|
Post by Joe Durnavich on Jun 14, 2006 23:14:53 GMT -4
The Difference between the scientist and the CT, well for me, would be that the Scientist looks at the data he has and then attempts to position the line so that it reflects the data as accurately as possible, only removing the obviously out of place data points after that.I sometimes envision the process this way, too, but in the end it I find that it makes the purpose of investigation look to be just an artistic exercise. Events do leave behind items of evidence, and these items do serve as sample points for one or more coherent scenarios. But it can sound like the "line" or the scenario is simply something made up. Oswald was last seen on the sixth floor of the TSBD, just before the rest of the employees broke for lunch and went outside to view the motorcade. Immediately afterward, Officer Marion Baker rushed into the building and spotted Oswald in the lunch room on the second floor, which was near the stairs. Here we have two sample points--Oswald on the sixth floor and then on the second floor--that we can, "connect", so to speak, by claiming that Oswald walked down the stairs from the sixth floor to the second. We connect them this way because in the world, people often move from floor to floor by walking stairs. If there is a line here that connects them, so to speak, it is constituted by the acts of stair walking that have occurred. The success of the investigator in catching murderers and getting them off the streets is determined by how well he can relate small sets of evidentiary data points to various forms of human activity and so on. This is why experience pays off. It is not so much that the investigator comes to the case full of "beliefs", but that he has identified the types of things that tend to occur in the world. The CT on the other hand has already drawn his line prior to plotting the data and so as such then removes any data that doesn't agree with his line so that the data fits the line, rather then the line fitting the data.Conspiracists are stubborn, and they don't listen to reason, but I am not so sure there is any sort of "line" in what they say. Hold your nose and take a look at Cosmic Dave's 32 questions again: www.clavius.org/bibdave32.htmlIs there any sequence of events or actions that you know people do, or could do, that have as a subset the points touched on by Dave's 32 questions? It is not clear what coherent scenario could emerge from such a random list of points (never mind the fact that all of them are based on a misunderstanding of how the world works). There might be a single thread that runs through them all, but it is hard to say what that could be. Compare the list of Dave's questions to any one of Jay's answers. The collection of points touched on in each answer are coherent because they derive from the world and the way it functions, and the world is the standard for coherency. What I haven't been able to understand though, is why do people, often people that are reasonably intelligent, do this? Why is the CT so attractive to people that they would so willingly switch off their minds to accept it? That I don't understand.I guess if there was a reason for people to act crazy, their behavior wouldn't be considered crazy. I don't know why people want to be conspiracists, but what let's them get away with it is that they don't investigate crimes, aircraft accidents, or explore space for a living. There is little penalty for them in being so wrong about these areas. I bet very few of them, if any, argue that their paychecks are a hoax and refuse to cash them.
|
|
|
Post by JayUtah on Jun 15, 2006 0:19:47 GMT -4
We connect them this way because in the world, people often move from floor to floor by walking stairs.
And of course this is how we do literal curve-fitting. If I know from basic principles that two variables are probably related in linear fashion, I won't fit a Bezier curve through them. I'll fit the best-fitting line.
There might be a single thread that runs through them all, but it is hard to say what that could be.
Actually this is a good point. Conspiracists don't necessarily take as great pains in their curve-fitting as we would. Certainly they do propose some limited scenarios, such as the tortuous darkroom method by which the crosshairs were allegedly produced and consequently messed up.
But primarily the conspiracists don't really draw their own lines. Instead they look at the outlying data that supposedly doesn't our line exactly. These "anomalous" data points are held up as evidence that the line we draw somehow can't possibly be correct because it misses some particular data point. That procedure is based on the wrong assumption that data in a true event must be perfectly consistent with theoretically expected results. They draw high-order curves that locally interpret all the data points, but that high order maintained over the breadth of the data reveals no trend.
In other cases the conspiracists insist a line must be drawn, when the data instead suggest a higher order. And so data points are considered outlying when they really aren't.
|
|
|
Post by gwiz on Jun 15, 2006 3:05:05 GMT -4
What I haven't been able to understand though, is why do people, often people that are reasonably intelligent, do this? Why is the CT so attractive to people that they would so willingly switch off their minds to accept it? That I don't understand. There is the obvious pleasure of being in on the secret, of knowing you're right and all the experts are wrong. "They all laughed at Christopher Columbus" is well established as the iconic example, and people like to see themselves on Columbus' side. Unfortunately, most of the people that "they" have all laughed at deserve the ridicule, and the chances that the independent thinker that you're following is another Columbus are minimal. David Icke, to name but one. As Michael Shermer has pointed out, intelligence is no bar to belief in the weird, in fact it gives the believer a better chance of being able to find ways of justifying the belief.
|
|
|
Post by Joe Durnavich on Jun 15, 2006 11:20:04 GMT -4
If I know from basic principles that two variables are probably related in linear fashion, I won't fit a Bezier curve through them. I'll fit the best-fitting line.
OK. Let's flesh this out a bit:
It depends on what you are trying to accomplish. If you picked up a piece of scrap paper that had two points you knew were linearly related, you could draw a big "J" through them anyway just to amuse yourself.
If, however, you had to determine the current a 48-blade server draws, but knew only the currents 32-blade and a 64-blade servers draw, and knew the relation between current and blade count was linear, you could plot the two points, draw a line between them, and get your result by interpolation. In this context, you are not just drawing a line, but building a power system. There is a wide range of productive activity that may follow this that might include somebody in a factory winding copper wire into a transformer, somebody else loading it onto a truck, a driver hauling it across the country, electricians installing it, managers and accounts overseeing the whole process, and so on, with the end result of a functional, high-powered computing system. You are not "fitting" a line to a set of points, but fitting a computer system to a power grid.
Let's say you were developing the Apollo guidance computer software. Let's say that the LM needs a computer system to land. Without it, there is no lunar landing. There is not enough CPU power to calculate trig functions and the like to high accuracy, so you store a list precomputed values in a table and linearly interpolate between them even though you know the relationship is not linear. The approximation does cause errors in guidance, but they are found to be acceptable. They are acceptable because your strategy allows a landing to take place. The type of "line fitting" that is best is determined by the context of what you are trying to accomplish.
But primarily the conspiracists don't really draw their own lines. Instead they look at the outlying data that supposedly doesn't our line exactly. These "anomalous" data points are held up as evidence that the line we draw somehow can't possibly be correct because it misses some particular data point.
Yes, we see this a lot with witness testimony. There are almost always plenty of outliers to be found in witness testimony.
That procedure is based on the wrong assumption that data in a true event must be perfectly consistent with theoretically expected results.
Also, as I try to demonstrate above, even though you may have data points that you cannot explain or fit, you will tend to be better off by having a theory that explains more points than any other competing theory.
They draw high-order curves that locally interpret all the data points, but that high order maintained over the breadth of the data reveals no trend.
Here is where I seem to differ from most of my non-conspiracist brethren. I have a hard time picturing Jack White (or Cosmic Dave, etc.) doing anything that can be considered even by analogy as connecting their points, especially by drawing a high-order curve. That attributes, shall we say, more brainpower to White then I think he makes available when he has his conspiracy theorist hat donned.
Here are three of Dave's points. The act of line fitting, in this context, would involve describing the series of actions and events the left points 21, 22, and 23 behind as evidence. This is precisely what Dave does not do. I think what happens is that we assume Dave has connected them, and try to connect them ourselves and find ourselves trying to fit a high order curve to his points. (Notice too: What is the point of item 23? Notice that Dave makes you fill in the point that the administrator left because of the sinister activities at NASA. That is supposed to be Dave's responsibility; not yours.)
If we all went to a JFK conspiracy convention and talked to many of attendees, we might find their beliefs, many, varied, but ultimately a conflicting mess. The attendees on the other hand, listen to all these beliefs and think instead that the amount of evidence is really piling up against the Warren Commission.
They are not bothered by contradictions in the beliefs because they never consider any two together. The "evidence", of course, is not evidence like we deal with it, but "smoking gun" proofs. Costella described his job in the conspiracy theorist community like this:
Each item is not a point to be fitted to a line, but a "self-contained proof". The lists these folks present are not meant to be fit to a line. Each item is supposed to be considered in isolation.
The conspiracists apply this same narrow-focus thinking when criticizing. I have seen someone describe the Single Bullet Theory to one of them, whereupon the conspiracist took each point of evidence, isolated it, and then gave reasons why he could doubt that single item. At the end, he proclaimed that there is no evidence that can prove the SBT.
|
|
|
Post by nomuse on Jun 15, 2006 13:45:43 GMT -4
I sometimes describe it as the "shotgun fallacy." The CT's come up with a number of "points" or "anomalies"; observations they ascribe significance to. Their fallacy is that sheer weight of numbers counts. Five weak explanations for how low gravity movement was simulated are stronger, in their view, then one good explanation. The fact that the explanations may be contradictory is not important. All that is important to them is that they add one more bit of weight to the idea that a hoax was possible -- or that they chip away one more little bit at the official story.
|
|
|
Post by JayUtah on Jun 15, 2006 15:29:42 GMT -4
There are almost always plenty of outliers to be found in witness testimony.
And in a strange quirk of interpretation, the outliers are generally considered to be more important than those who remain consistent. So if you have 10 data points, 9 of which fit a simple curve and one of which lies significantly outside it, the conspiracy approach considers that one outlier to be as important as the 9 consistent ones.
...you will tend to be better off by having a theory that explains more points than any other competing theory.
Yes. Conspiracy theorists want to focus on the "absolute" credibility of some particular curve rather than comparing whether one curve fits the data better than some other one. Some data are naturally inconsistent, and fitting a line through them may be entirely problematic, introducing many outliers. But if that is still the line that minimizes error in the prediction, it is still the best theory even if it is objectively unappealing.
The act of line fitting, in this context, would involve describing the series of actions and events that left points 21, 22, and 23 behind as evidence.
You're not so divergent as you suppose.
By "local fit" I mean that the conspiracy theories have a convoluted, high-order curve that may fit three or for "adjacent" points and interpolate them all acceptably, but that same strategy doesn't scale up to the entire data set -- in other words, Dave may have some theory that explains how the photographs were faked and produced, as a side effect, all the eroded or misplaced fiducials (e.g., cutting and pasting, plastic overlays that slipped), but that theory doesn't really go beyond those particular observations.
If you have 100 data points, and you look at points x[54] through x[59] whose coordinates require high-order curves in order to pass through them, you can probably devise such a curve that interpolates x[54-59], but that same curve departs immediately from x[53] and x[60] and surrounding points. The theory holds only for that subset of data and would have to be significantly revised and extended to an even more absurdly high order in order to account for all available data.
Conspiracy theorists work on the wrong presumption that every record of a true event must be explicable and consistent, which simply does not occur. The existence of "noise" in the evidence does not render the evidence unsuitable for analysis nor the theories that judiciously ignore inconsistency untenable.
If we all went to a JFK conspiracy convention and talked to many of attendees, we might find their beliefs, many, varied, but ultimately a conflicting mess.
Exactly. The "conflicting mess" you describe is the qualitative equivalent of the high-order curve that you would need if you insisted that it pass through all points in noisy data. The world, evidentiarily speaking, is "noisy" containing many things that are possible to observe, but which defy simple explanation as causes precipitated by other observations. We can note a cigarette butt on the subway seat, but we do not necessarily have the means convincingly to connect it to or dissociate it from a mugging that occurs in the same car. A theory that explains the mugging can variously incorporate or ignore the cigarette butt, as parsimony dictates.
The attendees on the other hand, listen to all these beliefs and think instead that the amount of evidence is really piling up against the Warren Commission.
And that is simply the natural outcome of an unnatural level of scrutiny applied to the problem. The more you look for details and the more you wrongly assume that a viable theory must account for all the details you uncover, the more you will become wrongly convinced that a theory is untenable even if it is still the best one. If you were to follow up any investigation with such a level of scrutiny you would inevitably increase the number of things that "don't add up". That does not necessarily constitute a meaningful challenge to the validity of the original findings.
Each item is supposed to be considered in isolation.
Exactly. Instead of induction you get merely re-expression. Costella puts together high-order curves that interpolate precisely a handful of "anomalies" and nothing else; but he provides no overarching insight that is typically the goal of investigation.
As I've told Michiel Brumsen repeated on BAUT, I will have some respect for these "investigators" when they submit themselves to be held accountable if they are wrong. Nothing galvanizes one's quest for reliable methods than self-preservation.
|
|