Back to previous post: A mysterious hatred of fedoras

Go to Making Light's front page.

Forward to next post: Chupacabra

Subscribe (via RSS) to this post's comment thread. (What does this mean? Here's a quick introduction.)

October 5, 2012

Correlation and causation: still just good friends
Posted by Teresa at 06:00 AM * 89 comments

First thing you have to know is that Patrick and I completely disagreed about this article.

So, anyway, Daniel Engber has declared war on “Correlation doesn’t imply causation,” calling it The Internet Blowhard’s Favorite Phrase. He’s wrong. Blowhards have entire suites of favorite phrases, which is useful insofar as it lets us spot them quickly and classify them according to type. “Correlation doesn’t imply causation” is only one of them, and it’s also used by smart, thoughtful, engaged commenters who are acting in good faith.

What’s really going on is that there’s a large population of demi-trolls on the internet who aren’t all that smart, and don’t have much to say that’s original or interesting, but who passionately want people to pay attention to them and act like their opinions are worth something. Mind the gap.*

This means they’re always in the market for universally applicable arguments. When they find one, they make excessive use of it. Some of them have discovered that causality is very hard to demonstrate. I wouldn’t say they understand the principles involved. They just know that people who do understand causality will stop in their tracks if you invoke it.


Some of their other all-purpose arguments and comments:

It’s a slippery slope.
You’re arguing from emotion.
The internet changes everything.
I challenge you to prove me wrong.
I’m not going to do your research for you.
You wouldn’t react with this much hostility if I wasn’t right.
If you cared about this as much as I do, you’d be saying the same thing.
The ultimate in insanity is to keep doing the same things while expecting different results.
(Not a complete list.)

There’s an even easier way to spot Engber’s bête noire, Zyxwvutsr, as a demi-troll. Note this paragraph from one of Zyxwvutsr’s comments:

Engber … correctly noted that risky behavior causes bad health, but failed again to break out of the standard liberal paradigm that says the poor are not responsible for their own plight. He does this by citing a paper that “low income, education, powerlessness, discrimination, and social exclusion” cause risky behavior rather than the other way around.
Stop and ask yourself: is there such a thing as a “standard liberal paradigm that says the poor are not responsible for their own plight”? There is not; so QED, you’ve got a troll or demi-troll using “wrong things liberals think, which I just invented on the spot or am parroting from some right-wing source” as an all-purpose argument.

Pinky swear: a whole lot of what appears to be right-wing grassroots argument online is actually doofs like Zyxwvutsr using “wrong things I fantasize liberals doing” as an all-purpose fill-in. Guys like him are generally clueless about what centrists, liberals, and leftists actually think, and their own ideologies are an incoherent patchwork of borrowed opinions. The one thing they’re sure about is that being a right-winger means they’re right — and they desperately want to be right, even if they’re not sure what they’re right about.*

So how do you deal with them? My favorite method is to calmly engage with the factual content of their comments. Correlation isn’t an on/off leap of faith; it’s a technique for assessing data. Talk about what it’s doing in this case.

One of the cute things about trolls and demi-trolls is that most of them have difficulty unpacking, extending, or modifying their sacred opinions. If you put them in a position where respect is clearly available, but the price is that they have to actually know and understand what they’re talking about, they’ll often just evaporate.

Is this a guaranteed technique? It is not. It’s merely a good one. But it sure as hell beats taking up arms against “correlation is not causality.”

(Note: a sketchy draft of this post accidentally went live a couple of mornings ago, then got taken down. Apologies to anyone who was confused by that.)


Comment #20, from James Moar:

“It’s a slippery slope”

Well, you’ve got to remember that if you use a slippery-slope argument once, you’ll end up using it for absolutely everything.

Comments on Correlation and causation: still just good friends:
#1 ::: Avram ::: (view all by) ::: October 03, 2012, 01:29 AM:

What's that about the Higgs boson?

#2 ::: Tom Whitmore ::: (view all by) ::: October 03, 2012, 02:53 AM:

As you might expect, T, I agree with you here. And the fellow doesn't even prove his point well, with his examples.

Does he believe in global warming/climate change?

Yeah, a correlation does indicate that it's worth spending a little time investigating the correlation, to winnow out if there's some sort of causality involved. But in cases where it's likely to be an artifact, the question of whether it's worth investing the time becomes a real and important one, at least as important as the correlation itself. And remember: about one result in 100 of a P less than .01 is an entirely spurious correlation, an artifact of probability.

#3 ::: Zander ::: (view all by) ::: October 03, 2012, 02:57 AM:

Yes, I was wondering that. Also how does the writer know there aren't any uses for it? Aren't we still at the "look what happens when I rub this piece of amber with this cat" stage, boson-wise? "No, no, Eratocrates, there's no practical use for this discovery of yours. The risk of being scratched to buggery is far too high."

But that may not be relevant to the thrust of the main article...

#4 ::: Niall McAuley ::: (view all by) ::: October 03, 2012, 07:06 AM:

I wonder what happened in August 2008?

(That's a Google Trends graph showing the phrase "Correlation does not imply causation" taking off from zero that month, and never dropping back to zero since).

#5 ::: David Goldfarb ::: (view all by) ::: October 05, 2012, 07:07 AM:

As a card-carrying crazed nitpicker, I will note that every instance of "Zyxvutsr" should actually be "Zyxwvutsr".

#6 ::: paul ::: (view all by) ::: October 05, 2012, 10:14 AM:

I'm in favor of Pascal's Wager, myself. But I realize that doesn't cut much melting ice with true-believer style trolls.

(I'm also very much in favor of Shalizi's Law of Two Tails: any time someone invokes the uncertainty in some estimate or other as a reason to believe the real number is in the part of the distribution they like, I ask whether it might equally likely be in the part they don't.)

#7 ::: Alice Bentley ::: (view all by) ::: October 05, 2012, 10:15 AM:

Niall @#4 - Shadow Unit makes a veritable theme of the concept that correlation does not imply causation, and launched in 2007. I'm not at all sure that this is causal, but it's interesting.

#8 ::: Craig ::: (view all by) ::: October 05, 2012, 10:25 AM:

Niall@4: If I had to make a guess, I'd say it was this. He's currently up to his ~1100th strip, at ~3 per week, so the timing works out approximately.

#9 ::: albatross ::: (view all by) ::: October 05, 2012, 10:59 AM:

One interesting thing about these trollery markers is that they're very common among people who don't intend to be trolling. There is this common pattern in arguments:

a. Alice and Bob are arguing about X. Alice "knows" X is true, based on something other than actual knowledge. Like, X is the conventional wisdom she's always seen upheld, or it's the teaching of her church, or the common belief of people on her side of politics, or the belief assumed by respectable media, or whatever. She's certain without actual knowledge.

b. Suddenly, *any* argument for X seems quite convincing to Alice. Similarly, *any* argument for throwing out evidence against X seems rock solid. Thus, when Bob points out statistics that disagree with X, Alice will say "correlation != causation" or "lies, damned lies, and statistics" or "biased questions" or whatever.

This is all independent of the truth of X. X may well be true and well-understood to be so, but Alice may still accept weak arguments for it and discard arguments against it for dumb reasons. And this looks to me to happen *all the time*.

#10 ::: paul ::: (view all by) ::: October 05, 2012, 11:48 AM:

albatross @#9:

Often (but not always) it's possible to tell the difference between troll-marked arguing and trolling by the response to counter-arguments. Trolls' purpose is to "win" or to derail, rather than to get closer to an accurate picture of the world, so when confronted with respectful refutations (partial or complete, direct or with links) they'll either go away or disparage the response or switch arguments.

Of course, certain communities of discourse tend to produce lots of people who do this kind of stuff without intending to troll or thinking of themselves as trolls, but at some point that's no more interesting than the fact that some people think of themselves as "not racist" or "not sexist".

#11 ::: rea ::: (view all by) ::: October 05, 2012, 11:55 AM:

I see bad "correlation implies causation" arguments a lot more frequently than I see bad "correlation does not imply causation" arguments.

#12 ::: Dave Bell ::: (view all by) ::: October 05, 2012, 12:29 PM:

Correlation doesn't imply anything, but it allows an inference of causation to be made.

#13 ::: Serge Broom ::: (view all by) ::: October 05, 2012, 12:34 PM:

Cotillion does not imply saltation.

#14 ::: Terry Karney ::: (view all by) ::: October 05, 2012, 12:39 PM:

Paul how are you using Pascal's Wager?

#15 ::: Lars ::: (view all by) ::: October 05, 2012, 12:51 PM:

I’m not going to do your research for you.

But this is sometimes a legitimate response to a sort of muffling technique employed by trolls who want to, not argue a point, but waste the time of an opponent who actually knows what they're talking about. You see it all the time in clashes between climate-change denialists and, well, anyone who knows anything at all about climate science. The idea is to ensnare the knowledgeable person in a bottomless vortex of picayune, long-ago-answered quibbles, which lead to further quibbles, and so on, and so on.

#16 ::: JohnAspinall ::: (view all by) ::: October 05, 2012, 01:00 PM:

Here's my approach to a bad (e.g. trolling) "correlation does not imply causation" argument:

When A is correlated with B, there are (simplifying, a bit) 4 possibilities:
(1) A causes B;
(2) B causes A;
(3) Previously unmentioned third factor C causes both A and B;
(4) It's pure coincidence.

Then you ask Mr. Troll to pick one. It encourages the rest of the audience to explicitly formulate the other possibilities.

"So, Mr Troll, you're here to defend the notion that lung cancer causes people to smoke cigarettes?"

"So, Mr Troll, you're here to suggest the notion that an addictive personality not only causes people to smoke cigarettes, but also increases their likelihood of lung cancer?"

"So, Mr Troll, you're here to promote the 0.000...0001% possibility that all those lung cancer patients just happened to also be smokers?" (Large portions of the statistician's toolbox are devoted to estimating the likelihood of the coincidence in possibility (4).)

It engages with the trollery just enough so that Mr. Troll cannot claim to be being ignored, yet it says "put up or shut up".

#17 ::: lorax ::: (view all by) ::: October 05, 2012, 01:45 PM:

Craig @8, that was my first thought too, but the XKCD archive provides the original dates of strips if you hover over them, and that wasn't until March 2009.

#18 ::: Caroline ::: (view all by) ::: October 05, 2012, 01:47 PM:

A friend sent me the Engber article yesterday. The first thing that struck me: The NYT article about how depressed people use the internet was very careful not to make claims about causation. The most they did was speculate about underlying psychological reasons for certain internet behaviors that they saw were correlated with depression.

So I'm not sure what the commenters quoted by Engber were complaining about -- maybe some blogger or reporter who decided to frame the story as "The Internet makes people depressed!"

And in fact, there is no evidence in this study for that claim -- and the study authors aren't making that claim. What they're saying is, certain patterns of internet use are more likely to occur among people who are depressed. And therefore if you notice that someone is using the internet in those patterns, you might want to encourage them to get screened for depression, because they are more likely to have it than someone who doesn't use the internet in those ways. That's what correlation is.

You don't actually need to know the causation in order to use that correlation to take some action. The suggested action is "If you're checking your e-mail every two seconds, maybe go to a doctor and get screened for depression," not "Quit checking your e-mail and you won't be depressed anymore."

Hell, most of medicine is correlation. There's a lot we don't know about the exact mechanisms of causality in the body.

Even genotyping is often correlation. I research a particular heart condition, called Long QT Syndrome Type 1, that can cause dangerous cardiac arrhythmias (where different parts of your heart start beating all out of sync, which makes your heart not able to pump blood properly). You can test people for particular genetic mutations associated with LQT1. But 70% of the people known for a fact to have a LQT1 mutation won't have a cardiac arrhythmia*. They'll just go right on as though they don't have the mutation.

Causation is clearly more complicated than "mutation = arrhythmia". There are probably all sorts of interactions, epigenetic factors, etc. You get deep into molecular biology fast, trying to track down mechanisms. It's fascinating and well worth investigating.

But somebody with a LQT1 mutation does have a higher probability of having a cardiac arrhythmia than someone without a LQT1 mutation, and that information does materially change how their doctors will treat them. The correlation between the mutation and the arrhythmia does provide useful information.

Anyone who says "Correlation doesn't imply causation!" to mean "Correlation is completely meaningless in all circumstances! Only proof of the exact details of causation is worth anything!" -- doesn't understand science or statistics or the actual meaning of what they're saying. They should be handled accordingly, as Teresa suggests.

But it's just silly to draw grand conclusions about how people are overusing the catchphrase as some defense against powerful science or "Big Data" running wild. People are overusing the catchphrase because it's the only thing they remember from statistics class and it makes them feel like big smart guys on the Internet. That's all.

(In the same way, it's silly to draw grand conclusions about what 50 Shades of Grey means about "modern sexuality.")

*Priori et al. "Risk Stratification in the Long QT Syndrome." New England Journal of Medicine, 2003; 348:1866-1874.

#19 ::: paul ::: (view all by) ::: October 05, 2012, 01:49 PM:

Terry Karney:

I (like to) use Pascal's wager in a pretty straightforward way: looking at the gains and losses if different positions are right or wrong. E.g. on climate change: If the deniers are right and we act as if they're right, we gain oodles of self-esteem and about 3% of GDP, which is roughly equivalent to the amount that got cratered by the housing crash; if they're right and we act as if they're wrong, we lose about 3% of GPD that will be applied to carbon-reduction and other mitigation tools instead of 3D TVs, champagne and caviar. If the deniers are wrong and we act as if they're wrong, we spend 3% of GDP to gain a secure future; if they're wrong and we act as if they're right, our greatgrandchildren live under the same conditions as humans in the Younger Dryas.

as with John Aspinall @#16, it's in significant part a way to frame what the actual options might be.

#20 ::: James Moar ::: (view all by) ::: October 05, 2012, 01:58 PM:

"It’s a slippery slope"

Well, you've got to remember that if you use a slippery-slope argument once, you'll end up using it for absolutely everything.

#21 ::: Mary Aileen ::: (view all by) ::: October 05, 2012, 02:25 PM:

James Moar (20): That made me laugh.

#22 ::: Christopher Wright ::: (view all by) ::: October 05, 2012, 02:58 PM:

I hate that trolls have co-opted the "slippery slope" argument, because there are times when it's a valid and important argument to make. But I think on a practical level it's almost impossible to use it these days.

#23 ::: albatross ::: (view all by) ::: October 05, 2012, 03:15 PM:


That style of argument creates a great incentive for apocalyptic predictions of all kinds, as with what you did there, or with Condoleeza Rice's famous line "our first warning may be a mushroom cloud rising over an American city." It's related to the common scheme of con men where they offer you such a huge hypothetical payoff that your brain kind-of can't deal with assessing the expected benefit vs costs.

#24 ::: ACW ::: (view all by) ::: October 05, 2012, 03:23 PM:

JohnAspinall (#16), your taxonomy actually made me tear up a little, because I immediately thought, "You know who would be interested in this discussion? DLW. Oh, no, dammit."

(I'm assuming this is a particular JohnAspinall; if it's not, then this post will make no sense, until I explain that DLW is a recently-deceased mutual friend of mine and that John Aspinall's.)

#25 ::: cleek ::: (view all by) ::: October 05, 2012, 04:28 PM:

"Trolls' purpose is to "win" or to derail, rather than to get closer to an accurate picture of the world"

in my experience, a troll is more about derailing, name-calling, bomb-throwing, etc.. any response they can get is a good one, as long as they're driving the conversation (typically away from a topic they don't like).

winning for the sake of winning is the activity of the hyper-competitive jerk.

a person can be both, sure.

and then there's the pointless pedant...

#26 ::: Fragano Ledgister ::: (view all by) ::: October 05, 2012, 04:51 PM:

Paul #6:

I took Pascal's wager for true
But misfortune soon caused me to rue.
Sad indeed was my fate,
As I needs must relate,
One god? It turns out there are two.

#27 ::: paul ::: (view all by) ::: October 05, 2012, 05:48 PM:

albatross @#23:

Although the possibility of an apocalyptic arm does make pascal's wager into a slippery slope (ahem) what's sometimes useful about it is seeing which arm different people view as apocalyptic. (Even as I typed my example, I was dimly aware of the issues. On the other hand, the whole point of pascal's wager is that the result of being wrong in one direction is Really Bad while the point of being wrong in the other direction is Not So Bad.)

#28 ::: Lee ::: (view all by) ::: October 05, 2012, 06:11 PM:

albatross, #9: Troll arguments and flamer-bingo phrases share certain similarities. One of them is that a number of them can be used in good faith; context is critical. It's not the argument that makes the troll, it's the way it's deployed -- which is exactly the point that Engber seems to have missed in the article which inspired this post.

Lars, #15: You're talking about a fractally wrong argument. And yes, shutting that sort of thing down with "I'm not going to do your research for you" is perfectly appropriate; see my comment above.

James, #20: *snerk*

Fragano, #26: Outstanding, and delineates the flaw in Pascal's Wager -- namely, that it assumes a binary set of alternatives, which may not be the case.

#29 ::: Sancho ::: (view all by) ::: October 05, 2012, 07:34 PM:

Really enjoyed this post. It explicates a bunch of tricks and traps trolls and Dunning-Krugeroids use to prevent meaningful discussion.

The Stuff Liberals Think line of attack is gold. Common applications include "liberals have no objective morality based on religion, therefore they can't object to [insert atrocity]" and "liberals back gay marriage when two people love each other, so if [non-consenting pair] loves each other liberals can't deny them marriage".

I love the slippery slope argument because it requires the troll to describe a version of reality that's never existed, where people simply shrug and permit any sort of change without thought or objection because they had previously accepted a vaguely similar change after a long fight.

That's behind the claims that if we permit same sex marriage, then we must permit interspecies marriage, as though we can't evaluate them separately and make decisions on merit.

As for dealing with them, you generally can't if your goal is to make them admit error or accept a proposition that conflicts with their ideas. When I'm wound up by a troll with no end in sight, I often point out that the discussion is not for their benefit, but for others reading, who will see that the debate has consisted of one participant laying out verifiable facts, and the other countering with evasion and falsehoods, and the reader will reach their own conclusions about who is correct.

#30 ::: Serge Broom ::: (view all by) ::: October 05, 2012, 08:46 PM:

If two women are allowed to marry each other and have kids, they might raise a boy who'll grow up to like women. The horror. The HORR... Wait. Oh. Nevermind.

#31 ::: Lee ::: (view all by) ::: October 05, 2012, 11:06 PM:

Serge, #30: That's not as much of a joke as you might think. There are an awful lot of men who will swear up and down that they like women, when what they mean is that they like to fuck women*, but would never consider one to be a friend like their male buddies. A boy raised by two women who like and respect each other might very well grow up to consider women in general worthy of being liked and respected, and that's something a lot of the anti-gay crowd doesn't even want to think about.

* I got that formulation from someone here. I don't remember who it was, but THANK YOU -- it suddenly made a lot of things much clearer.

#32 ::: Rob Rusick ::: (view all by) ::: October 06, 2012, 06:38 AM:

Lee @31: I got that formulation from someone here.

I recall Xopher saying that; I don't recall in what thread.

#33 ::: rea ::: (view all by) ::: October 06, 2012, 07:54 AM:

Sancho @29--goodness, if we allow slippery slope arguments, before you know it, people will be accusing us of wanting to allow marriage with box turtles! :)

#34 ::: lorax ::: (view all by) ::: October 06, 2012, 10:21 AM:

Sancho @29:

Odd though it sounds, I find the "if we permit same-sex marriage, we must permit interspecies marriage" line of argument to be encouraging. For years same-sex marriage was the horrible thing that the slippery slope would lead to, and opponents felt that it was so self-evidently bad that no further argument was necessary. (I've heard politicians make the argument that allowing an LGBT group to 'adopt a highway' - pick up trash on the roadside, in exchange for which the name of the organization would appear on a sign - would slippery-slope to same-sex marriage.) So if opponents of marriage equality now feel that they won't win people over just by arguing that same-sex marriage in itself is bad, but need to resort to laughable slippery-slope arguments, I see that as real progress.

#35 ::: Kip W ::: (view all by) ::: October 06, 2012, 05:39 PM:

David Goldfarb @5: He's probably a Republican, and constitutionally unable to bring himself to mention the existence of W.

#37 ::: Fragano Ledgister ::: (view all by) ::: October 06, 2012, 08:06 PM:

Many moons ago, when I was a grad student and learning how to be a political scientist, I plugged two sets of information into a chi-square program to see if there was a relationship: rural population of Latin American and Caribbean countries, and literacy rates of said countries. Blow me down if I didn't find that there was a relationship. The higher the rural population, the lower the literacy rate.

Now the arrow of causation doesn't go in two directions. I can't hypothesise that low literacy rates cause high rural populations, but the likelihood is that there is (or was) a causative factor lurking in the fact of high rural population. That factor was not far to seek, in policies that kept rural populations poor, subservient to the lords of mine and soil, and, in many countries,* in bondage to the land.

*Debt peonage (huasipungo) has only fairly recently been outlawed in Ecuador.

#38 ::: Jordin ::: (view all by) ::: October 06, 2012, 10:41 PM:

Paul #19:

Nitpick: What you're describing is not quite Pascal's Wager. Pascal's Wager is specifically about the case where the probability matrix (specfically P(go to heaven|belief)) is unknowable, but the potential reward (or penalty) is infinite.

(One could argue that rendering the Earth uninhabitable is close enough to an infinite penalty for all practical purposes, but that's a separate debate.)

There are lots of problems with Pascal's wager, including the one Frangano so elegantly points out, but one I haven't seen is that if you really believe the reward is infinite, then *any* finite cost or effort is negligible in comparison, and you should rationally devote every waking moment to religion, or whatever you think increases your chances of going to heaven -- and that path leads to martyrdom or madness.

#39 ::: Chris Lawson ::: (view all by) ::: October 06, 2012, 11:56 PM:

In maths/logic "implies" means "can be logically shown to entail" rather than the more common-use meaning of the word. This is not usually appreciated in causation-vs-correlation arguments on the inter webs.

#40 ::: Josh Berkus ::: (view all by) ::: October 07, 2012, 02:58 AM:

Rea @33:

If we outlaw slippery slopes, only outlaws will have box turtles.

#41 ::: NelC ::: (view all by) ::: October 07, 2012, 08:43 AM:

Josh @40:

If we outlaw slippery slopes, only outlaws will marry box turtles.

#42 ::: albatross ::: (view all by) ::: October 07, 2012, 10:15 AM:


The funny thing is, they were right. Treating gays decently, not arresting them for being gay or involuntarily subjecting them to some sort of treatment to cure them, was indeed the first step toward letting gays marry, raise kids, and generally be part of the normal community.

Similar things are true of civil rights for blacks--stop making blacks sit in the back of the bus, allow them into every school and business and club, and someday, we'll end up with a black president. And womens' rights--if we start letting women work as doctors and lawyers and own businesses and sign contracts, why someday there will be women giving orders to men, women in very high places in government and industry.

Sometimes, the downhill direction on the slippery slope is actually where you want to go.

#43 ::: Fragano Ledgister ::: (view all by) ::: October 07, 2012, 11:01 AM:

Josh Berkus # 40/NelC #41: The solution is obvious -- outlaw box turtles.

#44 ::: Serge Broom ::: (view all by) ::: October 07, 2012, 12:34 PM:

albastross @ 42.. Good point. Then again it's a slippery slope only if one doesn't like what's at the other end.

#45 ::: Lee ::: (view all by) ::: October 07, 2012, 12:46 PM:

albatross, #42: I notice one thing in common about all your examples. Ultimately, they boil down to, "Someday we might have to start treating Those People* like Real People! The horror, the horror!"

Anything that makes people be treated like people rather than property or animals is an improvement.

* Adams: "They're people, and they're here. That's the only requirement I've ever heard of."
Rutledge: "They are here, but they are not people, sir. They are property."

#46 ::: Bruce Cohen (Speaker To Managers) ::: (view all by) ::: October 07, 2012, 03:10 PM:

Jordin @ 38:

Even if the reward for that cell of the matrix is infinite there's still a problem: there's another cell (in trying to propitiate some god in some particular way, you pissed off the real one with the wrong kind of ceremony/sacrifice/act of worship) that can have an infinite negative consequence.

#47 ::: Xopher HalfTongue ::: (view all by) ::: October 07, 2012, 06:11 PM:

I agree, albatross, but the slope has to be ond of real logical consequence. Your first slippery slope doesn't lead to marrying your dog, and your second one doesn't lead to the enslavement of white people.

Which you know. Just pointing out where the spurious divides from the real.

#48 ::: David Goldfarb ::: (view all by) ::: October 07, 2012, 06:33 PM:

My own answer to Pascal's wager is that I think the probability of the infinite reward is zero, or at least infinitesimally small; the expected payoff of that cell is then finite (and, in my opinion, small enough not to outweigh the costs involved).

#49 ::: Steve C. ::: (view all by) ::: October 07, 2012, 08:21 PM:

The main problem with the slippery slope is that it involves prediction, and people pretty much suck at that. If we are presented something that's already happened, we can see the events and forces that led up that. Prophesying the past is a piece of cake.

So I think the Slippery Slope argument is brought up so much because we see the path that led to historical events and think that we can predict as easily as we can post-dict. Not gonna happen. We just aren't that smart.

#50 ::: Steve C has been gnomed ::: (view all by) ::: October 07, 2012, 08:22 PM:

Oh, what may have been that word of power? I promise to use it only for good.....

#51 ::: Terry Karney ::: (view all by) ::: October 08, 2012, 01:17 AM:

I always thought Pascal got it backwards.

If there is a God, and I do my best, it will work out (because God, as posited, isn't supposed to be an asshole, at root)

So if there isn't a God, it's incumbent to make the world the best sort of place, because this is the only shot we get.

If we do that, then God will be pleased, and Hell will be avoided.

But that's independent of devotion to outward forms.

#52 ::: Serge Broom ::: (view all by) ::: October 08, 2012, 03:58 AM:

Lee @ 45...

"Fortunately there are not enough men of property in America to dictate policy."
- John Hancock

#53 ::: Serge Broom ::: (view all by) ::: October 08, 2012, 04:04 AM:

Besides the Slippery Slope, let's not forget the Law of Unintended Consequences.

#54 ::: Dave Harmon ::: (view all by) ::: October 08, 2012, 09:30 AM:

lorax #34: Odd though it sounds, I find the "if we permit same-sex marriage, we must permit interspecies marriage" line of argument to be encouraging.

And then too, perhaps someday we will be creating, or even meeting, new intelligent species....

#55 ::: Serge Broom ::: (view all by) ::: October 08, 2012, 10:56 AM:

Dave Harmon @ 54... Sarek kept marrying Earth Girls.

#56 ::: JohnAspinall ::: (view all by) ::: October 08, 2012, 11:25 AM:

ACW@24: No name collisions here, I'm that John. I miss DLW too.

Dave Harmon@54: I feel there is a nice Tiptree-esqe story lurking near the overlap of Pascal's Wager and creation of a new intelligent species.

#57 ::: albatross ::: (view all by) ::: October 08, 2012, 11:34 AM:


Yeah, but the point is, to many of the people bringing up those slippery slopes, the outcomes we've gotten to or are getting to are indeed things that sounded horrible to them. I mean, there's a fairly large set of people who find the idea of treating gays like normal people, with legal rights to marry and divorce and adopt and raise kids and all that, absolutely horrifying and offensive. And some subset of those people probably made the entirely correct slippery slope argument that a policy of quiet acceptance of gays (not raiding gay bars, repealing or ignoring antisodomy laws, not bothering to harass long-term gay and lesbian couples who were being reasonably discrete about it) would lead, eventually, to the world we will have in another decade or two. Their slippery slope argument was right, and for them, it was also a bad place to end up.

Similarly, at the beginning of the civil rights movement for blacks, some people must have correctly pointed out that, if blacks were no longer kept from voting, intimidated into second class citizenship, etc., that eventually, they would be marrying whites, being promoted to powerful positions, being elected to positions of power, and over time, many white men would have to take orders from black men. That was a correct prediction. And for the people worrying about it and employing the slippery-slope argument, they were right--all those things happened, and we ended up where we are now, with many generals, CEOs, cabinet secretaries, police chiefs, mayors, and the president of the United States black. Their prediction of gloom and doom came true. We see it as a good outcome, but by their values, it's a disaster.

#58 ::: Heather Rose Jones ::: (view all by) ::: October 08, 2012, 11:37 AM:

Lars @ 15

re: I’m not going to do your research for you, and its valid uses

(Sorry if someone has covered this point already -- I looked and didn't spot any.)

The trollish use of this phrase, in my experience, generally follows this template:

Troll: "Claim based on wildly improbable factoid that no one else has heard of."

Respondent: "Factoid cannot be responded to unless contextualized. Could you cite your source?"

Troll: "I'm not going to do your research for you. And if you can't find it to contradict it, I must be right. Ha, ha, I win."

(Some of my most memorable encounters with this technique were in the usenet group alt.legend.king-arthur. Talk about a topic where highly improbable uncontextualized factoids reign!)

#59 ::: Heather Rose Jones has been gnomed ::: (view all by) ::: October 08, 2012, 11:38 AM:

Perhaps some home-made split-pea soup on this blustery fall day?

#60 ::: albatross ::: (view all by) ::: October 08, 2012, 11:39 AM:


Your comment made me think of a poem I really like, Davis Matlock by Edgar Lee Masters. The end of the poem goes like this:

Well, I say to live it out like a god
Sure of immortal life, though you are in doubt,
Is the way to live it.
If that doesn't make God proud of you,
Then God is nothing but gravitation,
Or sleep is the golden goal.

#61 ::: Dave Luckett ::: (view all by) ::: October 08, 2012, 12:30 PM:

Slippery slopes. One of those can be deployed against capital punishment, and I think that it's valid there.

That murder in Melbourne. There's no doubt that the bloke they arrested did it. There's no real doubt that he's a deadly danger to women - he'd already served eight years of a ten year sentence for a particularly nasty rape, and this was just a bit more violent, which fits the profile. There's no way to know whether he can ever be reformed and rehabilitated. Nobody could ever trust any such assessment.

Basically, he's going to go to jail until he dies. Even then, he's going to be a danger. He could escape. He could kill in jail. Nobody knows.

I can surely say I sympathise with those who think he should die. I do. I think it myself. But the problem is, it's not that he'll die. It's that we'll have to kill him.

And then the slippery slope argument: if we were to bring back the death penalty - which isn't going to happen - but if we did, the slope would be right there, steep, glistening, icy. There'd be no barrier at the top of it, and the degrees of the drop are imperceptible, at first. Until one fine day you find yourself hanging an eighteen year old for a robbery gone wrong.

No. Can't have it. Won't have it. Slippery slope.

Works for me.

#62 ::: Terry Karney ::: (view all by) ::: October 08, 2012, 12:33 PM:

albatross: Your, "they were right, and awful things happened" suffers from accepting that it truly is awful.

But that moves into questions of ethics.

#63 ::: Lee ::: (view all by) ::: October 08, 2012, 03:10 PM:

albatross, #57: That they were right according to their own lights does not negate the fact that their desired outcome is barbaric and unconscionable.

The next argument usually deployed in those cases is the fake-tolerance one: "If you liberals are supposed to be so tolerant, then you have to let me have my way too!" Which is dead wrong; there is no moral obligation to tolerate people who refuse to grant you the same courtesy.

#64 ::: Jacque ::: (view all by) ::: October 08, 2012, 05:27 PM:

Ah, yes.

In my current RL Troll Bingo game, my pet troll (let's call him Alphonse), kicked Bartholomew off of the committe, because B was "telling lies and half-truths."

When I asked A to specify exactly which things B had said were lies, and what was the truth, A came back with, "Well, okay, B can be on the committee. But he has to mind his manners!"


#65 ::: Jordin ::: (view all by) ::: October 08, 2012, 05:53 PM:

Serge @ 53: I thought the thing that was beside the Slippery Slope was the Path of Least Resistance.

#66 ::: Carol Kimball ::: (view all by) ::: October 09, 2012, 12:10 AM:

...and both of them are paved with good intentions.

#67 ::: albatross ::: (view all by) ::: October 09, 2012, 10:40 AM:

Terry and Lee:

Someone who predicted, on slippery-slope grounds, that ending the legal and extralegal persecution of gays would lead to gay marriage and adoption, was *right*. The things they predicted, in this case, actually happened or are in the process of happening now.

Someone who predicted, on slippery slope grounds, that giving women the vote and the right to sign contracts and own property would eventually end up with women being legally equal to men, and often being in positions of great power, was *right*. The stuff they said would happen actually happened.

The question of whether or not the predictions were right has *nothing whatsoever* to do with whether the predictors' values were good ones. Nor with whether I like their values.

#68 ::: Serge Broom ::: (view all by) ::: October 09, 2012, 10:45 AM:

Jordin @ 65... I thought the thing that was beside the Slippery Slope was the Path of Least Resistance.

There is also the Road to Hell, and its well-intentionned pavement.

#69 ::: albatross ::: (view all by) ::: October 09, 2012, 11:01 AM:


I think that turns on the definition of tolerance.

IMO, we never have an obligation to believe someone is right or pretend we do. It can be impolite and counterproductive to constantly point out how wrongity-wrong-wrong some other person's beliefs are, if you want to live in relative peace with them, but there's no *obligation* to refrain from pointing it out. That's true whether you're Richard Dawkins pointing out how wrongity-wrong-wrong religious people are, or Pat Robertson pointing out how wrongity-wrong-wrong atheists are, or whatever.

On the other hand, IMO, we always have an obligation to refrain from coercing people into changing their expressed beliefs. We can and should use violence and the threat of violence as needed to keep the peace, but not to demand that bad or evil ideas be silenced. Refraining from violence and coercion, even wrt people we can't stand, is one flavor of tolerance. This is the kind of tolerance that puts a cordon of police around the KKK march, to keep the Klansmen from being mobbed. "I despise you, but I'll remain at peace with you."

Peaceful coexistence with people whose ideas we think are largely screwy is possible mainly when we agree to disagree--which is to say, we agree not to spend too much of our interactions telling them the many ways in which they are being idiots for believing the obviously silly stuff they believe. That's a different kind of tolerance--a decision to keep some level of politeness and kindness in interactions with people whose beliefs seem wrong to us, allowing us to be friends with people of different religious and political and social beliefs. "I can accept that you have weird beliefs in some areas and still be your friend, and we'll just avoid the subject."

Still another flavor of tolerance is to accept that some people are different from us, but that all paths are more-or-less the same. For example, a Baptist might think that his church speaks more to him than a Lutheran church does, but still think that he and a Lutheran are on the same basic path, on the same basic side. He may not feel any great need to talk the Lutheran over to the Baptist side. That's a third flavor of tolerance, like "You and I are different, but there's nothing wrong with your kind of different."

We owe everyone the first kind of tolerance. The second and third kinds make interactions easier--they make it nicer to work together or be neighbors or be friends despite important differences. But we don't owe them to anyone, and nobody owes them to us.

#70 ::: Bill Higgins-- Beam Jockey ::: (view all by) ::: October 09, 2012, 11:46 AM:

Albatross writes in #42:

And womens' rights--if we start letting women work as doctors and lawyers and own businesses and sign contracts, why someday there will be women giving orders to men, women in very high places in government and industry.

Here's an obscure, but possibly interesting, example of women in "very high" places in government and industry.

#71 ::: Lee ::: (view all by) ::: October 09, 2012, 12:49 PM:

albatross, #67: You haven't grasped the point that Terry and I are making, which is that whether or not they were right does not matter.

and @69: Again, you're missing my point. If someone is actively calling for my civil rights to be removed or curtailed, I am under no obligation whatsoever to "tolerate" them. And I extend that to anyone who is calling for the civil rights of anybody else to be removed or curtailed. If I am required to allow that behavior in the name of "tolerance", it puts all the power in their hands and leaves me with no recourse.

People can think whatever they want. It's what they DO that matters.

#72 ::: lorax ::: (view all by) ::: October 09, 2012, 02:27 PM:

albatross @69:

The problem with the second type of 'tolerance' as you outline it is that the consequences of 'agreeing to disagree' are often very asymmetric.

Let's stick with the example of same-sex marriage for the moment. I have relatives who disagree with me on this issue. Let me be clear that
it's not what they say at Thanksgiving that upsets me; it's what they do in the ballot box. For them, it's not a terribly important issue; they'll vote against marriage equality if it comes on the ballot, or count it as a point in a candidate's favor if he campaigns on rolling back domestic partner benefits, but they don't really spend a lot of time thinking about the issue; it's easy for them to talk about 'agreeing to disagree' because it doesn't really matter to them, and they're as happy to not discuss the issue in the name of family harmony as they are to not discuss any other political issue.

For me, though, it matters a great deal. It's the difference between being on my wife's health insurance and not (she works for the federal government; it would quite literally require an act of Congress for me to be covered by her insurance), the difference between being able to jointly adopt a child and having a kid with only one legal parent, the difference between being able to inherit the house if anything happens to her and needing to pay inheritance taxes on it. I cannot afford to smile and treat my civil rights as a minor, academic issue, and expecting me to do so as an ordinary matter of courtesy is ignoring the very real consequences of many of these issues.

#73 ::: Xopher HalfTongue ::: (view all by) ::: October 09, 2012, 04:06 PM:

In fact, issues of civil rights always have that property: in general, the dominant group in society can afford to just not discuss them for the sake of peace, or politeness, or whatever.

I resent them being given credit for politeness because they just won't discuss issues that could be quite literally life or death to ME.

#74 ::: albatross ::: (view all by) ::: October 11, 2012, 02:53 PM:

Xopher and lorax:

Tolerating other peoples' views doesn't mean never letting them know you disagree with them, it means being willing to let the issue lie, and still interacting with them in positive ways in other areas. It means accepting that people can be decent and valuable human beings even when they disagree with you on really important stuff. It also means accepting that they have the same right to disagree as you do.

We may live to see a world where gay rights becomes relatively uncontroversial. I hope so. But we certainly won't see a world where everyone agrees, on this or any number of other moral issues, or political issues with big and important impacts on real human beings. (Pretty much all major political issues matter deeply to some real human beings; most will have a big impact on some peoples' lives one way or another.) So our choices are either to refuse to associate with people who disagree with us on some issues, or to agree to disagree on those issues.

That is, in fact, what we do here, right? We don't rehash brtn or gn cntrl regularly, because those are issues that split the community, and that have led in the past to unproductive flamewars. We mostly leave those issues alone. That feels to me like one flavor of tolerance.

#75 ::: albatross ::: (view all by) ::: October 11, 2012, 03:06 PM:

It seems to me that looking at arguments in terms of slippery slope or correlation!=causation runs into the same sort of problem as journalists doing "balanced" reporting by reporting what the Republican and Democratic mouthpieces had to say about the issue, and then considering their job done. What's needed is not an algorithm for accepting or tossing out arguments, it's judgment.

And this kind of one-line argument or dismissal of a claim is a flaming warning sign for a particular kind of short circuit in peoples' thinking: I start out knowing what answer I want to get, and then I'm willing to accept even very weak arguments that let me stop having to think about the question. Thus, when I don't want to believe that there's a relationship between X and Y, I can dismiss statistical evidence of that relationship with a handwave and a "correlation isn't causation," and maybe a throwaway reference to the positive correlation between ice cream sales and murder rates. Similarly, I can dismiss a cite to a media story claiming X by a reference to media bias (which kind depends on what side I'm arguing). Or I can dismiss the source of the information as biased or racist or anti-American or anti-Semetic or whatever other label is useful.

And the tricky part is, any of those may be true. Sometimes, official statistics are cooked to make things look better, media reports are hammered into a predetermined narrative, an advocate cites only the 1/20 studies that found a positive correlation with p>.05, etc. But it's really easy to use those sorts of claims to let yourself stop thinking about something when you already know what answer you intend to get.

#76 ::: Tom Whitmore ::: (view all by) ::: October 11, 2012, 04:25 PM:

albatross @75: going back to the original post, though, what T is pointing out is that objecting to people pointing out that correlation is not causation can be used in exactly the way you see people using the phrase itself. That is, objecting to the use of the phrase allows people to dismiss the user of the phrase, no matter how intelligently that person does so.

Because the people using it that the objector was pointing to: not doing so in a simply dismissive manner. T's much more on your side than this response leads me to think you know.

#77 ::: Tracy Lunquist ::: (view all by) ::: October 11, 2012, 10:46 PM:

albatross @74: The sense I got from Xopher and lorax was that the disagreement was not the issue, but rather the specific consequences of that disagreement for them personally.

I'd be astonished if any one of us could say that every single member of our families has congruent political beliefs, and I've never encountered a blood family that didn't have certain hot-button issues they tacitly (or explicitly) agreed never to discuss over dinner. But I can certainly sympathize with it being very hurtful for a family member to cast an opposing ballot on an issue that directly affects your life. If I were gay, and my family voted to deny me some civil right (marriage, property rights, etc.), that is them saying that they believe that I am not worthy of the same rights as they are -- even though I am their flesh and blood. It's precisely because the issue in that case is NOT academic that it would be very hard to just "agree to disagree."

Sure, there are always going to be people who are offended by homosexuality, just like there are always going to be people who are offended by women or offended by people of color. But we have, as a society, come around to accept that women and people of color are full citizens of the United States and entitled to the same rights as white males. For some nutty reason, it's still OK to send LGBTQ people to the back of the bus. I don't blame them for being grumpy about that.

#78 ::: Tracy Lunquist ::: (view all by) ::: October 11, 2012, 10:47 PM:

(With apologies for the fact that my comment is probably unnecessarily far off the track of the original post.)

#79 ::: Bill Stewart ::: (view all by) ::: October 12, 2012, 03:05 AM:

Steve C.@49 "Prophesying the past is a piece of cake."

Nah - hindsight's about 20/40, at best. Hard to get the past half right even if you were there, and harder when you weren't.

#80 ::: Jennifer Baughman ::: (view all by) ::: October 12, 2012, 07:11 PM:

Lee, albatross, lorax, Xopher, Tracy Lundquist on tolerance:

It occurs to me that the biggest problem of the "let it lie" definition of tolerance is in classification of beliefs. albatross #69 refers to beliefs that could be worked around, or "let lie", as "largely screwy". To me, "largely screwy" suggests "Uncle Amos swears he was abducted by aliens" or "Grandma has odd ideas about what children like to wear, but we only see her once a year/she doesn't have many years left in her, and she'll never know you took the sweater to Goodwill."

"Largely Screwy" does not include falsehoods ("Obama is really a Kenyan Muslim of illegitimate descent", "Vaccines cause autism"), or beliefs which, when put into practice, are actively harmful ("Gays are unnatural", "Illegal immigrants are putting citizens out of jobs"). And the people affected by those who act on harmful beliefs have the right not to let them slide; in fact, speaking up is sometimes their only recourse.

I don't think it's coercive to call out falsehoods and harmful beliefs, even if it has to be done multiple times, in multiple ways. Nor do I believe it's coercive to tell someone that, if they want to consider interacting with you, they need to observe the minimum amount of courtesy involved in not saying hurtful things. That's not coercion, that's consequence.

#81 ::: Kip W ::: (view all by) ::: October 13, 2012, 04:58 PM:

Jennifer, I'm reminded somehow of a quote in The Shorter Bartlett's Familiar Quotations, first quoted by Zephaniah W. Pease and said to be said by the mate of a whaler to his ill-humored captain: "All I want of you is a little see-vility, and that of the commonest goddamnedest kind."

It seems to me that this is the essence of what Republicans love to caricature as "Political Correctness run amuck." Just politeness. Civility. Of the commonest kind.

Is it the worst it's ever been? I couldn't say, but I'm quite certain its worse than bad enough.

#82 ::: Avram ::: (view all by) ::: October 13, 2012, 08:31 PM:

Albatross @67, thing is, equality is a virtue, and a more important one than tolerance.

Assuming tolerance is even a virtue! Tom Paine didn't think it was one: "Toleration is not the opposite of intoleration, but is the counterfeit of it. Both are despotisms. The one assumes to itself the right of withholding liberty of conscience, and the other of granting it." (The Rights of Man) He was talking about religious liberty, and contrasting the English Acts of Toleration (which reduced the extent to which non-Anglicans were oppressed, while leaving Anglicanism as England's Established Church, and Anglicans with special privileges) with the French Constitution's Universal Right of Conscience, and America's anti-establishment principle. But we can see the parallel with our own established gender model (heterosexual, cis-gender) having legal privileges over other models.

Tolerance is only a desirable state when contrasted with intolerance. Compared to equality, it's a step backwards.

What's going on with same-sex marriage (and what happened with the civil rights struggle decades past) is that the group that seeks to perpetuate established privilege, the state under which the established group has the option of tolerating or not-tolerating the dissenting group, is recasting their desire as something to be tolerated by an establishment. They are saying, ultimately, that their desire to keep their boot on another person's neck should be seen in the same light as the other person's right to have the boot off their neck. They're saying we need to tolerate inequality.

#83 ::: Jules ::: (view all by) ::: October 14, 2012, 12:03 AM:

Caroline @18 -- there have been rather a lot of media reports of the "internet use causes depression" type recently, however most seem to be linked to other studies (e.g. this one in the Daily Mail).

#84 ::: Nancy Lebovitz ::: (view all by) ::: October 14, 2012, 10:04 AM:

Overview of the poor quality of a high proportion of scientific papers

This is mostly about publication bias, lack of replication, and fraud. People are working on the problems, but the current situation is pretty horrifying.

It's all very well to say that "correlation doesn't equal causation", but there's some reason to think that the correlation you've heard about wasn't even there in the first place.

#85 ::: Nancy Lebovitz ::: (view all by) ::: October 14, 2012, 10:06 AM:

Gnomed again.

Chicken with experimental yogurt sauce (reasonably successful-- yogurt, lime juice, many spices, chopped tomatoes) offered as a bribe.

#86 ::: Shane ::: (view all by) ::: October 14, 2012, 08:16 PM:

Nancy Lebovitz #85 - peer reviewed experimental yoghurt sauce?

#87 ::: Tom Whitmore ::: (view all by) ::: October 14, 2012, 10:04 PM:

Personally reviewed -- what's more peer-ish than that?

#88 ::: Benjamin Wolfe sees spam ::: (view all by) ::: February 26, 2014, 04:02 PM:

While pounding the general idea into people's heads is good, the previous post certainly smells like spam.

Post should be rejected by peer review.

#89 ::: Cally Soukup spots more old SPAM ::: (view all by) ::: March 03, 2014, 01:02 AM:

Compliment spam at 88.

Welcome to Making Light's comment section. The moderators are Avram Grumer, Teresa & Patrick Nielsen Hayden, and Abi Sutherland. Abi is the moderator most frequently onsite. She's also the kindest. Teresa is the theoretician. Are you feeling lucky?

Comments containing more than seven URLs will be held for approval. If you want to comment on a thread that's been closed, please post to the most recent "Open Thread" discussion.

You can subscribe (via RSS) to this particular comment thread. (If this option is baffling, here's a quick introduction.)

Post a comment.
(Real e-mail addresses and URLs only, please.)

HTML Tags:
<strong>Strong</strong> = Strong
<em>Emphasized</em> = Emphasized
<a href="">Linked text</a> = Linked text

Spelling reference:
Tolkien. Minuscule. Gandhi. Millennium. Delany. Embarrassment. Publishers Weekly. Occurrence. Asimov. Weird. Connoisseur. Accommodate. Hierarchy. Deity. Etiquette. Pharaoh. Teresa. Its. Macdonald. Nielsen Hayden. It's. Fluorosphere. Barack. More here.

(You must preview before posting.)

Dire legal notice
Making Light copyright 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2011, 2012, 2013, 2014, 2015, 2016, 2017 by Patrick & Teresa Nielsen Hayden. All rights reserved.