users create, then the companies can immediately file a summary judgment motion (or a motion to dismiss) and end the case before discovery begins. If that line becomes less bright, then it’s harder to shut down frivolous (or at least losing) litigation early.

As Crawford notes, this decision clearly blurs that line. The real-world implication is that litigation will become easier to start and harder to finish. If Google attempts to dismiss a lawsuit, the plaintiff can always respond that broad, expensive discovery is needed to look at the inner workings of the search engine algorithms, or to depose the people who engineered the prompts, or whatever. In fact, the mere potential of expensive litigation will require companies to spend scarce resources on compliance and oversight to make sure their programs don’t cross the line.

Appellate judges need to realize that making bright lines fuzzy vastly increases litigation costs and business expenses. These increased costs are not necessarily a sufficient reason to abandon bright lines. But when you think about the immense benefits of keeping the Internet humming, we need to be very wary of imposing these sorts of costs.

Case below - (by the way, this feature seems pretty cool):

Read this doc on Scribd: fair housing 9th cir

Comments

[Jesurgislac]

Sebastian: Also 20 points is the difference between a 'C+' student and a 'B+' student. Are you saying that grades are SO USELESS in terms of academic merit that the difference between C+ and B+ is totally erased by race?

Quite possibly: certainly the Michigan University admissions board seem to feel that it is, and I suspect that they've probably given more time, attention, and have more actual experience of what things are like in Michigan high schools than either you or I do.

I'm perfectly willing to admit that hairline differences in academic merit aren't likely to show up on tests, but we aren't actually worrying about hairline differences.

The difference between a 'C+' student and a 'B+' student in high school? Yes. Actually. I would say that is a hairline difference. Is the B+ student very bright and coasting? Is the C+ student very bright and up till 1am every night taking care of baby brothers and sisters? High school is a study environment wholly different from college: a student that does well in one won't necessarily do well in another. Grades are not meaningless, but arguing back and forth about the difference between B and C? Yeah.

[Gary Farber]

"Also, what is IIRC?"

If I Recall Correctly.

I direct the gentleman's attention, as general reference for future use to here and here, and in this case, here.

The expression has actually been used widely in science fiction fandom since at least the early 1960s.

[Sebastian]

"Grades are not meaningless, but arguing back and forth about the difference between B and C?"

I think you don't know the US grading system very well if you believe that the difference between a B and C (on average over all of your classes) is a hairline difference.

And the difference between the worst possible score and the best possible score on the SAT? Also a hairline difference? Are you seriously go to claim that?

[liberal japonicus]

Yes, but aren't we discussing what those conditions should reasonably be?

Well, I guess that is a call, so I'd have to say that anything that makes the business a profit should be regulated. This is just a first pass, but that is where I would start.

[Jesurgislac]

I think you don't know the US grading system very well if you believe that the difference between a B and C (on average over all of your classes) is a hairline difference.

In measuring the merit of a teenager at high school with regard to their ability at college? Yes.

[Jesurgislac]

Sebastian: Ok, so if I say that I want to ban legacy considerations can we kill affirmative action?

Certainly: but you'll have to wait at least a generation after you ban legacy considerations. The whole point of affirmative action is to balance out the legacy considerations of racial privilege.

[Sebastian]

"In measuring the merit of a teenager at high school with regard to their ability at college? Yes."

You say this, but in fact grades are an excellent indicator of how a teenager is likely to do in college.

[Bernard Yomtov]

Sebastian,

You continue to assume that my comments are simply a rationalization for race preferences in admissions. They are not. They are a response to the claim that admissions should be "purely meritocratic, end of story."

I have simply been trying to say that merit is not clearcut, and in fact cannot necessarily be measured by a single number, which can then be used to rank applicants.

Suppose we wanted to accept applicants into some program or other based strictly on "athletic ability." What does that mean? Strength, speed, hand-eye coordination, reflexes, spatial sense, throwing accuracy? All these things are part of athletic ability. Similarly, lots of things are "merit." Until we decide whether we're after sprinters or defensive tackles we can't compare. And if we want some of both, and a centerfielder to boot, the choices get tough. This is not to say that there is no broad attribute we can call "athletic ability." I know this in part because I have little such ability. It just means that it's not like measuring height, and that it may be a meaningless metric in many cases.

I think, that Dr. Ngo explains all this much better than I can in his 2:16 comment. I urge you to reread it.

In fact, BTW, your assumption is wrong. I am not a huge fan of race-based preferences in admissions. I think that the process should take economic status and related matters into account, for a variety of reasons. Applicants from private schools, or wealthy suburban public ones, are going to do better on SAT's, have more AP classes, etc. than those from poor areas. If you want to call this "class-based AA" go ahead.

Nor do I consider legacy admissions an unmitigated evil. To the extent that they stimulate donations and thus enable the university's to admit students who could not otherwise afford to attend, or to improve its programs, they have some value. I suspect they are overdone, but that does not mean they are worthless.

So admissions is complex, and my main point from the beginning of this discussion is that blustering about "all admissions should be merit-based" is empty rhetoric.

[Sebastian]

"Until we decide whether we're after sprinters or defensive tackles we can't compare. And if we want some of both, and a centerfielder to boot, the choices get tough. This is not to say that there is no broad attribute we can call "athletic ability." I know this in part because I have little such ability. It just means that it's not like measuring height, and that it may be a meaningless metric in many cases."

No it still isn't a meaningless metric. Taking your athletic ability example, we would still know that hair color, number of teeth, ear size, and regional dialect aren't 'athletic ability'.

And I think you are on to something here with the sprinters and defensive tackles thing. If you want mathematicians you measure math ability. If you want English teachers you measure writing ability. If you want language experts you measure language ability. But all of these have merit-based measurements even if they aren't all the same measurements. Saying that different disciplines have different merit-based dimensions is nothing like saying that merit-based dimensions can't be the central focus of admissions. (And again, when we are talking about an entire GPA level for race we aren't talking about a small 'plus').

Similarly, no matter how hard it is to precisely define academic merit, we know that it doesn't include hair color, number of teeth, ear size, or race. Like many ideas that are difficult to precisely define, they can still exclude all sorts of things that aren't anywhere near the margin.

Race isn't near the margin.

"So admissions is complex, and my main point from the beginning of this discussion is that blustering about "all admissions should be merit-based" is empty rhetoric."

What you mean is that it would be empty rhetoric coming from you. That doesn't make it empty in general.

[felix culpa]

Gary:
Thanks for the links. Perfect.

[Jesurgislac]

Race isn't near the margin.

I have no idea what that means.

Anarch has a point that a lot of racial discrimination can be overcome by class-based discrimination.

But straightforwardly, if a college decides that part of its goal is to have an ethnically-diverse campus, that this will have a positive effect on students from diverse backgrounds, I don't quite see what the issue is unless there's evidence that the college is admitting substandard students.

For example, if a college decides to admit students strictly by grade/SAT score, top down, pure "merit"... but that doing so would mean that the college campus would be 99% white because there is a top layer of top students who went to schools which taught them how to pass the SATs with top marks and pushed them to get top grades - and well, gosh, 99% of the kids who get to go to those schools just happen to be white - well then, the college will have to add race as a factor to ensure that they also get a diverse ethnic mix. And your problem with that?

[Bernard Yomtov]

If you want mathematicians you measure math ability. If you want English teachers you measure writing ability. If you want language experts you measure language ability. But all of these have merit-based measurements even if they aren't all the same measurements. Saying that different disciplines have different merit-based dimensions is nothing like saying that merit-based dimensions can't be the central focus of admissions. (And again, when we are talking about an entire GPA level for race we aren't talking about a small 'plus').

Are you concerned about the size of the plus, or its existence?

If we are trying to educate students is there anything at all to be said for attaching some value to providing an environment that deliberately includes people from diverse backgrounds? (I tried to avoid the "D-word, because it tends to set people off, but there it is). Suppose, in the extreme, your "merit-based" standard produced an all-white class. Would you simply shrug and proceed?

...it would be empty rhetoric coming from you. That doesn't make it empty in general.

It would be empty coming from anyone who declined to explore the issue further. I point out that your notions implicitly measure "merit" in terms of expected outcomes. You want to admit the students who will perform the best. That's a defensible policy, but not a slam-dunk, and it hardly covers all notions of merit.

College admission has a strong element of prize to it. Going to Michigan, or Harvard, confers advantages that go beyond the content of the education received. How that prize should awarded does not seem like a simple question to me.

[Sebastian]

"Suppose, in the extreme, your "merit-based" standard produced an all-white class."

Suppose in actual practice the non-merit-based standard discriminated very strongly against Asians? In fact, they do. Did Asians do something to deserve being strongly discriminated against in college admissions?

You have to reach for extreme hypotheticals, I get to use the actual facts.

In extreme hypotheticals, the merit-based approach could be bad IF people of a particular racial heritage were incapable of doing well.

That isn't actually true. I know of no studies which suggest that any particular racial heritage is less capable of doing well enough to compete for college admissions in a meritocratic system. I know of very few people who argue that even without studies.

Also in actual fact, merit-based admissions have a history of making it more difficult to discriminate against people of particular ethnicities.

So in very counter-factual hypotheticals my approach might break down. But in reality, it doesn't.

The real question is why people of particular racial categories actually perform differently on meritocratic measures(both before and in college) despite the fact that they do not appear to have innate deficiencies. I suspect that whatever the answer is, correcting it would be best done long before they have accumulated 10-12 years of doing worse academically than similarly situated Asian students who in fact have to be discriminated against in order to create modern racialized admissions policies.

The problem with changing to a plus for economic status is that it wouldn't come to the 'right' outcome because poor and middle class Asians tend to perform better than poor and middle class people of other races, and as such would need to still be discriminated against in order to get the proper racial balance.

(Please note that I don't actually believe in a proper racial balance and am raising the problem from a point of view that I don't hold.)

[Gary Farber]

Possibly this and this and this might be of interest to some.

I may or may not get to comments on a lot of stuff said in this thread in the past few days, but geez, an awful lot of talking past each other, mind-reading, responses to things nobody said, but assumptions that they must have meant it, and other wastes of time going on here, I think, as well as a lot of very well-put and thoughtful points.

[Gary Farber]

"I know of no studies which suggest that any particular racial heritage is less capable of doing well enough to compete for college admissions in a meritocratic system."

As you say, few would argue otherwise; I sure wouldn't.

But that's not, in my view, the relevant question. IMO the relevant question is whether people viewed as one particular type of person or another are actually in circumstances where they can, in fact, compete equally, and from an equal position of support, as people of seen to be of another, more privileged, background and environment. What do you think?

Now, this isn't an argument for mandating equality of outcomes, nor anything more than a single question, amongst countless equally relevant questions and points, but I do think it is a necessarily relevant question.

[Jesurgislac]

The real question is why people of particular racial categories actually perform differently on meritocratic measures(both before and in college) despite the fact that they do not appear to have innate deficiencies.

Sounds like an argument for providing free, high-standard nursery education for all children from the age of 2, followed by free, high-standard grade school education for all children from the age of 5, followed by free, high-standard high school education for all children...

...of course, that would entail not just one but a whole series of administrations determined to get high-quality education for all children from an early age, all at the taxpayer's expense. Which would meet with determined, concerted conservative opposition. Not least because the easiest way of ensuring that all children get to go to nursery school is to provide nursery school as part of free daycare for working mothers. And my goodness, the socialist/feminist implications of that...

Plus, if you're determined to provide good-quality education for all children, you also have to provide free basic health care for all children: a kid who needs specs to read will need to have those specs provided for free: a kid can't concentrate in class if tooth decay is rotting his jaw...

In short, Sebastian: if you want everyone to exploit their full potential, you have to live in a socialist country. As a conservative, I don't doubt you'd fight the idea tooth, nail, and screaming about taxes.

[Anarch]

Sebastian: I think you don't know the US grading system very well if you believe that the difference between a B and C (on average over all of your classes) is a hairline difference.

I do, and I think that such a difference is not in any way dispositive of a student's success in college. Mind, I've been working with a relatively small sample here, but I'd say there's going to be more impact from race than a B-C differential, almost certainly.

What impact is an entirely different question, though...

Jes: I know you're a gorram furriner, but it's the University of Michigan, or just Michigan, not Michigan University. That's just... wrong.

Sebastian again: And I think you are on to something here with the sprinters and defensive tackles thing. If you want mathematicians you measure math ability. If you want English teachers you measure writing ability. If you want language experts you measure language ability. But all of these have merit-based measurements even if they aren't all the same measurements.

Well, this is where things get dicey. If you want mathematicians, yes, you measure math ability -- assuming you can do so which, even in such a precisely-defined field, is bloody difficult to do. If you want math teachers, though, it's an entirely different story. [Ditto for English teachers AFAIK, but since I'm familiar with Math Ed. I'll stick with that.] Measuring that... I don't even know how I'd begin, and I've been on our TA Evaluation Committee for a several years now.

[Which means I know how we do it, and that it ain't right. And that I lack the power to do anything about it.]

But I'm not sure what the larger point is here. The American system isn't the British system, where one applies not just to a university but to a "major" within that university. [E.g. Warwick for Maths, Bristol for Geography, etc.] Admissions Officers aren't selecting members for a given athletic team, they're trying to select individuals who will integrate together to form the best student body for the school. Whatever that means, which it doesn't. What the heck is the analogue of "student ability" in this more general context? Regardless, it's not at all clear to me that increasing the number of B-C students while decreasing the number of black students -- assuming zero sum, which everyone seems to be -- speaks to Michigan's desire for a better student body if one's conception of "better" is broader merely than GPA.

Obviously YMMV... but that's one of the reasons why we're not on Michigan's Admissions Committee and they are. We don't get to make those decisions because we don't represent the University of Michigan. Period. And I have no respect or sympathy whatsoever for the spoiled brats who sued; life ain't fair, honey, and if you're the kind of petulant whiner who can't stomach that the privilege of your race didn't extend all the way to admission at one particular university, well, you lack the character to have gone there and they were right to have bounced your sorry ass in the first place.

Sheesh.

Anyway, for the general saga of "merit-based" and the metrics under which admissions might be considered, I'd suggest everyone re-read this, which despite being a touch combative -- and despite my personal stake in it ;) -- is pretty damn identical to what I would've written if I could be so cogent, and so trenchant.

[Gary Farber]

"I'd suggest everyone re-read this"

Yer link is busted.

Otherwise I agree completely.

[Jesurgislac]

I know you're a gorram furriner, but it's the University of Michigan, or just Michigan, not Michigan University. That's just... wrong.

*facepalms* Sorry.

[Anarch]

Damn you, linkbustery! Meant this by dr ngo.