What are predictions in science
FORECAST AND THEORIES
About dubious forecasts
How long will the world be populated by humans?
We can assume that today is picked out purely by chance from the total period of the existence of Homo sapiens. Therefore, there is a 95 percent probability that this day will not be in the first 2.5 percent and also not in the last 2.5 percent of this period. Since around 200,000 years have passed so far, there is a 95 percent probability that humanity will populate the world for at least 5 128 years and at most 7 800 000 years.
What is to be made of the "theory"? Apart from the entertainment value - nothing! It is true that it makes reasonable-sounding assumptions; it makes legitimate use of mathematics, and it provides a number result - even with a serious-sounding confidence interval. But it cannot be refuted under any circumstances; their predictive value is ridiculous; it does not lead to any consequences and it does not enable anyone to cope better with life. Their main weaknesses can be summarized in one sentence: There is a lack of experience with comparable issues.
In the Reliability and safety assessment it is about the prediction of the behavior of technical systems - that is, about forecasting methods. There is always the risk that nothing more will come out of the theory than in the case just described.
The classic reliability theory deals with hardware failures. Experience is available about the type and frequency of such failures. Reliability models on this basis enable reliable predictions.
In the area of Reliability of complex systems But there are also theories that are no better than those described above for the "survival of mankind". In my essays from 1990 and 1993, I count the following as missing reliability theories:
- Reliability growth models. The large number of models alone is astonishing: Poisson model, Musa model, Jelinski / Moranda model, Littlewood models (!) And many more. A short summary of my criticism can be found on my Reliability Growth Models Criticized page.
- X-Ware Reliability. This "theory" takes failures into account and Design flaw; Nevertheless, constant failure rates are assumed. This can not go well.
- Fault tree models for diverse systems. This flawed modeling persists and has some tradition in the evaluation of parallel redundant software.
How can the wheat be separated from the chaff? When it comes to forecasting methods and scientific theories, it is best to seek advice from Karl Raimund Popper.
What distinguishes useful prognoses or theories?
Knowledge is synthetic insofar as certain assumptions are made which are not valid a priori and which must be confirmed by experience. It is analytically in the parts that are based solely on logical inference and math. The synthetic (empirical) content of forecasting methods is best measured against the criteria for scientific theories. According to K. R. Popper, these include the following.
- Falsifiability: A forecasting method must have a predictive value and experience, at least in principle, must be able to fail. (So not something like: "If the cock crows on the dung, the weather changes or it stays as it is.")
- Proven: We consider a (falsifiable) prognosis method, the prognostic value of which has proven itself in practice in many different cases, to be tried and tested.
- Objectivity: Statements (forecasts) must be able to be checked intersubjectively.
- Simplicity: A forecasting method must not depend on too many adjustable parameters, otherwise it escapes refutation too easily and has too low a forecast value (see reliability growth models)
Carnap, R .: Introduction to the Philosophy of Natural Sciences. Ullstein 1986 (1966)
Popper, K .: Logic of Research. Mohr, Tübingen 1982 (1934)
Two influential critics of Popper:
Feyerabend, P .: Knowledge for free people. Suhrkamp 1980
Kuhn, T .: The Structure of Scientific Revolutions. Suhrkamp 1976 (1966)
From the origin of our prognostic skills
Introduction to the 4th Fulda Electrical Engineering Colloquium 1999 on October 29, 1999
On one of my forays through the Internet, I came across a page that asked "What is simulation?" and the answer is: "Realization of reality in the computer". But: What is real
It was maybe 1 million years ago. Several types of people coexisted on earth. Not just Neanderthals. Then a new species appeared - one with an oddly high forehead. When jumping, running and hunting, the large head was more of a hindrance. But it also offered advantages.
The great brain endowed these modern humans or Homo Sapiens - as we say today - with a special ability (Dawkins, 1978, p. 71): The new human could imaginations evolve from objects. And he could Be-aware-of-yourself. He could see all the things in his head featured space move. He could try this and that in his head and see what would come of it for him. And all of this before he took action. Short: He could think!
This ability was found to be extremely beneficial. This new breed of people was so successful that they ousted all of their competitors. This is probably one of the reasons why we humans, unlike most other living beings, no longer have any close relatives on this earth (TIME, 23.8.99).
Thinking is a way Simulation in the head: Our ideas about the objects are Models. And the trial treatment in the presented room is that experiment.
What role does the computer play in this context? In fact, it doesn't bring anything essentially new. It only does one thing: it expands our ability to simulate in the sense of faster, higher, further - nothing else.
The computer only "knows" the models that we enter into it. But we ourselves have no direct access to reality - we don't even know whether there is such a thing as reality at all.
This is an old philosophical problem (Poundstone, 1988). It is also the subject of the film MATRIX - whoever has seen it knows what I'm talking about.
That is why we are completely unable to plant a picture of reality on the computer. The definition "simulation is the reproduction of reality in the computer" is nonsense. It is better to characterize them Computer simulation as a kind of extension of our thinking skills.
Dawkins, Richard: The egoistic Gen. Springer, Berlin, Heidelberg, New York 1978
TIME, 8/23/99: How Man Began. Up from the apes
Poundstone, William: Labyrinths of Reason. 1988
© Timm Grams, December 1st, 1999
- Is blockchain in the cloud
- How children found a band
- Slow drivers are less skilled and incompetent
- Why did Walt need Jesse
- Is it difficult to program in Python
- Tangle is necessarily more central than blockchain
- What about fishing makes you happy
- Which body has the constitution of India established?
- What is hemp extract 1
- What are the best part time jobs
- How long has blockchain been around?
- Which art store is your favorite
- How much do Cuban cigars cost
- Craniosacal therapy is dangerous
- Where do I fit in this world
- How does radioactivity affect spiders?
- Why are some people obsessed with love
- What is the importance of water binding angles
- What politicians the US don't deserve
- Why does an electron have a charge
- What is the stamen of a flower
- Which musical instruments prefer to sound like language?
- Can I speak while meditating?
- Winston Churchill was a terrible person