Join 3,552 readers in helping fund MetaFilter (Hide)


You know; Arsenic is "Natural..."
June 16, 2014 5:18 PM   Subscribe

10 Scientific Ideas That Scientists Wish You Would Stop Misusing.
posted by quin (111 comments total) 50 users marked this as a favorite

 
Good article. io9 continues to be the most non-awful part of the Gawker Media empire.

"It doesn't help that those little tubes of plastic toy dinosaurs often include cave people or mammoths."

I still blame Hanna-Barbera for most of the common 'pre-historic' misconceptions... not to mention the obsession with flying cars.
posted by oneswellfoop at 5:27 PM on June 16 [3 favorites]


Theory vs Hypothesis is definitely up there for me. I say the former all the time when I mean the latter, and am trying to switch.
posted by Phredward at 5:38 PM on June 16 [1 favorite]


It wasn't at all clear to me if the i09 writers were making a distinction between colloquialisms versus malapropisms (or, in the case of the ridiculously dumb What The Bleep Do We Know) outright misrepresentations.

Most of these terms are colloquialisms that are used for rhetorical effect (eg, "survival of the fittest") in some sort of informal armchair argument (for example, any Op-Ed column in any newspaper) that is intended to win points more than anything else.
posted by KokuRyu at 5:40 PM on June 16 [1 favorite]


10. How very restrained.
posted by edd at 5:50 PM on June 16 [3 favorites]


The solution is to make science more like a TED talk
posted by thelonius at 5:54 PM on June 16 [6 favorites]


Oh sweet, 2/10 are from Jacquelyn Gill! She's one of the best ecology bloggers out there. The "all traits must be adaptive" assumption drives me bonkers and I'm so glad she called it out. She's also super right about peoples' utter lack of perspective on the geologic timescale; the first draft of my masters project included a bunch of stuff about the Last Glacial Maximum, and people just don't even know where to start with that. Their eyes would glaze over before I could even say what I was doing.

The "statistically significant" one is the best of the lot, though. Maybe I'll just start saying statistically discernible and see if it catches on. When you frame it that way, it becomes clear that you also need to care about effect size - sure, you can tell it's different, but does it matter?
posted by dialetheia at 6:00 PM on June 16 [15 favorites]


I feel like "correlation does not imply causation" should be on there.
posted by Sangermaine at 6:00 PM on June 16 [7 favorites]


Also note that 7 is from MeFi's own escabeche. I've never heard "statistically discernable" suggested as an alternative for "statistically significant" before but it is much better.
posted by grouse at 6:04 PM on June 16 [12 favorites]


I wonder if I can convince my committee to let me start writing "statistically discernible" in all my papers. It's so much better.
posted by pemberkins at 6:06 PM on June 16 [5 favorites]


Happy to see that "organic" got a mention, but disappointed that they failed to remark on the fact that the term has nearly the opposite meaning in chemistry. Plastics and other fun things made from petroleum distillates (carbon chains)? Yeah, that's organic chemistry!

I'm not sure how I feel about quantum weirdness. The Uncertainty Principle gives us limits on what we can measure. So if you want to construct a supernatural theory (which, being supernatural, is going to be beyond the reach of science and thus non-falsifiable. unless you screw up and try to place your supernatural theory within the domain of science, in which case you have only yourself to blame), then doing so beneath the planck length is probably as good a place as any. The bigger problem is when these theories are framed such that quantum mechanics seems to validate these supernatural theories. It doesn't - it just explains the data that we get when we attempt to observe really tiny things. QM doesn't tell us how things are it just provides a general description of our observations so far.
posted by b1tr0t at 6:08 PM on June 16 [3 favorites]


Not a scientific term per se, but one that makes me cringe is "processed", as in food, as in no good, very bad stuff.

Chopping, mixing, cooking, freezing: just a few processes I can think of off the top of my head.

If that's not what is meant by the word "processed", then the word has no meaning.
posted by Pararrayos at 6:10 PM on June 16 [11 favorites]


Speaking as a Bayesian, even 'statistically discernible' is wrong.
posted by edd at 6:10 PM on June 16 [4 favorites]


I don't know what the Last Global Maximum is but if you were at this bar I would buy you a drink.right now because I need to know about this.
posted by sio42 at 6:19 PM on June 16 [3 favorites]


"That is cool, of course, but only in the sense that all of physics is cool" made me giggle.
posted by jaguar at 6:21 PM on June 16 [4 favorites]


Please do not overlook the exponential growth in misuse of the word "exponential".
posted by Nelson at 6:22 PM on June 16 [7 favorites]


Honestly, I've yet to find a scientist or engineer that could strongly commit to the behavior of gravity, except as a localized case with possible exceptions.

As such, "Scientists agree that..." is my least favorite pop science expression. It's always more complicated than that.
posted by underflow at 6:24 PM on June 16 [6 favorites]


'Natural' kills me sometimes. I follow a couple of skincare blogs, and while most of them are science-based/run by cosmetic chemists, I still occasionally run into something like DIY sunscreen, along with the claim that because it is 'natural' (i.e. full of stuff you don't have to google) it is automatically better than the stuff you can buy in a store.

I understand the desire to know exactly what goes on/in one's body, but there are certain expectations attached to the word in some circles that I think can sometimes do more harm than good.
posted by supermassive at 6:27 PM on June 16 [1 favorite]


Speaking as a Bayesian, even 'statistically discernible' is wrong.

Care to speak as a Bayesian about what's right?
posted by clockzero at 6:28 PM on June 16 [2 favorites]


I feel like "correlation does not imply causation" should be on there.

Yeah, I think the prevalence of this gem has caused some people actually believe that correlations are meaningless and shouldn't even be studied.

A correlation between A and B means that A causes B, B causes A, a separate factor causes A and B, or there was some sort of flaw in the study that identified the correlation. Next time, before dropping the "correlation does not imply causation" truth-bomb, maybe it'd be better to discuss which of those three alternatives you're presenting as more likely.
posted by zixyer at 6:29 PM on June 16 [13 favorites]


Is there a Bayesian interpretation of significance? I thought significance was pretty much relegated to the frequentist world anyway.
n.b. I am totally Bayes-ignorant although I try to be aware of its arguments re: the limitations of frequentism

I feel like "correlation does not imply causation" should be on there.

If anything, I'd like to see them call this phrase itself out. I see non-scientists egregiously misapplying it constantly. No, correlation doesn't imply causation, but you can't just say that to dismiss every scientific paper you read. In a properly designed study, multiple correlations provide the basis for a (hopefully) convincing argument for causation.

I don't know what the Last Glacial Maximum is but if you were at this bar I would buy you a drink.right now because I need to know about this.

Yay! It marks the point at which the last glacial ice sheets were at their fullest extent, which happened roughly between 20,000 and 26,500 years ago, depending on where you're measuring and how.
posted by dialetheia at 6:31 PM on June 16 [7 favorites]


this is why scientists are so popular
posted by Reasonably Everything Happens at 6:35 PM on June 16 [9 favorites]


I like this, but at the same time, it is useful to have straightforward terms for things like foods grown without pesticides or petrochemical fertilizers, as long as the words have an agreed upon meaning in that context. There are lots of words that have multiple meanings, so if organic meant something specific related to food, it could still also have the its chemistry meaning. The problem is that organic as a food term is not necessarily defined consistently, not that the word can also have a different meaning in a different context.
posted by snofoam at 6:47 PM on June 16 [7 favorites]


Honestly, I've yet to find a scientist or engineer that could strongly commit to the behavior of gravity, except as a localized case with possible exceptions.

Physicists might argue about what causes gravity, but I think most scientists and engineers are comfortable with the behavior. With Newton's laws we can calculate the effects of gravity over pretty substantial time and length scales to a high degree of accuracy.
posted by anifinder at 6:47 PM on June 16 [1 favorite]


The misuse of Occam's Razor wasn't on the list. I'm sure there's a simple explanation for that.
posted by klarck at 7:01 PM on June 16 [33 favorites]


I would have gotten all shirty about "stop misusing the word prove" except it was just weirdly titled -- the argument is a more reasonable "please accept that proof means something different in science than in everyday life (also theory)".
posted by jeather at 7:02 PM on June 16 [2 favorites]


How about "evolving"? Like I read this crappy news article a few years ago talking about how human thumbs are evolving because we are using them so much to text.

lamarck much?
posted by hal_c_on at 7:07 PM on June 16 [8 favorites]


I feel like "correlation does not imply causation" should be on there.

If anything, I'd like to see them call this phrase itself out. I see non-scientists egregiously misapplying it constantly. No, correlation doesn't imply causation, but you can't just say that to dismiss every scientific paper you read. In a properly designed study, multiple correlations provide the basis for a (hopefully) convincing argument for causation.


I heard this reasoning in a criminal case once.

The dude was pretty much found with 'the smoking gun' in his location, and the lawyer was all "correlation does not mean causation: just because his location can be correlated with the 'smoking gun', it doesnt mean he caused 'the crime'.

And I was like wow. Law is the anti-science.
posted by hal_c_on at 7:10 PM on June 16 [1 favorite]


I'll admit it: I use "theory" wrong all the time (when the more precise "hypothesis" is what I really mean). But it's not out of ignorance, but habit bred from pragmatism. Most listeners I encounter in daily life--even the scientifically-literate ones--would know what I mean if I say "theory," while many honestly wouldn't know what I'm saying or would just think I'm trying to show off my vocabulary if I said I had a "hypothesis."
posted by saulgoodman at 7:13 PM on June 16 [1 favorite]


The lawyer is right? The possesion of the murder weapon is not usually sufficient to prove someone guilty beyond a reasonable doubt.
posted by RustyBrooks at 7:14 PM on June 16 [1 favorite]


That's an abuse of "correlation", as well. One observed set of paired data cannot a correlation make.
posted by gingerest at 7:14 PM on June 16 [5 favorites]


Because science education is really weak. I barely got any in my college career.
posted by saulgoodman at 7:16 PM on June 16 [1 favorite]


I agree it's a bad place to use the phrase "correlation doesn't not prove causation" but it is still true that possesion of the gun does not prove that the guy killed someone with it.
posted by RustyBrooks at 7:16 PM on June 16 [3 favorites]


Next time I eat convenience store nachos with fake cheese I will take comfort knowing I am eating 'organic'. Heh.
posted by mazola at 7:43 PM on June 16 [6 favorites]


I use "theory" wrong all the time (when the more precise "hypothesis" is what I really mean)

"I've got a hypothesis
it could be bunnies"

... nope.
posted by tzikeh at 7:47 PM on June 16 [13 favorites]


The misuse of Occam's Razor wasn't on the list. I'm sure there's a simple explanation for that.

What people don't understand about Occam's Razor is that it's a moral principle, not a logical principle. You can't apply it to other people's arguments. You apply it to your own arguments, because it's the right thing to do.
posted by grog at 8:09 PM on June 16 [4 favorites]


A correlation between A and B means that A causes B, B causes A, a separate factor causes A and B, or there was some sort of flaw in the study that identified the correlation.

Or you were just really unlucky.
posted by BungaDunga at 8:11 PM on June 16 [5 favorites]


I'll admit it: I use "theory" wrong all the time (when the more precise "hypothesis" is what I really mean). But it's not out of ignorance, but habit bred from pragmatism. Most listeners I encounter in daily life--even the scientifically-literate ones--would know what I mean if I say "theory," while many honestly wouldn't know what I'm saying or would just think I'm trying to show off my vocabulary if I said I had a "hypothesis."

I think it's cool to use "theory" as in "I have a theory about why all my socks are going missing" during a casual conversation. The issue is when folks take that casual conversation definition and assume it applies in scientific contexts, e.g. "Evolution is just a theory, man!"
posted by Gymnopedist at 9:01 PM on June 16 [7 favorites]


What people don't understand about Occam's Razor is that it's a moral principle, not a logical principle.

How so? Isn't it well-suited to describing a fruitful way to think about checking for parsimony?

You can't apply it to other people's arguments. You apply it to your own arguments, because it's the right thing to do.

You can't apply it to others' arguments? Is that...illegal?
posted by clockzero at 9:15 PM on June 16 [1 favorite]


It ain't ignorance causes so much trouble; it's folks knowing so much that ain't so.

I've really tried to stop myself lately from talking over my pay grade, as it were. If everybody did it, I feel like we might evolve a little faster. Yay me.
posted by Camofrog at 9:19 PM on June 16 [3 favorites]


I agree it's a bad place to use the phrase "correlation doesn't not prove causation" but it is still true that possesion of the gun does not prove that the guy killed someone with it.

Oh, I agree that being near a smoking gun does not mean guilt. But that's what sucked; the reasoning was correct, but the words used to come to that conclusion were wrong. But because the reasoning was correct, it seemed as if the causation correlation concept was applied correctly.

It wasn't.
posted by hal_c_on at 9:28 PM on June 16


This "correlation doesn't not prove causation" thing has never been proven. I think it's just a theory.

Albeit a theory that often agrees with evidence. But that's just correlation.
posted by twoleftfeet at 9:47 PM on June 16 [5 favorites]


Reasonably Everything Happens - "this is why scientists are so popular"

Naw, this is a better example of why scientists get pissed off at people with "American" style elementary and secondary educations that actively forstall curiosity and inquisitiveness. It's almost like an arms race against the Chinese to make an even more sheepish and compliant populace.

Its the secondary effect that makes us Scientists oh so popular with the "in" crowd. You're taught to fear/dismiss us, they're taught to not be like us. Question authority, but you're going to have to be "smarter" than authority in order to win.
posted by porpoise at 9:57 PM on June 16 [1 favorite]


What people don't understand about Occam's Razor is that it's a moral principle, not a logical principle. You can't apply it to other people's arguments. You apply it to your own arguments, because it's the right thing to do.

I think you must be thinking of grog's Razor.
posted by Segundus at 10:26 PM on June 16 [1 favorite]


My favorite "suddenly realizing you're the nerdiest person in the room full of nerds" moment came about during some tedious case presentation and then the pharm rep started having it out with some fighty, sleep-deprived intern, who said "But correlation doesn't imply causation!" with an edge of hysteria to his voice.

Out of sheer reflexive habit I woke up enough to dredge some long submerged and dimly recalled phrase from a statistics textbook, "Wait...You can't possibly be implying causation is not a statistical subset of correlation?"

Then I realized to my horror I was defending the pharm rep by mistake, and took some danish to finish my nap in the car.

But yeah. Causation is totally a subset of correlation. And people who learned how to sound smart from reddit don't really know how to respond to that, I guess. We're all just parroting shit we heard once at a party, that sounds pretty cool and sciencey now.

Plus bro, my cousin like dropped out of a PhD program? and he says these vaccines contain 200% of the recommended daily value for dolphin mercury.

This post brought to you by sleep deprivation.
posted by hobo gitano de queretaro at 10:28 PM on June 16 [18 favorites]


The above post only correlated with sleep deprivation.
posted by converge at 11:26 PM on June 16 [7 favorites]


A correlation between A and B means that A causes B, B causes A, a separate factor causes A and B, or there was some sort of flaw in the study that identified the correlation.

E. someone cherry picked the data showing correlation is statistically discernable* - i.e. the margarine linked to divorce example - yet it's pure coincidence and is probably just being used to try to con you.

or F. the journalist couldn't understood the study, and so over-simplified the conclusions that the article is a complete bastardisation of what the study actually said, i.e is really just complete bollocks. Which is depressingly common.

I do hate pop use of that phrase though, and you're right that most users of it haven't actually thought through what they actually mean by it. But I think I'd rather that than have more people blindly swallow the mass-market pablum that passes for much science reporting that often just seems to be intended to bilk them out of money or make them afraid of something.

* I love this
posted by ArkhanJG at 12:19 AM on June 17


If that's not what is meant by the word "processed", then the word has no meaning.

This idea that if a word has a specific technical meaning in one context, it cannot be used in any other way in any other context, and if it's used in a different way, it cannot have a specific technical meaning, has very little to do with how humans actually use language.
posted by effbot at 12:39 AM on June 17 [9 favorites]


Schrödinger's Cat is not an attempt to explain quantum mechanics and shouldn't be used as one; it is a reductio ad absurdum (Erwin himself calls it a "ridiculous case") illustrating the issues he had with one interpretation of quantum entanglement.

My high school physics teacher was very, very, very adamant that we understood the difference.
posted by Spatch at 12:44 AM on June 17 [5 favorites]


Correlation simply refers to the extent to which specified variables vary. It has nothing to do with causation.
posted by carping demon at 12:55 AM on June 17 [2 favorites]


Statistical significance says this.

I have an effect A, and an effect B, and I say that A>B and that that is a statistically significantly difference (at 5%). This means that if A=B the data I observed would occur about 5% (or less) of the time. So if I'm testing whether my homeopathic medicine is more effective than placebo, about 5% of the studies I do will be significant.

Lets suppose I run 100 experiments, testing 100 different hypotheses. Only 2 of these are correct, because generating correct hypotheses is hard. Lets say my power is 100% (note, this never, ever happens!), so I detect one of those as significant. Of the remaining 98, I have a significance level of 5%, so I detect 5 (well 4.9) of these as significant. Whoops! This is why replication is important, as if I do the test again on these 6 I should only find one of them significant.

What we actually want significance to give us is the probability, given that the result is significant, that A>B. But this is much harder to do, and we need to be Bayesian to do it. That is, we need to assign a probability that A=B before we've done any experimentation, and then, having observed some evidence we now have the probability that we want! In that example above, if we had correctly guessed that only 2% of my hypotheses were correct I would then have identified that the probability of there being an actual effect when I saw one was 1/6. Sadly, we have to do that first step, which is not always easy! I am fairly confident homeopathic medicines are ineffective, so would assign a high probability that A=B a priori... others feel differently.

The other thing I haven't mentioned is effect size. Practically speaking, given a large enough sample size, you will always be able to show A is not equal to B. Its deeply unlikely that A=B unless you have a true like with like comparison. What we should be worried about most of the time is the size of the difference between A and B rather than the size of the p value.
posted by Cannon Fodder at 2:45 AM on June 17 [2 favorites]


Care to speak as a Bayesian about what's right?

Cannon Fodder has already commented on this above, but I'll write my thoughts anyway.

First off, my earlier comment was late-night snark (2:10am here), and shouldn't be taken too seriously. I do think the usual framework is right, as long as you know what it is right about. A lot of the people who laid down the rules for this knew what they were doing, but many people applying it don't, or don't apply it properly.

The conventionally taught framework is, as Cannon Fodder says, to take a hypothesis (a model that predicts some probabilities for your experiment or observations) and calculate the probability given that hypothesis of getting your results or more extreme ones, and seeing if this probability is smaller than some level (typically 5% is used). If it is, you call it 'statistically significant'.

The problems with this are (this list not being exhaustive):
* as Cannon Fodder said, this is the probability given that hypothesis. It doesn't tell you the probability that the hypothesis is wrong, since it assumes it is right. The probability of getting your results or more extreme ones (called the p-value) is often wrongly conflated with the probability the hypothesis is wrong - this is a problem with people, not the mathematics though, which is quite clear about what it is saying.
* 5% is a rubbish level of significance. It means that when the hypothesis is right, 1 in 20 times you will find the results don't fit it in a statistically significant way. You can choose a lower level, and in some fields you very often do (particle physicists choose '5 sigma', or 0.000003% if I've counted my zeros right - that's the significance they wanted for demonstrating the existence of the Higgs boson for example)
* it doesn't tell you anything about any other hypothesis - it only considers the probability from the default model you are testing
* some people feel that it is kind of daft to count the probability of a 'more extreme result', since you didn't actually get that result.
* a low probability given a hypothesis alone doesn't mean for sure it didn't actually come about by chance, and you should properly contemplate that possibility. As Sherlock puts it "when you have eliminated the impossible, whatever remains, however improbable, must be the truth"

A Bayesian will compute directly the probability of the hypothesis given the data, and compare it to the probability of another hypothesis given the same data, and compute the ratio of the two. This gives you a direct comparison of two competing ideas. It has the benefit that in the computation you can incorporate information about how likely it is that a hypothesis is right from what you already know (the 'prior probability). You can therefore include ideas like "extraordinary claims require extraordinary evidence" (often attributed to Carl Sagan, but really goes back as far as Laplace who said "The weight of evidence for an extraordinary claim must be proportioned to its strangeness."), and not just consider the improbable after eliminating the impossible, but weigh up two improbable options against one another. It also naturally incorporates Occam's Razor, by penalising hypotheses with free parameters. I personally feel it is a more natural and more easily understood approach, but it is often computationally more expensive, and as I said can be subjective.

The downside is that the prior probability can be subjective, so whereas a test of statistical significance is clear cut and can be just inserted into your paper without much clarification, if you want to be Bayesian about it you have to specify your prior, and other people may not agree with that. If I'm testing homeopathy I'm going to put a very different prior on my results than a homeopath would.

Usually the two approaches will agree - a strongly significant result will usually find its parent hypothesis strongly disfavoured by a Bayesian analysis in favour of another idea, but this isn't always the case. As an example, see Lindley's paradox. In the example on that Wikipedia page there's a population of 49,581 boys and 48,870 girls born in a certain time, and you're looking at the hypothesis that the probability of a boy being born is 0.5. It has a p-value of 0.0235 but compared to the alternative hypothesis that the probability is something different (anywhere between 0 and 1, uniformly) assuming equal prior probability to both, you favour the 0.5 probability hypothesis very strongly (the ratio of the probabilities is 0.95). This is basically because we've got one very predictive hypothesis (probability is 0.5) versus a very unpredictive one (no idea what the probability is).

Anyway, the main thing I'd want to get across is that the usual significance testing makes a calculation based on a certain thing being true, and it doesn't necessarily mean if a result reaches statistical significance that some other idea must be better. There's plenty of reading matter linked off Wikipedia here.
posted by edd at 3:51 AM on June 17 [6 favorites]


Is there a Bayesian interpretation of significance?
The closest analogue to a significance level would be the somewhat arbitrary 'lines in the sand' drawn for the Bayes factor (the ratio of the two probabilities I mention above) as listed here, I think.
posted by edd at 3:56 AM on June 17 [2 favorites]


I'm one of those science-people... but language doesn't work like these other science-people seem to think it does.

Also "organic" had a broader meaning before scientists squashed it down to mean tiny-stuff-with-carbon-backbone.
posted by zennie at 4:03 AM on June 17 [1 favorite]


Speaking as a Bayesian,

Is that like a cult or a club one could join? ;-)
posted by sammyo at 4:05 AM on June 17


Clickbait with delusions of grandeur
posted by Fists O'Fury at 4:07 AM on June 17 [6 favorites]


Not to be anti-dogmatic or anything but I've noticed a subset of folks in the scientific community that have trouble with the "common talk". I remember an old roommate getting upset when I said "don't be paranoid" and he reacted like I was making a clinical diagnosis. Um, folks, I totally agree that language needs to be precise in certain situations, but like dudes and dudettes, just don't be paranoid.
posted by sammyo at 4:09 AM on June 17 [8 favorites]


To say "Survival of the Fittest" are not Darwin's own words is a bit misleading. Darwin included them and a chapter by this name in his fifth edition of Origin of the Species. He did point out that he used this description from Spencer in economics (who, in turn, was commenting on Darwin). But, yes, he did point out that he was describing any advantage and ability to match one's environment.
posted by dances_with_sneetches at 4:11 AM on June 17 [1 favorite]


sammyo: pretty much yes :-)
posted by edd at 4:13 AM on June 17 [1 favorite]


Some of these are simply words that have multiple meanings. In literary and cultural criticism, a theory is an abstract framework that suggests meanings in texts and groups of texts. In law, proof indicates a fact that has been corroborated by evidence sufficient for the context in which proof is being sought. Natural in the context of food refers to a lack of certain forms of processing defined by regulation, as noted in the article itself. The fact that scientists use words with specific meanings in mind in no way invalidates other meanings. It causes confusion when the word is used out of the proper context.
posted by mobunited at 4:15 AM on June 17 [2 favorites]


The fact that scientists use words with specific meanings in mind in no way invalidates other meanings. It causes confusion when the word is used out of the proper context.

As a scientist (and a linguist) this is exactly what I think the problem is. Words can have multiple meanings. The problem arises when people don't know that there are multiple meanings - often, it means people mistaking the colloquial meaning for the scientific meaning, but it can be the other way around, too.

But another problem is that the context isn't always clear. Casual discussions of topics where scientific thinking is relevant but may not be the only way of seeing/discussing an issue, for example.

Though, many people reject scientific meanings of terms at the same time they are rejecting science as a way to understand the world. Creationists are probably the prototypical "you can't prove it"-iest group there is. In that case it really just is not a disagreement over the definition of the word "prove." It's a complete misunderstanding of how the scientific process works.

(Knee-jerk "correlation doesn't equal causation" can make me grar no matter the context.)
posted by Kutsuwamushi at 4:38 AM on June 17 [1 favorite]


Hell, I'm a MARKETING major and the way people use words* makes me crazy. They could only come up with 10?


*why I'm not in advertising...
posted by randomkeystrike at 5:26 AM on June 17 [1 favorite]


I don't have a problem with "organic". After all, "calorie" is already a term that has been badly broken, and we just kind of sigh and accept it.
posted by Foosnark at 5:26 AM on June 17 [1 favorite]


After glancing at that again, I think it's even worse than I thought it was when I looked at it the first time.

The stuff about 'theory' is pretty disingenuous and crappy, I IMO. As others have noted, words are often ambiguous or semi- so. Theories are not necessarily "big." There are small theories--roughly: hypotheses. There's some suggestion that theories are bigger than merely hypotheses--but that's not some inviolable part of a definition of the term. Perhaps scientists have elected to use 'theory' only to refer to big complexes of hypotheses and conclusions--and that's fine. But that's not what 'theory' means. An it's certainly not the only legitimate meaning.

And much of what the quoted piece says isn't even about the definition/misuse of the term. Rather, it's about some interesting characteristics of "the best" theories. Which has nothing at all to do with misusing the term.

Anyway, this isn't even the now-apparently-orthodox line about 'theory.' The thing you most commonly hear is: theories are hypotheses that have a lot of evidence in support of them. That's wrong, too, but for different reasons. Near as I can tell, that's a neologistic use specifically instituted to win the "just a theory" fight with creationists. It allows the simple response: you're confused if you say that something is just a theory, because "just" doesn't go with theory--to be a theory is to hold an elevated status, and to have already been supported by tons of evidence. That's just a rhetorical trick, and it's not the way 'theory' has always been used. (The right response to the "just a theory" complaint is: no, evolution is not *just* a theory--it's a theory with a shit-ton of evidence in its favor. (I'll admit, though--it's easier to shut people up with that little rhetorical trick, dishonest though it is...))

And don't even get me started on that "proof" section...

This really is just clickbait.

Mission accomplished, I guess...
posted by Fists O'Fury at 5:44 AM on June 17


I don't have a problem with "organic". After all, "calorie" is already a term that has been badly broken, and we just kind of sigh and accept it.

How is calorie broken? It is the amount of energy required to head one gram of water one degree kelvin. I can see how Calorie vs calorie could be misunderstood, but the simple change would be to replace Calorie with kilocalorie (or even better kcal).
posted by koolkat at 6:01 AM on June 17 [1 favorite]


good luck trying to get the world to understand our language. In the mean time, perhaps science types should substitute in words that enable correct understanding

proof -> possible evidence
theory -> systematic understanding
Quantum uncertainty -> (I don't think there's any hope for this one)
Learned vs. Innate (nature vs. nurture) -> Feasibly Extinguishable/trainable
Natural -> non-toxic
Gene for X -> Gene that might affect X (not much hope here, either)
Statistically Significant -> Statistically Discernible (FTA) / Likely not coincidence
Survival of the Fittest -> Survival of the Best Adapted / Survival of the Survivors
Geologic Timescales -> (We humans have puny brains, I don't think this can be fixed via language)
Organic -> Carbonic
posted by rebent at 6:16 AM on June 17 [2 favorites]


(Minor corrective note. Up thread I should either have had a power of 50% or had only one correct hypothesis to make my numbers work. I changed my mind about what I was doing midway through writing....)

Thinking about correlation not equalling causation, the thing that bugs me about that is that while causation matters in lots of fields, sometimes all you need is a strong correlation to predict things. If I wanted to guess when there would be more shark attacks, and had access to no data except ice cream sales I'd be pretty good at guessing when there would be more, even though the two do not cause the other.

That website with spurious correlations always interests me, because while I'm fairly sure most of the correlations are spurious (i.e. caused by random fluctuations and will not continue outside the selected time periods), if some of them are genuine theres some really useful data there!
posted by Cannon Fodder at 6:21 AM on June 17 [1 favorite]


Scientists agree that it's always more complicated than that.
posted by Kirth Gerson at 6:26 AM on June 17 [1 favorite]


What people don't understand about Occam's Razor is that it's a moral principle, not a logical principle. You can't apply it to other people's arguments. You apply it to your own arguments, because it's the right thing to do.

This seems almost right, but you've phrased it weirdly — I assume you don't really mean a "moral principle." It's a heuristic, not a logical rule.
posted by John Cohen at 6:29 AM on June 17 [2 favorites]


Stop torturing natural languages in your papers, and when attempting to explain your work to media and we'll talk. Until then, I'm ok with most of these having lay meanings.
posted by clvrmnky at 6:38 AM on June 17


the simple change would be to replace Calorie with kilocalorie

The better change is to abandon substance-specific calorimetric units entirely and switch to kilojoules, leaving the nutrition folks their customary unit. Joules are more useful generally, anyway.
posted by bonehead at 6:39 AM on June 17 [2 favorites]


Good thing I use 15% of my brain, so I understand all this stuff intuitively.
posted by robocop is bleeding at 6:45 AM on June 17 [4 favorites]


Geologic Timescales -> (We humans have puny brains, I don't think this can be fixed via language)

I wonder if this is an artifact of our education combined with a difficult concept? We've moved from a relatively short view of human history to a context where the age of the Earth or the universe, rather than some guesstimation of the start of human civilization, is the lower boundary.

I think we learned the clock analogy/cosmic calendar in first year university, but my son got it in grade seven or eight. It's one of those things that takes a long time to soak in.

It seems like the inability to understand this concept is a driver of both Creationism and climate science denial - people have a hard time imagining huge time periods, therefore they probably don't exist.
posted by sneebler at 6:48 AM on June 17 [1 favorite]


I still blame Hanna-Barbera for most of the common 'pre-historic' misconceptions... not to mention the obsession with flying cars.

So I should stop trying to teach my dog Kung Fu?
posted by Billiken at 6:58 AM on June 17 [3 favorites]


Until then, I'm ok with most of these having lay meanings.

Most of the items in the list aren't just words that have different "lay meanings" and technical meanings, they are cases where the "lay" interpretation is a significant misunderstanding of the concepts involved. The title alludes to this, "10 ideas", not just "10 words". This is also why the suggestions of word replacement are not the answer; the words themselves are only symptoms of a more difficult problem.
posted by kiltedtaco at 7:06 AM on June 17 [2 favorites]


On the subject of statistics and correlation != causation, this article by Peter Norvig is pretty interesting.
posted by Axle at 7:09 AM on June 17 [1 favorite]


Not only is #3 a complete straw man argument, but it means the author totally misses the point of #1 and #2. Let's review
#1: Science never really proves anything
#2: Scientific theories are entire systems of testable ideas which are potentially refutable
...
#3: In the end, we are made of quantum particles (protons, neutrons, electrons) and are part of the quantum universe.

Uh...no we aren't. Science never proves anything anything (#1), including the existence of quantum particles, which is a potentially refutable theory (#2), nothing more. Currently, our best model is to say that it is AS IF there were these entities called quantum particles, and that is the most we can say about it. The materialist fallacy is to mistakenly impute truth to a theory such as quantum mechanics, which is a great theory, just like Newton's theories were and are pretty awesome theories, despite having been superseded in some contexts by other, more useful ones.

Moving on, the author eludes the problem of the necessity of a totally undefined entity called an "observer" in quantum mechanics by referencing the more or less loony theories of the New Agers, including the theory that we "create" everything. That's not a very good theory from a scientific perspective (see #2). But it's also not the argument made by those, like myself, who believe (understand) that awareness precedes formal knowledge, including all of science, and that while from one limited perspective it is currently valid to say that our minds are made out of quantum particles, from another limited perspective it is equally valid to say that quantum particles are made out of our minds.
posted by haricotvert at 7:41 AM on June 17 [1 favorite]


Most of the items in the list aren't just words that have different "lay meanings" and technical meanings, they are cases where the "lay" interpretation is a significant misunderstanding of the concepts involved.

It's more than that; there are often a bunch of different meanings of a word, some "technical" and some "lay." So the word organic has a specific meaning to a chemist but a different specific meaning to someone dealing with food law/regulation (and this meaning is likely to vary by state in the US), and the general meaning of "healthy" to consumers. Leaving out the last sense, which is kind of what this list is fulminating against, the specific meaning for the chemist does not invalidate the specific meaning for the lawyer or farmer.
posted by GenjiandProust at 7:44 AM on June 17


kiltedtaco, I couldn't agree more. A large part of the problem behind most of these concerns is simplification of almost everything in normal use into dualities or a small number of categories. The classic example is nature vs nurture. This is also at the heart of most of the difficulties in statistics. We are forced to draw lines of "significance" to answer questions like "Is it or isn't it proven/toxic/safe?" where there are in reality ranges of probability, of effect, of factors.

When turning scientific results into something that can be used, we are force our floppy analogue continua of measures and observations into digital boxes, so that people can say "this is that and not the other thing". So we get arbitrary bounds of significance (or discernibility, I don't think the word matters), lowest observable effect levels, t-tests and PCA, all to sort results into categories that people can use. It's a messy process, and poorly understood even by many practitioners (aka abuse of the p-value).

Science communication is really hard to do right because most people want "simple" answers, preferably binaries. Things don't always fit into neat boxes, and the idea of continuous variability (with multiple factors) is not one that is easy to explain.
posted by bonehead at 7:51 AM on June 17 [2 favorites]


I like the many definitions of the word "organic". It lets me make bad puns in the supermarket.
posted by bonehead at 7:52 AM on June 17


So I should stop trying to teach my dog Kung Fu? Billiken.

Pony up and buy a panda, for God's sake!
posted by dances_with_sneetches at 7:56 AM on June 17


It's a bit wierd that the article uses "you" in the title since I would guess that io9's readership is by and large aware of these issues and identifies with the scientists. I get the feeling they're conjuring an imagined audience of uninitiated doofuses for their readers to snigger at. But, dear reader, the product is you.
posted by dmh at 8:15 AM on June 17


Science communication is really hard to do right because most people want "simple" answers, preferably binaries.

Tertiary facts? True/false/maybe or perhaps False/True/Truthier?

This thread could devolve into a discussion of science reporting or the p(?) value usage in medical journal articles or what what was that rejection of a bunch of journal articles constructed by an algorithm? It does seem like if the ideas are really really hard the is just a smidgen more of a burden by the science community to get the ideas presented in form that is digestible by folks other than specialists. A non-trivial request but I think warranted.
posted by sammyo at 8:16 AM on June 17


If the word "proof" is understood differently by scientists than by most other people, that doesn't necessarily mean that everyone but the scientists has it wrong. Prove has existed as a common word since long before the invention of the scientific method as we know it.
posted by sfenders at 8:20 AM on June 17


dmh: "It's a bit wierd that the article uses "you" in the title since I would guess that io9's readership is by and large aware of these issues and identifies with the scientists. I get the feeling they're conjuring an imagined audience of uninitiated doofuses for their readers to snigger at. But, dear reader, the product is you."

It's the clickbait formula: XX Things YOU need/will/should want/watch/cringe at/over.
posted by Big_B at 8:22 AM on June 17 [2 favorites]


If the word "proof" is understood differently by scientists than by most other people, that doesn't necessarily mean that everyone but the scientists has it wrong.

Yes, but people do often "have it wrong" when the apply the non-scientific sense to a scientific context, as in "Scientists have proven..."
posted by grumblebee at 8:23 AM on June 17


If someone claims that "science has proven" that asparagus causes cancer, or whatever, that is a clear sign that they're using "proof" in the more common, more ancient, and less mathematical way. There's nothing inherently wrong with doing that, and it's such a useful and well-established word that I think it fair to say that any misunderstanding comes at least equally from a mistaken tendency to understand it in the other sense.
posted by sfenders at 8:35 AM on June 17


If someone claims that "science has proven" that asparagus causes cancer, or whatever, that is a clear sign that they're using "proof" in the more common, more ancient, and less mathematical way.

Yes, but what's not clear is whether or not they understand what scientists mean when they say they've proven something. It's the mapping between those two definitions that's the problem, not the fact that there are two definitions.
posted by grumblebee at 8:41 AM on June 17


Science never really proves anything in the sense that the word "proof" doesn't even apply to the scientific method. But mathematics is also part of a scientist's toolkit, and mathematical propositions can be proven, as when for example a physicist proves that one "theory of everything" is equivalent to another, or a computer scientist proves that a particular sort can be performed in quadratic time. So, while "science never proves anything" might be strictly accurate, "scientists never prove anything" is not.

Moving on, the author eludes the problem of the necessity of a totally undefined entity called an "observer" in quantum mechanics

There's no problem of an observer in quantum mechanics in general. There's a problem of an observer in the traditional Copenhagen interpretation of QM, a problem which is basically solved by decoherent interpretations.

If someone claims that "science has proven" that asparagus causes cancer, or whatever, that is a clear sign that they're using "proof" in the more common, more ancient, and less mathematical way. There's nothing inherently wrong with doing that

Isn't there, though? What happens when it comes out later on that it's not asparagus after all, it's a microbe that lives on unwashed asparagus? "Oh, looks like the scientists were wrong after all!" Even though the original paper merely correlates consumption of asparagus with increased level of cancer and was factual in every respect. Sloppy reportage then causes people to equate good science with hokum. "Nobody knows. It's all guesswork anyway!"
posted by xigxag at 8:46 AM on June 17


Concerning correlation/causation etc. This came up recently toward the end of a JFK conspiracy thread.

impossible
unlikely
possible
plausible
probable
verifiable


So what would the "correct" correlation/causation continuum be?
posted by philip-random at 8:49 AM on June 17


on differing technical definitions: in stellar astronomy, "metal" means any element that isn't hydrogen or helium (there is a very good reason why you need a word to cover that, and "metal" was the best they could come up with I guess).

Chemists don't like it when you call Carbon a metal.
posted by vibratory manner of working at 8:52 AM on June 17


Scientists have a "natural" tendency to prefer the mathematician's definition of proof.
posted by sfenders at 9:08 AM on June 17


Billiken : I still blame Hanna-Barbera for most of the common 'pre-historic' misconceptions... not to mention the obsession with flying cars.

So I should stop trying to teach my dog Kung Fu?

Phooey!
posted by IAmBroom at 9:09 AM on June 17


It does seem like if the ideas are really really hard the is just a smidgen more of a burden by the science community to get the ideas presented in form that is digestible by folks other than specialists.

There have been all kinds of attempts to do exactly this, to redigest and reinterpret quantitative data. The simple fact though, is that most of the public and the media blows additional information off as too complicated. Does anyone really listen or understand when a news anchor relates the latest polling numbers and then says, mechanically, "with a margin of 3% error, 19 times out of 20"?

That's the core of the problem we're talking about here. How do you communicate complexity when your audience only cares about The Answer? How do you communicate, for example, an estimated age range of the planet based on multiple pieces of evidence to someone who thinks the earth was created on Sunday, October 23, 4004 BC at 6PM? Any range of variation is characterized as: "well you don't really know, do you?"
posted by bonehead at 9:53 AM on June 17 [4 favorites]


It's because most people still labor under the kinds of logical errors Bacon described as the "Idols of the Tribe." Every day common sense is based on implicit assumptions that aren't logically sound like belief in first and final causes operating in nature and various other logical fallacies. Every day common sense reasoning comes very easily and works well enough to get the job done even if it isn't actually right on the level of the underlying principles involved in a problem.

Scientific reasoning is a specialized kind of reasoning that can work to correct the problems in so-called common sense thinking, but most people can't realistically be expected to systematize their everyday thinking all the time and they don't really need to to function normally as humans, anyway, so there's no incentive to think more systematically for most people.
posted by saulgoodman at 10:11 AM on June 17 [1 favorite]


Related to the article's discussion of the misuse/misunderstanding of "theory":

I have an advanced degree in Ecology and Evolutionary Biology, and one thing I refuse to let go of is taking issue whenever I hear someone say that they "believe" (or, in rarer cases, "don't believe") in Evolution.

I don't believe in Evolution. But that's because Evolution is not a belief system. Evolutionary theory (and I mean theory in the correct way) is a means of understanding Earth's biota, and specifically the vector for change over time in those species that inhabit it. Nothing in our understanding of the natural world makes much sense without an understanding of Evolutionary theory.

However, "belief" in Evolution would imply it's immutable, and it's not. Refinements are made all of the time to Evolutionary theory, and while it's underpinnings have remained for the most part unchanged (variable populations, inheritance of characteristics, and competition for resources), other parts have been areas of both debate and increased understanding since Darwin's time (sexual selection, kin selection, nature vs. nurture, pre-emptive competition, etc.).

I like to hammer on the semantics of "belief", much to the chagrin of whomever I'm talking to on the subject, whether it's a scientist or a faith-based creationist. I think it's the fundamental point of getting past the stupid Creation vs. Evolution debate, which is not a debate at all. You can't pit a faith/belief system against an iterative scientific knowledge model and expect any progress. Get someone to grok science's way of knowing instead of the tired "did we come from apes?" and the possibility of understanding between the disparate parties improves markedly.
posted by mcstayinskool at 10:28 AM on June 17 [4 favorites]


That was a good little list.

My tiny little scientific/philosophical terminology-used-incorrectly pet peeve (and surely, it is a small-ish thing but it drives me bonkers) is people using 'beg the question' wrong - and it happens all. the. time. Like even very intelligent people do it and I cringe a bit every time. Usually people mean "strongly raises the question" or "leads to the question," not begs the question. "Red objects are red," begs the question.
posted by Lutoslawski at 11:33 AM on June 17 [2 favorites]


At this point, "begs the question" is a phrase that means "strongly raises the question" that happens to have another obscure use in the technical jargon of formal debate and logic. In ordinary conversation, it doesn't carry the technical debate meaning any more than "artificial" carries its original meaning of something like "artistic" or "skillful," or "terrific" carries its original meaning of "terrifying."
posted by ROU_Xenophobe at 12:12 PM on June 17 [3 favorites]


You can have my technical meaning when you pry it from my cold, dead, technical hands!

Surely we don't want 'natural' to come to mean 'bought it at the coop.' Just say no to change!
posted by Lutoslawski at 12:18 PM on June 17


"belief" in Evolution would imply it's immutable

Why? I believe in my friend John (both in the "He exists" sense and in the "You go, girl!" sense), but I don't believe he's immutable. He changes all the time.

I guess, if you want, you can define "belief" as "commitment to an the claim that such-and-such is immutable," but that's certainly not standard usage.

When I say that I believe in Evolution, I mean that I believe that the current theory is predictive. I don't mean that it can't or won't be refined. I'm sure it will.

I like to hammer on the semantics of "belief"


Me too, and we could have a fun discussion. I've been talking to people about it for three decades. In my experience, the word (even in everyday usage) has many meanings.
posted by grumblebee at 1:24 PM on June 17


Theory vs Hypothesis is definitely up there for me. I say the former all the time when I mean the latter, and am trying to switch.

It's unfortunate that the article never says this. The word "hypothesis" doesn't appear in it at all, nor does any other suitable alternative. This article has a few problems like that; some of it is due to being collected from multiple contributors but there are a number of places where the writer could have pulled it together a little better.
posted by George_Spiggott at 1:24 PM on June 17 [1 favorite]


any more than "artificial" carries its original meaning of something like "artistic" or "skillful," or "terrific" carries its original meaning of "terrifying."

When did "artificial" not mean "made with artifice"? Artifice may have acquired some slightly different connotations of crafty trickery, but it still carries what I would have thought was very close to its original meaning. Or at least a really old meaning. Isn't it "artisanal" that's changed more?

I think it's too early to count the incorrect usage of "begs the question" as correct. It's had a spate of popularity lately sure, but I'm not convinced it has staying power. Many people were saying and writing "for all intensive purposes" for a number of years, but that trend has nearly been exterminated by now (after a hundred years some say) due to it being a clear error (deliberate or not) that was eventually recognized as such despite (or because of) a surge to widespread use.
posted by sfenders at 2:01 PM on June 17


There's no problem of an observer in quantum mechanics in general. There's a problem of an observer in the traditional Copenhagen interpretation of QM, a problem which is basically solved by decoherent interpretations.

If I'm not mistaken, the philosophy of quantum decoherence posits multiple mutually unobservable universes splitting off from each other to create co-existing alternate histories. Anyone seriously suggesting this as a solution to the observer problem should be careful about how they talk about New Agers.
posted by haricotvert at 2:07 PM on June 17 [1 favorite]


Isn't there, though? What happens when it comes out later on that it's not asparagus after all, it's a microbe that lives on unwashed asparagus? "Oh, looks like the scientists were wrong after all!" Even though the original paper merely correlates consumption of asparagus with increased level of cancer and was factual in every respect. Sloppy reportage then causes people to equate good science with hokum. "Nobody knows. It's all guesswork anyway!"
Regardless of how much "science" bloggers and administrators of powerful scientific institutions wish otherwise, science isn't about a group of infallible priests interpreting the will of a distant god. Science is all about making mistakes and correcting them. Trying to hide that fact from the public so that science can fit into the old authoritarian-priesthood model is the problem, not a goal that we should be working towards.
That's the core of the problem we're talking about here. How do you communicate complexity when your audience only cares about The Answer? How do you communicate, for example, an estimated age range of the planet based on multiple pieces of evidence to someone who thinks the earth was created on Sunday, October 23, 4004 BC at 6PM? Any range of variation is characterized as: "well you don't really know, do you?"
But it is OK to not "really know". One of the really important things to understand about science is that you can make useful progress by describing what you don't know. As you look into the unknown, structure and pattern become recognizable. Knowledge advances.

So the best answer is something along the lines of "that's a great point! Let's talk about what it means to know something." And then you work your way to the point that we know the age of the universe within a certain margin of error, and both the age and margin can (and should) change as we acquire new knowledge.
posted by b1tr0t at 2:08 PM on June 17


But yeah. Causation is totally a subset of correlation.

Which would mean that causation implies correlation, but not the other way around. Not seeing the problem here (though the saying is certainly much-abused in practice)
posted by atoxyl at 2:40 PM on June 17


on differing technical definitions: in stellar astronomy, "metal" means any element that isn't hydrogen or helium

And reading up on that reminded me of one very important point: most things actually are chemical free. At least on an astronomical scale. The vast preponderance of the earth is chemicals.
posted by ambrosen at 3:01 PM on June 17


"belief" in Evolution would imply it's immutable

Why?


In the context of Evolution, because it's oft compared with Creationism/Religion/Faith, where the word "believe" means something very different.

I agree with you that belief can mean many things, and by a broad interpretation you can "believe" in Evolution. But, because of the nature of where it sits in public debate, using the word belief is, in my opinion, semantically incorrect. Further, separating a belief-based system from an evidence-based system is an educational point that's worth embracing and using as a teaching point.
posted by mcstayinskool at 7:37 PM on June 17 [3 favorites]


But it is OK to not "really know".

In a journal article, sure. Presenting a popular science tv show, perhaps. As a commentator "debating" on FoxNews, it makes you the punchingbag. In the courtroom, it puts you on the losing side.
posted by bonehead at 8:51 PM on June 17


If I'm not mistaken, the philosophy of quantum decoherence posits multiple mutually unobservable universes splitting off from each other to create co-existing alternate histories. Anyone seriously suggesting this as a solution to the observer problem should be careful about how they talk about New Agers.

That's not entirely accurate. Many Worlds is one version of decoherence. But in any event, your insinuation that Many Worlds is wackadoodle is unwarranted. Unintuitive as it may be, it is at least consistent with our current understanding of the way the universe works, and is as far as I know the prevailing interpretation among contemporary physicists working in QM.

science isn't about a group of infallible priests interpreting the will of a distant god. Science is all about making mistakes and correcting them. Trying to hide that fact from the public so that science can fit into the old authoritarian-priesthood model is the problem, not a goal that we should be working towards.

Unless I'm mistaken, we seem to be basically agreeing here. I wasn't in any way implying anything about infallibility or hiding the truth from the public. I was suggesting it's imprudent to pass off a gross misrepresentation or oversimplification as the truth, as in when published findings are inaccurately called "proofs." If anything, calling a scientific paper a "proof," as if it asserts a transcendental truth, is something that falsely plays into what you're calling the authoritarian-priesthood model. On the other hand, there's the very real problem that one side, the politically religious, are very confident and assertive in what they claim to be truths, and if a researcher goes up against that with a "the findings seem to indicate some likelihood that etc.," she's gonna have a difficult time making her case.
posted by xigxag at 9:11 PM on June 17 [1 favorite]


Unless I'm mistaken, we seem to be basically agreeing here.
I think we largely agree. Where we might disagree is that I'm suggesting that (a) we should avoid engaging the politically religious on their terms and (b) separately we should seek to redefine the debate so that it is not only OK to be less than 100% certain, but it is great to have error bars around your knowledge, when you need them.
posted by b1tr0t at 12:49 AM on June 18 [1 favorite]


b1tr0t: 100% in agreement. People often seem to want certainty more than they want the truth, is the problem, and I think even very intelligent and rational people are often also subject to the bias.
posted by saulgoodman at 6:31 AM on June 18


The point I am making is not that Many Worlds is "loony" or lacks internal consistency as a theory. I'm not in a position to evaluate the math, but I'm confident that if the QM community takes it seriously, it has internal validity. What I'm pointing out is the tendency to make unsupported projections from the theory into areas that are not within its purview, and thus to give the false impression that the theory is "true", rather than "extremely useful," which is the most any theory can ever be. There are not "Many Universes." The most we can say is that based on current understanding it is AS IF there were Many Universes. This is not an important distinction as far as the theory itself is concerned, but it is of crucial importance to understanding the world we live in, because if we forget that theories are AS IF propositions, we slip into a false materialist viewpoint that has powerful implications for how we actually live our lives and treat other people and the world. That's why I keep raising the issue. But I also understand the QM theorists fully warranted frustration with those who would make their own unsupported extrapolations (such as the ones the New Agers are so fond of).
posted by haricotvert at 8:04 AM on June 18


« Older Sub Commander...  |  Lou Reed's 1979 LP The Bells, ... Newer »


This thread has been archived and is closed to new comments