|
Author
|
Topic: Calibration of science's measuring tools
|
Wade
Member
Member # 1713
|
posted 07. August 2005 10:02
I'm a new poster to this group looking for threads related to calibration of the tools "science" uses to make observations in general, and observations of intelligent intervention in particular.
Is there a list, or review article, or summary of what tools science DOES have in its toolkit that do claim to measure interventions and have been empirically validated on synthetic test cases?
Staying entirely within the realm of natural science, I think there is a recognized need to calibrate any measuring system or tool or process before using it. A system designed to measure X should, one would hope, at a minimum, respond positively in a carefully crafted test case that includes X, and respond negatively in a carefully crafted test case that excludes X.
That constraint is necessary, but not sufficient. For example, radio astronomers for years missed detecting pulsars, third brightest objects in the radio sky after the sun and the galactic center, because they assumed (wrongly) that bright things had high AVERAGE brightness (instead of high PEAK brightness, as pulsars are basically strobe lights). Only when a graduate student finally removed the blocking filter could pulsars be seen. Oopsie.
So, I'd like to raise a question far reduced from the purpose of life and the nature of the universe, at least to start.Question: Has natural science provided us a tool that demonstrably can distinguish situations in which there was intelligent intervention from those in which there was no such intervention,and demonstrated the statistical power of this tool (or technique or method or process) on some unambiguous test cases, highlighting false positives, false negatives, etc?
My impression is that no such tool exists, pretty much regardless of how one defines "intelligent intervention". Should such a tool exist? Should the absence of such a tool be an item of some concern among scientists?Is there currently any generally accepted, validated tool that can tell, whether looking at a cake baking in an oven, or a social opinion forming regarding some issue, whether the activity is "spontaneous" or "forced"?
In thinking about this, I ran into the frustration of what probably has a name that I don't know, that I'll call the Preventive Maintenance Paradox: The better job a maintenance staff (for water, or sewer, or computers, or social order) does at preventive maintenance,the higher the odds they'll all be laid off because "nothing ever goes wrong."
Similarly, in health care or corporate management, the better job one does at seeing far down the road and heading off problems before they become visible, the lower one's reputation. For example, preventive health care is demonstrably far more efficient than acute rescue care, yet the country spends less than 2% of the health care budget on prevention.
Or, watching a student pilot versus a professional pilot fly an approach to an airport and landing - the student pilot will be extremely busy with massive course corrections, and the professional will barely touch the controls from time to time with fingertip nudges. So, neither the magnitude of an intervention, nor the micromanaged level of visibility is a very good guide for how "intelligent" the control intervention is.
Largely, what we take for "management" is visible efforts to dramatically solve issues that, if handled properly long ago, would never have arisen in the first place. In fact, looking at the pilot/plane model and trying to deduce whether there is intelligent control or an unguided missile, I'm not sure anything can be done in calm air - the two would behave identically.It's only when there is turbulence and cross-winds and other external forces trying to break the "homeostasis" of some plan that positive-control and non-control can be distinguished. The professional pilot's craft will continually return to the same course and never diverge far, the student's craft will be lucky to end up at the right airport shiny-side up, and an unguided craft will simply crash or wander off out of sight.
In that sense, maybe the simplest primitive element we could look at would be a single feedback-control loop (as a black box) and the inputs and outputs from it.And here, again, as with the pulsar, there seems to me to be a black-hole of perception, because most of our statistical measuring tools assume, pivotally, that there is a start, a possibly complicated set of paths, and an end. But in the real world, the end feeds back to the start, and so the terms "start", "end", and "cause" become ill-defined in classical terms. We need to borrow tools from analog electrical circuits, or from Engineering Systems Control to analyze such data sets, and that has generally not been done,even in areas such as genetics where "circuits" and "feedback" are everywhere, but feedback-controlled goal-seeking primitive elements are conspicuously lacking from published papers.
This seems an important and addressable gap in science that might be a place to focus some effort. Anyway, mabye other readers have some knowledge of such tools and I'm the only one in the dark here. Comments would be greatly appreciated.
IP: Logged
|
|
Wade
Member
Member # 1713
|
posted 09. August 2005 07:19
OK, let me rephrase that more simply.
Here's a question I suggest being posed to "science". Let's bring this down from the universe to household objects.
What tests do YOU have that can tell an intelligently designed object or process from one that's not?
I'm suggesting that this is a "middle ground" question, designed to improve the quality of dialog.
I also think it's a litmus test question that highlights a tendency of "science" to respond to questions on which its tools perform very poorly, and try to reframe them as being "bad problems" or "unscientific areas". This seems inexcusable "sour grapes" reasoning.
So we have the "hard sciences" in which the quantitative, isolationist, reductionist approach works relatively easily. Then we have the "soft sciences" (psych, economics, social science) in which the tools work poorly or not at all, but this is taken as proof that there is something wrong with the area, not with science.
Then we have problems of world hunger, poverty, greed, corruption, disease, injustice, employment, affordable housing, etc., at which science appears to have nothing to say at all, despite these being way more important problems to most of us than discovering a 10th planet.
So, what I hear "science" saying is, basically, "If we reduce the problems under consideration to those that science does really well on, science should get a grade of A or maybe even A+ !" And I hear the rest of society saying "If we look at problems most of us care about and understand, and progress made over the last 100 years, globally, maybe science should get a failing grade."
Measured by inputs, Science is soaring along with never so many papers per minute published.
Measured by outputs, Science is increasingly irrelevant.
That, it seems to me, is a fair issue to discuss.
IP: Logged
|
|
Christopher D. Beling
Member
Member # 723
|
posted 02. September 2005 23:54
Hi Wade, I wonder what you think of Dembski's Complexity-Specification Criterion (CSC) for detecting design (i.e. intelligent intervention)? You can read about this either in chapter 5 of the book "Intelligent Design" or chapter 2 of the book "No Free Lunch" by Dembski. According to this criterion - if a system has a specified function and at least 500 bits of information associated with that specified function (i.e. sufficient complexity) then we can infer design. Agree?
quote:
Or, watching a student pilot versus a professional pilot fly an approach to an airport and landing - the student pilot will be extremely busy with massive course corrections, and the professional will barely touch the controls from time to time with fingertip nudges. So, neither the magnitude of an intervention, nor the micromanaged level of visibility is a very good guide for how "intelligent" the control intervention is.
There seem to be two issues here:
(i) Can one detect intelligence in the landing system of the airplane?
(ii) Assuming an affirmative to (i) can one quantify (measure) the degree a degree of intelligence
It would be good to clarify which is your priority to answer. I would say that (i) is fairly easy using the CSC. It is clear that in the complete cybernetic system for landing the plane that you have a specified function (landing the plane). At the same time the system has a huge amount of complex information (bits) required for BOTH construction of the planes control system + the complexity of the human eyes, that part of the brain that deals with control, and the muscles of arms, hands and legs. The system registers positive in the CSC filter. The detection of the design, however, requires having a more detailed internal descriptions of the configurations of both pilot's and plane's operating systems. Just by observing a plane landing would not tell you that the system was designed.
With regard (ii) - things are more difficult, but perhaps the answer lies in how closely the flight path of the plane satisfies some specified target path for the "ideal" landing. Interested to know your views and those of others.
[ 03. September 2005, 00:16: Message edited by: Christopher D. Beling ]
IP: Logged
|
|
|
Printer-friendly view of this topic