Debate:Falsifiability/Archive2

This is a Debate page. Feel free to add your own spin on the story. Please keep it civil! This debate was created by signed users.

More Half Assed Science

May I refer you to the following paper on predictions on the spread of AIDs in the UK. You have to register to read the whole thing but the point I'm trying to make is visible from the introduction. The key phrase is that the paper is to determine to what extent extrapolation forecasts can be reliably made. Now this is pure stats and the results will be given in percentages. The end reslt will not be black and white but shades of grey with limited confidence in the outcomes. How limited? That's part of the story. The question is, is it science. The authors have analysed the data, applied the tools available to them, and made predictions from those data. Because the predictions are probabilities they cannot be, strictly speaking, falsified but is this really half assed science. If it is what exactly are you calling half assed about it? The methodology? The fact that it's not black and white? Would you have suggested to the British Government at the time that they disregard this work because not all of the variables had been eliminated? Were they doing a half assed job just to get published and avoid perishing - be careful who you insult here - or were they using their scientific skills to answer a very pressing question of the time. The FDA may be satisfied with half assed jobs - though I doubt it - but I'm sure I wouldn't like to read a half assed paper before such an august body as the RSS. Silver Sloth 12:34, 7 May 2009 (UTC)

Okay, there are a bunch of questions here. First, is it science? My opinion: It's a hypothesis. It'll be science when the projections can be mapped against the future progress of the disease. Second, is it useful? I guess insofar as it's an effort to mathematically model, it teaches us how to model. My highly unscientific prediction is that this hypothesis will be falsified by the data. However, in the meantime, it will be used as "science" for political purposes. Ungtss 13:36, 7 May 2009 (UTC)

My highly unscientific prediction is that this hypothesis - there you have it. You obviously didn't read the paper or you would have noticed that the date was 1987. Furthermore, because it's about AIDs you rate your guess over two experts in the field without even reading what they had to say. You don't want a debate - because you won't listen to anything that disagrees with your world view. You're not interested in Science, you're only interested in your own prejudices. Bye. Silver Sloth 13:56, 7 May 2009 (UTC)

Ooh, tricky game of gotcha. Were these projections ever mapped against the data? Ungtss 14:00, 7 May 2009 (UTC)

Gotcha? As I read the post I was able to infer that it was an old article - "at the time" clued me in. Your question, of course, makes sense. Can we haz data pleez? ħuman 03:56, 8 May 2009 (UTC)

The paper is not forecasting the spread of AIDs but a meta discussion about the reliability of such forecasts. It's high level stats and well outside my area of competence. I referenced it because I know one of the authors and, by implication, Ungtss is saying that he is not a scientist - and worse, that he only publishes to get noticed and avoid perishing which I find deeply offensive. Hey, you offend my dad and you offend me. Silver Sloth 08:59, 8 May 2009 (UTC)

Okay, let's evaluate your recent attempts at reasoning. "Ungtss didn't notice the date of a really old paper" --> "Ungtss obviously didn't read the paper." --> "Ungtss doesn't care about science." Second attempt: "Ungtss says papers that are content with minor statistical differences rather than actually isolating variables" --> "Silver Sloth's friend is not a scientist and only publishes to avoid perishing." Do you see a pattern there? Hypothesis: "Ungtss is an asshole." You are observing facts which are arguably consistent with that hypothesis, but also consistent with an alternative hypothesis ("Ungtss thinks differently than I do but is not an asshole"). If you were going to be scientific, you'd test your hypothesis by asking me if I mean the things you think I mean. But instead, you jump to a conclusion without test. That, my friend, is confirmation bias. You already think I'm an asshole, so every little thing I do is interpretted to confirm that pre-existing belief. That, and many other reasons, are why if you don't test a hypothesis, you aren't doing science. Ungtss 09:29, 8 May 2009 (UTC)

Ungtss says Experiments that produce 80/20 results are half assed and the experimenters haven't worked hard enough to remove the variables -> The vast majority of clinical trials are like this so they must be half assed and the experimenters haven't worked hard enough to remove the variables -> Medical Science is either half assed or the practitioners aren't working hard enough. Where's the fault in that logic? Your ridiculous definition of that only experiments that give black/white answers are good science is as insulting to those in Medical Research, and most other disciplines, as it is not true.

The flaw in your reasoning is called "straw man." You are creating an overgeneralization of what I said. Kangaxx said, "I agree that this is bad science in a way that it settles for a (relatively) enormous margin of error, that it relies on a not complete understanding of how things work. However, it is science." I agreed. Science that settles for huge margins of error is half-ass science. You then turn that into something bizarre -- "Medical science is either half assed or the practitioners aren't working hard enough." But I never said that. You're using strawman and overgeneralization. But whatever gets you off, I guess. Ungtss 17:23, 8 May 2009 (UTC)

OK, so going back to the 80/20 thing here's a hypothetical experiment. Drugs-r-us has produced an new treatment for the previously untreatable hurtsalotitis. This treatment, costalotium is claimed to have an 80% success rate. In the clinical trials 78% of the test group survived and 100% of the control group died.

Questions

1. Did the experiment confirm or deny the claim?
2. What variables should be reduced to get a 'black/white' answer?
3. Was the experiment 'good science' and, if not why not?

Well, that one is pretty clear cut - although NOT black and white. Now for something more real world. This time Drug-r-us has produced a new drug to mitigate the effects of MS. Note that this is not a cure. Of the test group 42% report remission. Of the control group 25% report remission. However, in the long run, both groups have the same prognosis. Same questions

1. Did the experiment confirm or deny the claim?
2. What variables should be reduced to get a 'black/white' answer?
3. Was the experiment 'good science' and, if not why not?

Not so easy now, is it? Does that make it bad science, here in the world of MS where the only certainty is a long slow undignified death? If you had MS would the results be good enough? How do you define good enough. This is real science in the real world. Silver Sloth 11:05, 8 May 2009 (UTC)

You're going to have to fix your facts, because it shows a profound misunderstanding of how medical treatment works. There is no "survive/not survive." There is "average survival rate during a given time period," and "average life-span with and without treatment." But there's no "die/don't die." We all die. So be more precise. Better yet, pick a real case. Ungtss 17:23, 8 May 2009 (UTC)

Nice try at ducking the question. As you seem to know about medical trials you know that survival rates are usually given as 'n' years from diagnosis, and, of course, this 'n' varies widely with the disease. So, in the first example s/survived/survived five years/g and s/died/didn't survive 5 years/g. Now, will you answer the question or will you further nit-pick tocontinue avoiding it? Was this trial, as described, good science or did the 80/20 outcome, and the fact that it wasn't bang on 80/20 make it half assed? What level of statistical significance makes it good enough. You seem to insist on 100% certainty which practically unachievable in clinical trials. And then, with the MS drug, here we have something that may or may not work with huge margins of error because of the multitude of unknowns surrounding this very complex disease. Science that settles for huge margins of error is half-ass science. Your very words, not quoted out of context, not over generalised, but how it has to be in certain areas of medicine. Or are the scientists investigating MS doing it just to get published and avoid perishing - your exact words again - and now who's guilty of over generalisations. Silver Sloth 18:13, 8 May 2009 (UTC)

Merits: Any paper that concluded "80% of the population that died after 5 years survived, 20% did not; 100% of the control population died" would absolutely be half-ass science. What happened at 6, 7, 8? Was the condition itself cured, or was it merely slowed, or were the symptoms merely mitigated? Let's say 80% of the population survived 5 years, 60% survived 6 years, 40% survived 7 years, etc. Have you extended life for an average of 3-5 years? Yes. But you haven't cured the disease. Much of modern medicine takes great pride in delaying the inevitable. That doesn't impress me. So I lived an extra two years with colon cancer, with my colostomy bag and my incontinence. So what? Looks great on paper, but it's a shitty life. Until you figure out what actually causes cancer and how to actually prevent / cure it, you're just mitigating the problem. Does that mean it's useless? Of course not. But anybody who was content with that is content with shoddy science.

I had a close friend that died of adult-onset leukemia. Her name was Heather. And the doctors pumped her full of this drug and that drug, locked her in a room sterile room, gave her chemo hoping it would kill the cancer before it killed her. The cancer actually went into remission for a while. But 3 months later, it came back and she was dead. All those efforts at "mitigating the disease" and "extending the lifespan" are well and good, but real science focuses not on how to eek a few more statistical months out of her, but how to actually end the disease. A toast to the people who are trying to do that.

Now to your questions.

1) Well if the claim was "This drug extends lifespans an average of 3-5 years," then the experiment confirmed the claim. But they still haven't figured out what causes the disease.

2) To get a black/white answer, you find the cause. Like finding poliovirus and developing a vaccine killed polio. 3-5 additional miserable years of life? You can keep it.

3) It was good science in the sense that it actually tested its hypothesis. It would be bad science to be content with this.

And finally, I'll address your problem with confirmation bias. "Ungtss asks you to clarify an unrealistic fact scenario" --> "Ungtss is trying to duck the question." Your conclusion only follows if you assume "Ungtss is an asshole." Take away that assumption, and your conclusion is ridiculous. Retain that assumption, and everything I do or say can be interpretted as further confirmation of it. Confirmation bias. Ungtss 12:15, 9 May 2009 (UTC)

"Ungtss" (I notice how you hide behind an alias), you have free will and it's clear to me that you are going to exercise it to the point of embracing absurdities. I looked at many of your edits before reverting them and all you do is repeat anti-American lies, so there's a 95% chance you're not here to glorify the Lord Jesus Christ. You also clearly deny that Obama refuses to salute the flag and that including the phrase "under God" in The Pledge of Allegiance stops autism. I repeat: Conservapedia has long observed that girls are injured in sports at much higher rates than boys. That's basic logic, though I'm not optimistic that you or other liberals will admit it. Open-minded people who respect logic have no difficulty elevating logic over personal whim. The FBI would be happy to do something about you.--Aschlafly 16:41, 21 December 2024 (UTC)

We're going to have to agree to differ. I think you're constantly shifting the goalposts. You ask for a definition of survival - I give the standard one - and then you want more. I hope that, for your sake, your newborn child is never, ever, in the position where you'll give anything on earth for 3-5 additional miserable years of life - Yes, I think the way you denigrate Medical Science and those that practice it along with your absurd definitions of what makes good or bad science make you an arrogant asshole. You probably think the same of me. Silver Sloth 20:11, 9 May 2009 (UTC)

I don't think you're an asshole at all. I think we're not communicating very well. A specific study would be helpful to analyze in this context. Also, for the 4th-5th time, I'm not dissing medical science. I'm cheering the folks that are dead-set on answering the essential questions and findings cures, and describing as half-ass anybody who is content with anything less. Finally, I'm not shifting goal posts -- the fact set is not the goal post, our analysis of the fact set is, and I'm trying to find a fact-set that will help us explore what good and bad science actually are. But I can tell your interest is fading, so I wish you well. Ungtss 20:24, 9 May 2009 (UTC)

Ungtss, I think you are the only one still looking for this so-called 'fact-set'. I think most of the rest of us accept that any inquiry which faithfully uses the scientific method qualifies as 'good' science, and we've offered plenty of examples. If that's not good enough for you, then I'm sorry, but I don't think we're going to be able to convince you otherwise. Jorge _{Radix Malorum Est Superbia} 21:15, 9 May 2009 (UTC)

My wife was diagnosed with multiple myeloma six years ago. She had radio- and chemotherapy and a stem-cell transplant all during the first year . Since then she has enjoyed an active live in good health. Yes it would be great if a magic bullet was found to cure all cancers and we could be sure that she would not succumb again, but you are low-life to insinuate that those who seek to improve palliative care or extend the active life of sufferers are half-assed scientists. Генгис^{Is the Pope a Catholic?} 19:46, 11 May 2009 (UTC)

Reading comprehension according to Ghengis: When Ungtss says "I'm cheering the folks that are dead-set on answering the essential questions and findings cures, and describing as half-ass anybody who is content with anything less," what he really means is "those who seek to improve palliative care or extend the active life of sufferers are half-assed scientists." Ungtss 20:21, 11 May 2009 (UTC)

Say, how are you doing on those examples of scientists who are content with "good enough", not counting of course those who only have "good enough" available currently for real-world needs and are willing to use whatever they can get while still pressing forward? --Kels 20:25, 11 May 2009 (UTC)

Well that leaves unanswered the question, "What is good enough for real-world needs?" And that's case specific. It's a waste of time for me to name examples unless you're willing to more carefully defined the criteria for what should be considered "good enough." Ungtss 22:46, 11 May 2009 (UTC)

So strawman, then? --Kels 22:51, 11 May 2009 (UTC)

What? Ungtss 22:52, 11 May 2009 (UTC)

You've been railing against these scientists who you've already said are content with "half-ass". Basically saying "eh, 80% is good enough, let's go research something else". I figured you'd have a lot of examples at hand, but it seems you don't, and are basically making these researchers up. --Kels 22:59, 11 May 2009 (UTC)

Did you read what I wrote? You framed in such a question as to assume some objective standard of what is "good enough" for public policy. That depends on the issue, of course. But I'm not going to bother with examples unless you can define what you think is "good enough." Well, never mind. Let me help you. I think that before you start using the word "pandemic" in press conferences, scientists should have evidence that the virus in question is more deadly than your run of the mill flu virus. Not only new, not only spreading, not only deadly, but more deadly than your average, run of the mill flu virus. 300-400k people do die of the flu every year, after all. And when the CDC and WHO start raising pandemic levels, governments and people are foreseeably going to panic. But the WHO and CDC were content with less before they started throwing out the P word. Mexico City shut down for days, billions spent on a vaccine for this bug, trade blocks in Eastern Europe, and every pig farmer in Egypt out of business. For nothing. Ungtss 23:07, 11 May 2009 (UTC)

Well first off, you apparently don't know what "pandemic" means. It doesn't mean "deadly", it means "widespread". And yes, the more widespread a virus is, the more potential there is to become deadly, that's not really controversial. Not to mention, if you don't catch it in the early stages and it does turn deadly, then it's too late after you've waited a while. Yes, you take the chance of having egg on your face, but it's a whole lot better to take extra precautions than not enough. Perhaps this video might help you understand a bit better (he's got others on the same subject as well). Scientists don't have a lot of control over the media and politicians after the findings are released, and I don't see that the WHO was out of line with that. They did not recommend killing pigs, and the Egyptian government has been roundly criticized for the act. Say, how long do you think the WHO should have waited? How many days? How many deaths? What are you "content" with as a death toll before you decide it's actually dangerous?

To go back to the "good enough" business, you are the one who's been bitching about scientists settling for 80%, good enough, or whatever you define as below the benchmark of good science, whatever that is. So by your definition of "content with anything less" (your own phrase), can you name actual researchers who were content that way and didn't keep researching to try to improve on those levels? Ones you actually understand, that is. --Kels 23:27, 11 May 2009 (UTC)

So many layers of stupid to dissect here.

1) As the WHO defines "pandemic," it is useless for public policy, because it does not speak to the dangerousness of the disease at all. It speaks only be being widespread, infectious, and new. But if a strain is widespread, infectious, new, and harmless, then why the hell should we take radical action to stop it? Answer: No reason. But they've defined pandemic so broadly that it applies in cases where no action need -- or even should -- be taken.

2) As much as these scientists would like to blame the media for their own failures to communicate effectively, they need to hearken back to that one class of mandatory English they took in college, and realize that words carry connotations, and as much as they like to say that "pandemic" only means "widespread and new" to the elitists, it means "widespread and deadly" to the public. When people think of Pandemic, they think of Plague and 1918. The WHO can hide behind their technical (and useless) definition all they want, but "pandemic" will always cause panic in the public. Anybody with any knowledge of the real world would know that.

3) The point is that "good enough" begs the question "good enough for what?" The WHO's evaluation of the situation was not good enough to support their actions, in my opinion. This is why I initially asked you to name a specific situation, and define what you thought was "good enough" evidence to draw the conclusion. I picked one. You want a different one, pick a different one. For now, I'm content to discuss this one. Ungtss 02:08, 12 May 2009 (UTC)

You've got a real axe to grind against the WHO here, don't you? So you blame them for the 24-hour news cycle which has been known to ramp all sorts of things way out of proportion, the emotional and foolish reaction of the Egyptian government, and people being ignorant of the terms even though they make efforts to actually educate the public. So what? You blaming them doesn't make it true. You also have a bizarre double standard on the go here. You demand science be precise as possible, and yet when they use precise language you bitch because it's not colloquial enough. That's messed up. The fact of the matter is, when the flu first broke out in Mexico, there were deaths, and it spread quickly. In those first couple of days, when it's critical to make a decision, the decision was made to take precautions. And honestly, most of the actually recommended precautions were sensible. Avoid crowds. Wash your hands. Wear a mask in some areas. Avoid travel to and from areas known to be infected. Now we dodged the bullet in that it wasn't as deadly as expected, great! But if it turned out to be deadly and they waited the several more days to be sure like you're asking, a lot more people would have died. You seem to be okay with that concept, which I find disturbing.

As to your point three, this makes no sense. YOU are the one who's been bitching for days now about scientists being, in your own words, "content with anything less". So why should I provide you with a specific situation that I don't think exists? You're the one who thinks there are lots of scientists "content with anything less", and presumably get to some arbitrary point and quit researching. The fact that you picked an example that does not apply to that concept is telling. Did the WHO just say "Okay, close enough, go with it" and stop researching and testing? Or did they make a decision when one needed to be made quickly using the best data available at the time, and kept researching to get better data? --Kels 02:31, 12 May 2009 (UTC)

I have no axe to grind. This is an example I chose to illustrate a principle: that scientists should not take or advise action based on evidence inadequate to support the significance of the action. When the WHO starts running around using the "P" word, people take notice. They should not do that lightly. In this case, the stupidity comes down to definitions -- a point you didn't address. Care to try again? If the WHO were being responsible, they would use the term "pandemic" as it is commonly used (widespread, infectious, and unusually harmful) rather than creating a rather useless and misleading definition by removing the significant element of harm, and replacing it with the insignificant element of novelty. They would therefore refer to this strain as a "new strain of the flu which has resulted in some deaths in Mexico City and therefore deserves watching" rather than "swine flu" (and the CDC did call it swine flu several times) with "pandemic potential." They gave the media and politicians all they needed to run with the insanity. And they got their press coverage and guaranteed funding, so all is well. Ungtss 03:45, 12 May 2009 (UTC)

1) Do you feel they used the term pandemic lightly?

I think they have defined the term in such a way as to make it meaningless, and then used their meaningless term to a public that understands the term differently, which resulted in a panic. The long-term danger here lies in "boy who cried wolf" syndrome. They lose a lot of credibility with these expensive, empty scares. Ungtss 13:02, 12 May 2009 (UTC)

2) Do you feel, in light of the early deaths, they had the luxury to wait several more days to be sure? Keep in mind, it was days in fact before it was clear the death toll would be low, while the virus did spread quickly to various countries.

I think they had an obligation to figure out the causes of the early deaths. Flu kills those with weak immune systems. Poverty, high altitude, cold, pollution, high altitude, exposure, and lack of rest are immunosuppressive. They needed to go to the family's of the dead within 6 hours to find out what conditions they were living in, to eliminate the potential that this was an ordinary flu which killed those living in poor conditions. But they didn't bother to falsify that alternative hypothesis -- they just ran with the "new bug might kill everybody" hypothesis. Ungtss 13:02, 12 May 2009 (UTC)

3) Do you believe the WHO and CDC are doing nothing to try to educate the public?

No. Ungtss 13:02, 12 May 2009 (UTC)

4) Do you feel the media has no culpability?

Who's to blame, the politician who says something stupid, or the media who exploits it? Both, of course. But if a politician thinks they can say something stupid and then point the finger at the media for exploiting it, he's an idiot. Ungtss 13:02, 12 May 2009 (UTC)

5) Given that it actually is a pandemic, and the definition of pandemic actually is a widespread virus covering multiple countries and populations at once, what term should they have used?

They need to use a useful definition of "pandemic." New infectious strains of diseases pop up all the time. They only deserve public policy attention if they're unusually harmful. As I said above, I think they should have said, "We have a new strain of the flu virus which is spreading. We have no evidence that it's more harmful than any other strain (as flu kills 1,000 people a day worldwide), but it might be, so here's what we recommend." That's called "being responsible and accurate."

6) Had the virus turned out to be lethal, as early reports suggested it could be, would the WHO have been negligent if they'd just told people to be careful of "swine flu"?

The virus did turn out to be lethal -- just like every other flu virus turns out to be lethal to thousands each year. As I said above, their responsibility was the report the facts, not their untested hypotheses. Ungtss 13:02, 12 May 2009 (UTC)

7) How does any of the above apply to your contention that there are scientists are "content with anything less"?

I suspect the answer to 7 is "not at all". --Kels 04:14, 12 May 2009 (UTC

You gotta work on your reading comprehension. Let me try again. "Good enough" requires an answer to the question "Good enough for what?" In this case, the death of 8 people to flu in Mexico City was "enough" for the CDC and WHO to hold press conferences calling it a deadly new disease likely to go pandemic. The evidence was not "good enough" to support the action. That's the third time I've explained that. Do you understand yet? Ungtss 13:02, 12 May 2009 (UTC)

You know, I typed out a lengthy response to every point, but got edit conflicted and you know what? Fuck it. It's clear that you haven't done your research, beyond maybe Ron Paul's blog or something. You seem quite okay with endangering lives by delaying announcements until everyone is 100% sure of everything, and have this bizarre requirement that WHO not follow procedures and use inaccurate language, despite the fact that the actual WHO press releases and statements to governments and the press actually say what you want, that there is no immediate cause for alarm but recommend increased monitoring and precautions. You clearly don't know what you're talking about, and the whole mess doesn't support your own argument. I understand quite well that you think the evidence wasn't "good enough", but the evidence shows it was what was available at the time when decisions had to be made and they didn't have the luxury of "oh, a few more days to investigate please". This is the real world, where sometimes decisions have to be made on the ground with what you have, not with what you wish you had. --Kels 13:38, 12 May 2009 (UTC)

Ah yes, the old "I had a good argument but lost it, so I'll just attack you instead of addressing the issue." Excellent work. Ungtss 14:14, 12 May 2009 (UTC)

Whatever you like to believe, man. Knock yourself out. --Kels 14:46, 12 May 2009 (UTC)

Round Seven

the question is, what is required to "faithfully follow the scientific method." from what I hear around hear, hyptheses need not be tested to constitute science. I think that's why y'all believe so much pseudoscience. Ungtss 01:37, 10 May 2009 (UTC)

Wow. And you wonder why people call you "science hater". --Kels 01:45, 10 May 2009 (UTC)

because I think hypotheses must be tested? Yeah, that is pretty unreasonable, isn't it. Ungtss 01:52, 10 May 2009 (UTC)

Hey, if that's what you've decided to see, who am I to stop you? --Kels 02:44, 10 May 2009 (UTC)

So what "pseudoscience" do "we all" "believe", oh master of all that is science? ħuman 03:01, 10 May 2009 (UTC)

Now why in hell would I waste my time telling you which of your beliefs I think are pseudoscience, when you don't think the scientific method requires that hypotheses be tested? Ungtss 03:14, 10 May 2009 (UTC)

When did I ever say that? ħuman 03:21, 10 May 2009 (UTC)

psst! Nobody but Ungtss actually said that. --Kels 03:23, 10 May 2009 (UTC)

This is what you said that I interpretted as meaning that experiment was not necessary for science: "I would comment on "4". Sometimes just collecting "more" data is essentially the same as "experiment [with controls]" - like the observation of gravitational lensing that strengthened the ole relativity thing way back when. (Or, when archeologists dig up more cool stuff and clarify what historians claim, for better or for worse.) ħuman 04:54, 6 May 2009 (UTC)"

My comment: Ah, yes. Sometimes you don't need to do an experiment with the controls necessary to test a hypothesis. Just find more data, and if you can interpret the data as consistent with your pre-existing belief, then you have confirmation bias science. Your first example, of course, doesn't support your thesis, as Eddington's observations were part of a highly controlled experiment deliberately designed to test the hypothesis of relativity. Your second example is so vague that there's not much I can do with it. But the initial statement (sometimes more data is the same as an experiment) means that sometimes you can have science without experiment. Ungtss 10:14, 10 May 2009 (UTC)

Collecting data for scientific analysis is not limited to designed experiments. Designed experiments are, in general, better able to reduce error and bias in their subsequent models, but this is not always possible. This does not mean that statistical analysis cannot be performed on observations made outside of controlled experiments, as their are plenty of mathematical tools designed to do exactly this. The results may have a larger marger of error than a controlled experiment, but that does not make it invalid, and I think this is the point you don't get. Scientists are able to quantify how good a given hypothesis is based on the available data, but it is extremely rare, if not outright impossible, that any hypothesis can be proven with 100% certainty. Therefore I can understand being skeptical of an assertion that is only, for example 80% certain.

OK, now try a thought experiment: You and 20 other people are in a burning building and you have to escape or you will die. There are two doors that lead to an exit. 10 people choose to exit through each door and you happen to be the last. Of the 10 who choose door #1, three of them are shot and killed by a sniper on the way out. Of the 10 who choose door #2, eight of them are shot and killed on the way out. Through which door do you try to escape? Jorge _{Gave real $$ to support RW} 16:44, 10 May 2009 (UTC)

In my mind, there's a very important difference between "Science" and "Reason." Science is what comes from the scientific method. Reason is what comes from evidence, logic, and proper philosophical presumptions and values. Science is a subset of reason, but there are things that are reasonable and yet not "scientific." Your thought experiment is a primo example of using reason. But it's not science, because there's no scientific method involved. In my opinion, scientists are way the fuck out of their lane when they start calling their "reasonable conclusions" science. Because reason depends on a lot of subjecting considerations, including values. Which door will I pick? Well, if I am self-interested, I'll pick door 1. If I'm an altruist, I'll pick door 2 to give another person a better chance at survival. If I'm McGyver, I'll build a Kevlar vest out of aluminum foil, and beat the bullet. Science can't tell me what door to pick. Only reason, stemming from my values, can decide that.

But science fanboys, unfortunately, don't seem able to distinguish between the two. Consider Eugenics. Hundreds of "scientists" running around claiming it was the "Scientific" way to approach reproduction. Hell no it wasn't science. It was an argument from reason -- dependent on an enormous number of philosophical and value assumptions. A lot of people punted to the scientists' judgment -- but the scientists weren't doing science -- they were out of their lane, trying to be the supergenius managers of society.

I despise that tendency, because it cheapens science. And I love science. Science should not be associated with anything except what it is -- observation, experiment, and analysis. Science tells us nothing about what we should do. It tells us nothing about how we should interpret evidence. It tells us only that when we control for variables in our material world, we can observe, and analyze, the results. Ungtss 18:08, 10 May 2009 (UTC)

"I love science. So long as the bitch does what I tell her." --Kels 18:26, 10 May 2009 (UTC)

I despise that tendency, because it cheapens science. And I love science. Science should not be associated with anything except what it is -- observation, experiment, and analysis. Science tells us nothing about what we should do. It tells us nothing about how we should interpret evidence. It tells us only that when we control for variables in our material world, we can observe, and analyze, the results. Ungtss 18:08, 10 May 2009 (UTC)

So tell me then, if a scientific hypothesis is known to be 80% certain, why does that make it "bad" science? Everyone agrees it's not perfect, the whole point of statistics is to account for error in scientific models. If all your doing is arguing semantics then when we enter the realm of making real life decisions based on those results, imperfect as they may be, we are merely being 'reasonable' and no longer purely scientific. Jorge _{is still sorry for whatever he did} 18:37, 10 May 2009 (UTC)

It's not just semantics -- it's a big deal. Science is as close to objective as possible, and as such carries a lot of credibility. Reason is subjective in many respects, and not nearly as reliable as science. Giving untested, untestable ideas (like eugenics) the credibility of the tested ideas of science is dangerous, because it leads to premature acceptance of untested ideas, premature rejection of alternative untested ideas, and the institutionalization of confirmation bias. As to whether "80%" is good science, let's take it down to some concretes. "This airplane doesn't crash 80% of the time." "This drug kills 20% of the time." "80% of mammals have placental reproductive systems." "This glue adheres 80% of the time." All half-ass. Even "This drug heals 80% of patients" is still pretty half-ass. What's different about the other 20%, and what is the drug doing to them? The 80% answer doesn't tell us. Ungtss 21:11, 10 May 2009 (UTC)

Ungtss, you are right. It's an argument from reason, not yet science. But let's change the scenario a bit. You are in a building, but it's not burning, and for some strange reason, you have an unlimited amount of people waiting to go through the two doors. You do eventually want to get out, as the food sucks and you miss your family. So, you start sending people out each door and taking data. Scenario 1 is really easy. You send 20 people out and figure out that there is a pattern, shoot,shoot,shoot,safe,shoot that repeats. You make your hypothesis and you make your conclusions that for you to get out, teh best time is number 4, in the series of 5. This is science, yet it still tells you "what to do", and gives you a "framework" for actions and behaviors. Scenario two, then is a bit harder, cause there is no clear pattern, but you have managed to figure out that 4 of 5 people die going through door one, and 3 of 7 people die going through door two. You test to make sure those numbers are consistent. It's STILL science, you still have a theory (4 of 5 in door a; 3 of 7 in door b), but you did not have a precise pattern. And you are *still* able to make choices for your behavior based on what you know. (edit con)--En attendant Godot_{"«Poetry involves the mysteries of the irrational perceived through rational words. V.Nabokov»} 18:41, 10 May 2009 (UTC)

I'd argue that you do have science, but it's pretty half-ass. Why are they less likely to shoot #4 in the series? Does he have five rounds in their gun, and so are delayed when they have to change clips? Or is it 5 shooters taking turns, and the 4th shooter is a terrible shot? Stats alone don't tell us. But if it's true, then everybody should run out at once at that point in the cycle, before he can reload. Stats alone don't get you to the heart of the issue -- what is ACTUALLY HAPPENING? They are, essentially, a way of quantifying the degree of our ignorance. But I'm not interested in quantifying it so much as eliminating it by investigating the problem further. Same with the second scenario. Why the difference? What's different about the terrain or the shooters? Until we know what's actually causing the difference, we don't know whether or not we can exploit that information. Ungtss 21:11, 10 May 2009 (UTC)

Yes, science is sometimes just 'quantifying the degree of our ignorance'. For example, I could pose a hypothesis that "Ungtss is ignorant of what science actually is". I could use the arguments you have made in this debate as my data. Although you occasionally manage to string together a coherent idea, most of what you say is utter nonsense. I would therefore conclude that there is a 99% probability that you are entirely full of shit. Jorge _{is still sorry for whatever he did} 22:15, 10 May 2009 (UTC)

Well, if you wanted to apply and communicate using reason, you would first have to evaluate my statements, determine if and why they are incorrect, and then articulate your reasons for rejecting them prior to stating your conclusion that I am full of shit. But that would, of course, require thinking. Ungtss 22:38, 10 May 2009 (UTC)

My hypothesis strengthens! Science WIN! Jorge _{noes teh maths} 22:40, 10 May 2009 (UTC)

"Your demand for an explanation of my argument is further evidence that you are full of shit!" Very nice. Ungtss 22:44, 10 May 2009 (UTC)

Ungtss, your argument wasn't about the quality of the science, but if it was science or not. It was, and you agreed. You want to find out "why", but science is by the very reality of life, far more practicle than that. If you need to get out of the house (or cure cancer, or stop a forest fire from jumping a boundary) then science is what you turn to. Before you worry about "why", you may need to know "how", so you can fix the here and now. We might not know what penicillin does yet, but we tested to find out that it really does kill illness. That is science at its core. "pure science" for science's sake is unrealistic and virtually wn-useful. rather than randomly finding out facts, most scientists (i'd argue) have a goal - like curing an illness, making a drought resistant wheat, or making a tool that predicts earthquakes and volcanic activity. to do otherwise is to live in your head and never see the real suffering in teh world.--En attendant Godot_{"«Poetry involves the mysteries of the irrational perceived through rational words. V.Nabokov»} 23:51, 10 May 2009 (UTC)

There are two parts to my argument -- 1) that untested hypotheses are not science, and 2) that settling for mixed statistical results in science is a fairly sorry-ass approach to scientific problems. In order to effectively solve problems, we must pursue the "why." Settling for the knowledge that something works even though we don't know why is also a fairly sorry-ass approach to science, in my opinion. Ungtss 00:40, 11 May 2009 (UTC)

Ah the powerful "why" question. I have been involved in a lot of discussion over this lately. There is a wonderful little statistical approach called Independent Component Analysis that is gaining ever increasing popularity because of its ability to pull apart components of a signal. I run into it because of its use in fMRI and meg/eeg neuroimaging. ICA is a great example of "model free" statistical techniques that discover "interesting" components in a data set without the need of any kind of a priori knowledge or modeling. A typical ICA analysis will pull out many dozens of "interesting" signals. The question is what to do with those signals? This is data mining in its purest form. Pulling out information from data without applying any hypotheses or tests. Some people are content with this, it allows for all kinds of interesting applications in categorizing and recognition of patterns. But if you want to know why those signals are interesting, what is creating them, and what do they mean you have to move one step further into modeling and hypothesis testing. If we think one particular signal is caused by one particular brain region responding to a particular stimuli we can create a model of what that signal would like like and see if we can match it with one of the components pulled out by the ICA. The ICA data mining provides us the information we need for hypothesis testing, while the hypotheses themselves, and the falsifiable claims are derived from model based reasoning and predictions. All of these components are needed for science to be done. The same thing can be applied to any of the "historical" sciences. Fossil strata, genetic comparisons, biogeographic distributions, etc. are all sources of data whose signals can be mined for "interesting" patterns. We can then take "why" questions and develop models that can answer these questions, derive predictions from these models and compare them to the signals from the data mining. I am sympathetic your idea that real science requires exploration into the "why" and that hypothesis testing is at the core. But I am confused how any of this precludes basic historical sciences and analysis. tmtoulouse 00:58, 11 May 2009 (UTC)

I agree with you, and I'm not saying that statistical analysis precludes good science. I'm saying that resting our hats on it is not good science. Stats are a good way to look for clues, but clues are not good science until the relevant variables are isolated and tested. I gave an example in the next subsection -- the doctor's decision to "diagnose" my baby with "lateral cerebral ventriculo-megaly" on a purely statistical basis (3 SD above the mean) without any papers showing any problems for well well above her size. Stats might give you a clue that size is associated with a problem, but relying on that (rather than testing it) is shitty science in my opinion. Especially when you use that shitty science to scare a pregnant woman out of her skull for a couple days. Ungtss 11:49, 11 May 2009 (UTC)

7.1

I must say, the ability to go through all of the above, including some very thoughtful descriptions by some very intelligent and well-spoken individuals, some of whom make their living doing hard science, and still default to the highly restrictive and inaccurate definition you started with is...well, let's go with "stunning". --Kels 23:07, 10 May 2009 (UTC)

Got it. My new criteria for truth will be "Things said by people who Kels thinks are smart, well spoken, scientists." We'll just have to add that step to the scientific method. Ungtss 00:40, 11 May 2009 (UTC)

I'm glad one of us learned something about the nature of science. Unfortunately, I think it was me. --Kels 00:48, 11 May 2009 (UTC)

Just a side bar: The scientific method is rarely a linear process. And why should it be? Sterile bat 01:43, 11 May 2009 (UTC) I also still don't get this 80-20 thing. How would you know before you did the experiment? And if those are your results, why would you throw them out or degrade them? You still know more about the natural world than you did before. (Disclaimer: I haven't read the last umpteen screens.) Sterile bat 01:45, 11 May 2009 (UTC) And, how are you going to know what uber-cool experiment to do or what hypothesis to explore or how to adjust your instrument if you don't have any preliminary results to follow up on? Sterile bat 10:51, 11 May 2009 (UTC)

I agree with you that it's not linear. I agree with you that you can't know (prior to an experiment) whether you'll get a purely statistical result. I also agree that statistical results can lead you to uber-cool experiments that truly solve the problem. All I'm saying is that considering purely statistical results alone to be high quality science is silly. I'll give you an example. My wife came back to the states a few months ago, because she didn't want to take the 26 hour flight alone and super pregnant. When she got here, she had an ultrasound, and the doctor noticed that our baby's cerebral atrial ventricle (a little cavity in the middle of the brain) was 10.2 mm across. That put it 3 standard deviations above the norm. According to standard medical practice these days, if the CAV is over 10mm, the child is diagnosed with "lateral cerebral ventriculo-megaly" -- a scary way of saying "This cavity in the brain is really big." This "diagnosis" is based on a few studies that show that extremely large atrial ventricles are associated with brain disorders or viral infections. But there are no studies showing any such problems between 10mm-12mm. They just pick 10 for their diagnosis cutoff because it's "really large." But are there other reasons it might be large? Why yes -- for one thing, her dad is 6'8" and has a big-ass head. For another thing, babies are constantly changing size in utero. But this half-ass, pure stats approach to science "diagnoses" her with a "condition" because one part of her brain is 3 standard deviations from the norm, even though there's no evidence of any problems at that cutoff, or even for much larger ones -- such as 12 mm. In other words, they're using "statistics" (she's bigger than most kids and that's weird) as a substitute for "science" (atrial ventricals of this side are known to cause X problems). So they "diagnose" our kid with this "condition" to my pregnant wife, who's all alone in the hospital, thinking her baby has brain disfunction. That is not only half-ass science. It's plain shitty. Ungtss 11:36, 11 May 2009 (UTC)

Now you're the one confusing science with speculation. Science only say your 'baby has a big head' and that it 'might cause problems' because there is a statistical correlation between big heads and viral/brain disorders. That's where it ends. You and your doctors can choose to use that information or not. I agree that no one should be forced to make decisions based on such uncertainties. Science does, however, give us the ability to make rational decisions based on the available evidence. Also, because one can use statistics to analyze virtually any set of observations, I think it is more apt to say any science that doesn't use statistical analysis is silly. Don't get me wrong, statistics can be misused and misinterpreted by people--scientists included--who don't fully understand it. Don't make the same mistake. Jorge _{Radix Malorum Est Superbia} 15:40, 11 May 2009 (UTC)

Science can say much more than that. That's why this statistical correlation is shitty. What is the common denominator between "big cerebral atrial ventricles" and certain brain problems? The two are associated, but obviously not causally. But GOOD science isolates causes. A good paper would say "X condition causes the cerebral ventricle to expand by Y mechanism, and also causes A brain disfunction by B mechanism. If you understand the causation, then you're doing good science and can develop a good treatment. Until you isolate the cause of the correlation through experiment, all you can say it "Well, people with big ventricles are more likely to have problems." How fucking worthless is that shit? Ungtss 15:50, 11 May 2009 (UTC)

What's your point? You can't look for a causation until you have some sort of association, some sort of temporal component of before and after, and some way to isolate the cause. (Thanks David Hume.) To think that one can come up with one experiment without any preliminary data to do all of these things is utterly absurd. To give up at looking at phenomena because you can't find the perfect experiment is also absurd. Sterile bat 16:08, 11 May 2009 (UTC)

My point was to respond to Jorge. He said, in part, "Science only say your 'baby has a big head' and that it 'might cause problems' because there is a statistical correlation between big heads and viral/brain disorders. That's where it ends." Bullshit. That's where it begins. Once you notice a correlation, you investigate for causes using the scientific method. But diagnosing people with a "brain condition" with "potential hazards" based purely on divergence from the mean is bullshit. But that's what they do, all the time. Ungtss 16:21, 11 May 2009 (UTC)

I refer you to my burning building analogy. Sometimes rational people make choices based on incomplete information. Science doesn't always produce all the answers immediately, and often when making important decisions we don't have the luxury of waiting for a verdict. Take it for what it is, but calling it junk science is like saying a movie sucked doesn't have a plot after only watching the first fifteen minutes. Jorge _{as if my opinion counts} 17:03, 11 May 2009 (UTC)

For the umpteenth time, dude. I'm saying it's bad to be satisfied after the first step. Not judging the movie after the first fifteen minutes. It's like RELEASING the movie after COMPLETING only the first 15 minutes. Diagnosing a "brain condition" based on size alone. Killing every pig in Egypt over a flu never observed in pigs. Calling abiogenesis "science" without testing the hypothesis. If you have to act based on incomplete information, fine, take your best guess and go. But at least recognize that you're making a decision based on half-baked evaluation of the evidence, rather than a "reliable scientific conclusion." And don't call people "unscientific" when the question your untested, unscientific, half-ass conclusion. Ungtss 17:08, 11 May 2009 (UTC)

Who, exactly, are these researchers who just say "eh, good enough" and stop researching? Do they actually exist? You keep talking about them a lot, so I figure you must have some names or something.

You're the one saying that scientific works-in-progress are "unscientific". Analyzing incomplete data does not make the analysis "half-baked". Its entirely logical to say "though this pastry is half-baked it will likely become a cake when fully baked"-this is not the same as saying "it's a cake". Analysis is hypothesis testing and is part of the scientific process. Jorge _{might get around to it} 17:23, 11 May 2009 (UTC)

Would you hire a baker who handed you half-baked pastries saying, "These will likely become fully-baked at some point?" Or a director who released half a movie, saying "This will likely become completed by someone else?" That's how I feel about a scientist who finds a bald correlation and releases a half-baked paper. Whoopdy doo. Yeah, maybe somebody will write a better one in the future. Doesn't make his incomplete one any better. Ungtss 17:28, 11 May 2009 (UTC)

You know, I rather like that whole "baking a cake" analogy. You apparently think a scientist releasing preliminary findings is like a baker mixing up the batter, then serving it and claiming it's a cake and maybe someone else will finish the job. The reality, of course, is the scientist/baker mixes the batter, says "yup, looks like a cake in the works, let's see how this turns out", then popping it in the oven. It comes out, the scientist/baker is pleased and says "yup, it's a cake", then passes it on to someone who specializes in decoration to finish the job. And we have a team effort that arrives, after some status reports along the way, at a complete, delicious cake. And dammit, now I want a slice of science. --Kels 17:39, 11 May 2009 (UTC)

And just because you're not satisfied doesn't make it bad science. If scientists were satisfied after the first step, they wouldn't get much done. Science is an ongoing process. Jorge _{Member of Vandal/Parodist Website} 17:32, 11 May 2009 (UTC)

What's wrong with a chef telling you about a cake he's baking? Or a director telling you about a movie he's working on? And where are these nefarious scientific straw men who claim their half-baked results mean more than they do? Most scientific articles I've read are abundantly clear as to the limitations of their results and almost always suggest further research before a definitive conclusion can be reached. Kindly give me a specific example. Jorge _{noes teh maths} 17:37, 11 May 2009 (UTC)

A little addition, and may be one things that's confusing our "science-loving" friend, is that science reporting in the popular press can be pretty awful, and often neglects to mention the limitations that are in the original papers. This leads to things like Egyptian pigs dying, your doctor who probably isn't a researcher making a diagnosis, and the waiter after having been told "the batter is mixed" turning around and saying "the cake is ready now". --Kels 17:42, 11 May 2009 (UTC)

Here's an example I would like to see you discuss with Ungtss. I can hardly read this, but Ungtss apparently has a complete grasp of it and it's shortcomings. Please discuss. ^{— Unsigned, by: Neveruse513 / talk / contribs} 17:49, 11 May 2009 (UTC)

Well, let's define our criteria for an adequate paper. To me, a paper must follow the scientific method all the way through with respect to an interesting question. Interesting question but you don't test your hypothesis = half ass. Uninteresting question for which you test your hypothesis = half-ass. Uninteresting question for which you don't test your hypothesis = whole ass. That's my opinion. You may say that "reporting on progress" makes for an adequate paper. Can't really argue, except to say it's a much lower threshold. What can I say? I think a paper should answer a meaningful question. I think papers that don't do so are half-ass. Ungtss 19:26, 11 May 2009 (UTC)

While I'm not saying that it's worth justifying scaring your wife about something at 3 SD's above the mean, it IS remarkable, like in the 99th percentile remarkable. Sterile bat 18:59, 11 May 2009 (UTC)

Agreed. Remarkable, but remarkable ≠ diagnosable. I'm in the 99th percentile for height, but that doesn't make it a "condition." Is extreme height associated with health problems? Sure -- but only if it's caused by something like a malfunctioning pituitary, in which case you're like 8 ft. tall. Same thing here. Somebody has to be in the 99th percentile. Doesn't mean there's anything wrong with it. Scientists writing the papers on the topic of atrial ventricles don't seem to understand that. Ungtss 19:26, 11 May 2009 (UTC)

Let me ask another question. If there were something wrong with your baby, and the doctor didn't investigate it and discover the 189th paper in the 1973 journal that explained the defect, would you be upset? I'm not saying there is or that I would ever wish there to be, just that there due-diligence might be a good thing. Also, have you considered that maybe this is just a doctor with a bad bedside manner, but still is something worth looking into? That is, have you considered a case for which you are detached from the situation? Again, I'm not trying to upset you, but sometimes we are too close to situations. Sterile bat 23:04, 11 May 2009 (UTC)

Good question. Obviously it's hypothetical, but my instinct is that I would only be upset if the paper demonstrated a basis for causation. In other words, if it were shown that size caused a problem, that would be one thing. But these papers only show that size is correlated with a problem, and provide no actual causation for the problem. It appears that some unknown cause results in both brain problems and an enlarged cerebral atrial ventricle. But there's no evidence that ventricle size causes any problems, nor that ventricle size is caused only (or even usually) by an independent problem. But none of the papers have investigated that. So no, I don't think I would be upset, because a) the "diagnosis" allows for no treatment as it determines no cause of the problem which can be treated, b) the "diagnosis" is based purely on size, rather than any correlation with actual problems, and c) a man can only know so much. Ungtss 03:35, 12 May 2009 (UTC)

Why?

Indeed. I thought that question was one that science has nothing to do with. How, what, when, where - seem more to me to be scientifically investigatable. Take gravity - how might be theories about particle exchanges, or curvature of space. What is the simpler description embodied in the m1/m2/r squared equation. When, is pretty much always in this case, and where is everywhere. Why? Why is there gravity? Because god doesn't like all her stuff floating around in free fall? If it is found to be inherent in the nature of the universe, then why is that? Why, indeed? ħuman 03:34, 11 May 2009 (UTC)

Why does the moon have phases? (Because it revolves around the earth). Why do the light and sound of lightning arrive at different times? (Because the speed of light and speed of sound are different?) Why do people die of the flu (Because their immune systems are not strong enough to fight it off). So now not only does science not have to experiment, but it also can't ask why? You're killing me, Smalls. Ungtss 11:42, 11 May 2009 (UTC)

Personally, I think you are confusing "why" and "how". Why has an inherent aspect of meaning. Why did he did (vs., how did he did). "Why is there life" vs. "What or how is life". "how does the Earth's gravity effect the moon" vs., (why is there interaction). In religious Studies & philo, we say that philo/religion's realm has been the why, and science has been the "how". --En attendant Godot_{"«Poetry involves the mysteries of the irrational perceived through rational words. V.Nabokov»} 15:51, 11 May 2009 (UTC)

I think there are different types of "why" questions. Certainly there are philosophical "why" questions. But I'd argue there are also scientific ones. Isn't it fair to look at an interesting observation (like the phases of the moon), ask why it occurs, and answer it scientifically (with reference to the revolution of the moon around the Earth and the moon as reflecting, rather than producing its own) light? How is that a "How" question? Ungtss 16:02, 11 May 2009 (UTC)

Presumably because it's "by what mechanism" rather than "for what purpose". Chalk it up to English being imprecise. --Kels 16:48, 11 May 2009 (UTC)

Fair enough. Ungtss 17:09, 11 May 2009 (UTC)

U., yes, your answers using "how" "why" [corrected dumb mistype] are grammatically correct and reasonable examples, however, the "whys" extend forever (why does the moon orbit the earth? why is there gravity? etc.). Also, I think Kels is right, in that those questions can be rephrased as "how" questions - "how is it that the moon has phases?" Anyway, with minor quibble put to rest, we continue... ħuman 19:25, 11 May 2009 (UTC)

Your point again?

Ungtss, is your point merely that statistical reasoning is not the be all and end all of science? I do not think there are many that would disagree with that. You seem to be suggesting though that there is some larger meaning in all this that leads "us" to believe in pseudoscience. Can you point to an example of a pseudoscience you believe most of us are enthralled with? And why our belief in its veracity is related to a misapplication of science? tmtoulouse 17:45, 11 May 2009 (UTC)

Evolution of antifreeze glycoprotein gene from a trypsinogen gene in Antarctic notothenioid fish(link). Ungtss will now enlighten us as to how these scientists have deceived us through their half-assedness. Jorge _{believes only in the Primal Oneness, the Immortal Tao} 18:00, 11 May 2009 (UTC)

“”These scientists observed that the two genes in question were fairly similar in their content. They then calculated what shifts, duplications, and deletions would turn the first gene into the second gene. Then (without testing whether this could happen, whether there would be any survival advantage each step of the way, much less whether it did happen, historically), their discussion concludes that it "could have happened." The flaw in the paper is not in the nitty gritty details -- it's in the conceptual approach itself. They invent a hypothesis, but they don't test it. Is that really science to you? A real experiment could use genetic engineering to duplicate the alleged evolution. Make the first switch, and see what happens. Make the second switch, see what happens, etc. See if there's even a feasible path from A to B. It wouldn't show that it did happen, but at least it would show that it could have happened. This sorry paper doesn't do anything but speculate with flashy jargon, pictures, and machines.

Ungtss 06:17, 7 May 2009 (UTC)

Who are you quoting, exactly? --Kels 19:18, 11 May 2009 (UTC)

He's quoting me. TMToulouse: that's all I'm saying regarding stats. I'm shocked that the argument has gone on so long -- I thought it was self-evident. My argument regarding pseudoscience stems from papers which fail to test their hypotheses (like the one referenced by Jorge, above). But that's a different issue than this silly thing about stats. Ungtss 19:29, 11 May 2009 (UTC)

Ah, the way it came out made it look like it was you quoting yourself, which seemed bizarre. --Kels 19:42, 11 May 2009 (UTC)

That would be my fault. Jorge _{as if my opinion counts} 20:05, 11 May 2009 (UTC)

What were their postdictions? Were they confirmed by experimentation/observation? ^{— Unsigned, by: Neveruse513 / talk / contribs} 19:40, 11 May 2009 (UTC)

I am a big fan of Bayesian inference as a good paradigm to explore what science is, and how we go about answering questions in a methodical proccess. You start with a given hypothesis say H1 that you think is the correct answer to a given question. There are then a range of alternative hypotheses H2...Hn. What we want to know is what are the relative probabilities of each of the hypotheses given what data we have p(Hj | D). Any observation can be used to improve our probability estimates. Say I have a given gene, and I have two hypotheses I am working with. The first says that the given gene is a mutated version of gene A the second hypothesis says that it is a mutated version of gene B. If I analyze the sequence and find that the only difference between the gene I am interested in and gene A is that a 'c' has been inserted in the middle, while at the same time showing that there is almost no overlap or similarity between gene B and my gene of interest I can adjust the relative probabilities for the correctness of the two hypotheses given this observational data. How have I not just performed science? tmtoulouse 20:59, 11 May 2009 (UTC)

Well, what hypothesis have we tested? We've tested the "Which of the 1-n hypotheses is most similar to the gene in question and therefore most probable, assuming descent" hypothesis. But we haven't tested the "Did this happen?" hypothesis, or the "Could any of stepwise mutants have survived" hypothesis, or the "Was this mutation so advantageous in the niche that it was able to displace all other alternates" hypothesis. Similarity is not a particularly interesting hypothesis to me. I'm more interested in whether it could have happened, and whether it did. And those interesting hypotheses are not tested by your method. Ungtss 22:23, 11 May 2009 (UTC)

Is this a "no one's ever observed evolution so it must not be true argument"? Because I'd like to pwn that right now. [User:Sterile|Sterile]] bat 21:47, 11 May 2009 (UTC)

Hard to tell, but I think it has something to do with cake. --Kels 21:51, 11 May 2009 (UTC)

The cake is a lie. Z3ro^talk 21:52, 11 May 2009 (UTC)

In Trent's example above, "descent" might not be an assumption, it might be an observation (IE, we might know critters with the "new" gene descended from critters without it, that have A and B (and etc.). ħuman 22:48, 11 May 2009 (UTC)

Just a question, Ungtss, if common descent is to be falsified, what is the observation that falsifies it? (Sorry, that goes back, like, a bit.) Sterile bat 22:55, 11 May 2009 (UTC)

I don't think a scientific test for it is yet possible. One might interpret the evidence in favor of it, but also against it, depending on one's views of the world. Ungtss 23:00, 11 May 2009 (UTC)

What if a genetic sequence were to emerge of two supposedly related organisms (by homology or geographic relationship) that couldn't be explained by the common mutation mechanisms and rate—they were just utterly different. Would that not falsify evolution? Sterile bat 23:07, 11 May 2009 (UTC)

That's not a test. That's a "well we might find it some day." One might as well say that the existence of God is falsifiable because he might show up in the sky someday. Ungtss 23:09, 11 May 2009 (UTC)

Then the existence of atoms are not falsifiable, then, by your logic? Sterile bat 23:14, 11 May 2009 (UTC)

I don't see the analogy. Can you elaborate? Ungtss 23:20, 11 May 2009 (UTC)

Most of the falsification of atoms is of the "well we might find it some day" type. Brownian motion, nuclear decay, etc. are consistent with atoms, but none was designed as a "test" of the existence of atoms. They were explained as post-observational evidence of atoms. Sterile bat 23:23, 11 May 2009 (UTC)

How are those instances of "we might find it someday" evidence like "we might find a bizarro alternative lifeform" or "we might find god?" Ungtss 23:31, 11 May 2009 (UTC)

(Howdy, doc!) What non-atomic theory to explain matter do you propose? (Phlogistonists unite!) Atoms are clearly in your realm of science that is falsifiable. Sterile bat 23:39, 11 May 2009 (UTC)

I'm still trying to understand your analogy between "we might find a bizarro life form someday to falsify common descent" and "brownian motion implies the existence of atoms." Ungtss 00:43, 12 May 2009 (UTC)

(How many colons to use? Conundrum!) You seem to imply that the existence of atoms is tested-and-validated hypothesis and therefore, by your logic, falsifiable. So there must be some experiment that could show that atoms do not exist. My point is that is it difficult to come up with a non-"bizarro" experiment that would falsify atoms, yet they were not accepted until the early 1900s through a lot of indirect detection in the early 1900s and certainly with no direct detection. Sterile bat 01:59, 12 May 2009 (UTC)

Got it. Here's what I think about atoms -- we know what matter is composed of to the extent to which we have experiments that provide information about matter. Brownian motion, electromagnetism, atom-splitting and atom-smashing, and atomic weights all provide a decent, though imperfect, picture of what matter is made of. It appears to be made something something that sometimes acts like a particle, sometimes like a wave, wiggles, carries charge, and increases in size and energy in definite quanta. On the other hand, the nuclear model of the atom (with protons and neutrons in the middle) is, in my opinion, fairly weak, as it requires the ad hoc introduction of a force to hold the repulsive protons together. There are other models for the atom that resolve this problem. The short answer, then, is that "We don't know that matter is made of atoms. The atomic model is a model we've developed to explain what we know, and it does so, except insofar as relativity and quantum physics contradict at the atomic level, leading to the drive for a "theory of everything" which, I suspect, will require a radical overhaul of our concept of matter, and potentially even a modification or rejection of our present model of the atom. Ungtss 13:13, 12 May 2009 (UTC)

A couple of quick points, after catching up on the thread. First, we need a real philosopher. I say this because Untgss, who is being very polite and is giving us an interesting conversation, is all kinds of wrong. There is not a goalpost he can't move. If he keeps redefining science and falisfiability, they will completely lose their meaning. What we have here is a system, which we call science, or sometimes methodological naturalism. This system is, so far, the best one we have to explain the universe we inhabit. It is both good for explanation, and good for prediction. It also refuses to overstep its bounds. "Science" (whomever that may be) will never claim that god(s) cannot or do not exist---only that they are unlikey to exist based on current evidence. This is where the beauty of bayes comes in. In medical science in particular, the prior probability (or "plausibility") of something is critical. For example, if I were to posit to explanations for myocardial infarction, one that it is caused by acute coronary artery occlusion, and another that it is caused by a melancholic nature, I know ahead of time that, based on the physical laws as we know them, the latter is unlikely to be true. This doesn't make it impossible, just very, very improbable. Of course, further testing is likely to increase or decrease the likelihood of one or the other hypotheses. This whole plausibility issue is important (and limiting) in that we don't need a false equality of considering every idea to be a priori equally valid. This of course bears on the ID/evolution thing. One of these ideas conforms to our understanding of the universe and makes useful predictions. One could easily be pulled out of my ass.-- -PalMD ^{--goat--the other white meat} 23:35, 11 May 2009 (UTC)

If you'd read the discussion above in greater detail and wished to make a comment that effectively addressed my argument, you would have addressed the distinction I'm trying to draw between "science" and "reason." Science being ideas which stem from experimental testing of hypotheses, and Reason being the evaluation of the evidence to determine what is most likely or more desirable. Science being a subset of Reason. The significance of this difference lies in the different epistemological standards of proof in the two realms of thought. Science carries a great deal of credibility, based on its unique and powerful methods of exploring the world. Reason is also powerful, but not nearly as reliable as the scientific method itself. Reason depends on one's view of what things are more likely and less likely -- on one's biases and emotional predisposition and value judgments. Science depends on none of these. Many things that are actually "reason" have been called "science" in the past, with disastrous results. Most notably eugenics. When you weed out "melancholy" as a cause of myocardial infarction, you have two ways of doing it. You can do it based on reason ("Based on the physical laws as you know them") or you can do it based on Science ("For starters, see if there's any correlation between mechancholy and the condition -- if you find one, investigate further into the potential causes."). Unfortunately, you and many others appear to be conflating the two means of knowing. But the danger comes in that your "reason" is subject to your philosophical biases -- and does not deserve the credibility inherent in "science." Ungtss 00:43, 12 May 2009 (UTC)

I think you are creating a razor's edge where there is none. I don't think you can so easily create separate realms of "science" and "reason", or worse, nest them. Sure, in a purely ethereal realm you can count angels on a pinhead all day, but I'm not sure how this advances our understanding of knowledge. "Reason" is, as you say, rather fallible. Science certainly is as well, but I'm not so sure it's as easy to put things under each's column as you say it is.-- -PalMD ^{--goat--the other white meat} 00:59, 12 May 2009 (UTC)

I'd agree that it's not easy to categorize, but I also think it is important to recognize the difference when it's clear. Ungtss 01:49, 12 May 2009 (UTC)

Another issue is the conflation here. I don't think that "reason" and "science" are as connected as you think (as concepts). It is a reasonable assumption that in order to develop a technological society, people must utilize the scientific method. The method itself is quite sound. Human reason is of course flawed inherently. It doesn't arrive at wrong answers by malfunctioning; it arrives at wrong answers by working flawlessly. I think where your (cogent) analysis breaks down is in your (not obviously) arbitrary assignment of ideas to "reason" or "science". Reason is essentially a form of assertion, but these assertions can be supported by the scientific method, and once validated, are no longer purely "reason".-- -PalMD ^{--goat--the other white meat} 02:05, 12 May 2009 (UTC)

I don't think the categorization is arbitrary. I think ideas are "science" when they are arrived at through the scientific method -- i.e. hypothesis tested through experiment. But not everything can be tested that way. For instance, "Were John and Abigail Adams married?" you have to use sources like history, judge their credibility, and infer an answer you find most reasonable. You can't test it with a repeatable experiment like you could test the gravitational constant. Those things, I think, constitute "reason." Not always a razor's edge, but I think generally doable. Can you test it through a repeatable experiment? If so, science. If not, not science. May still be true, but not science. Ungtss 02:15, 12 May 2009 (UTC)

Doesn't this bring you exactly back to the position you started this whole mess in? --Kels 02:33, 12 May 2009 (UTC)

Interesting. Of course, you are trying to set up the, "any knowledge that isn't (hard) science is equally valid/invalid" which is patently absurd. My "reason" may tell me that monkeys will fly out of my ass, or it may tell me that Thomas Jefferson was the primary author of the Declaration of Independence. Only one of these acts of reason has evidence to support it. The fact that you can't heat it with a bunson burner does not make it invalid, an certainly does not make the two assertions equal.-- -PalMD ^{--goat--the other white meat} 02:52, 12 May 2009 (UTC)

I am setting up no such thing. Of course some things are more reasonable than others, based on the rules of reason. Rules of evidence -- rules of philosophy -- knowledge of how the world works. Unless your ass is rather unusual or you've been very unkind to some monkeys, reason can tell me it's unlikely to happen. I'm not saying all things that are not science are equal. That's bullshit. I'm saying that reason and science operate by different rules. Science is experimental in nature. Reason is broader than that. It deals in likelihoods and probabilities and goals. And we use our reason all the time to make judgments about the world, because science can only answer a limited scope of questions, as experiment is not always possible. Ungtss 03:24, 12 May 2009 (UTC)

Scientific experiments inevitably deal with "probabilities and likelihoods". To say otherwise is to wholly misconstrue what an experiment is. What you are saying is false. Jorge _{knows you has a sock on CP} 03:36, 12 May 2009 (UTC)

Science is evidentiary by nature. Experiments are one kind of (often very good) evidence. So are many other observations. ħuman 04:18, 12 May 2009 (UTC)