You can always do enough research to defend your theory.
The technological process is dominated by two types of people:
1. People who understand that they are failing.
2. People who get what they don't understand.
If research isn't worth doing, then there's no point in doing it well.
After a painful and careful analysis of the sample, you suddenly find out that the sample was not taken and has nothing to do with the case.
If you are working on a solution to a problem, knowing the answer will always help solve it.
Every big problem has a little one that just wants to get out.
No problem is so huge that it cannot be simply dismissed.
If the facts do not support the theory, they need to get rid of them.
1. The more verbose the theory is, the better.
2. An experiment can be considered successful if no more than 50% of the measurements made can be discarded in order to reach agreement with theory.
The number of hypotheses explaining this phenomenon is inversely proportional to the amount of knowledge about it.
No matter how painstakingly and carefully you prepare a sample, you can always be told that it is wrong and unacceptable for this work.
When working on a solution to a problem, it is always helpful to know the correct answer.
Inside each big task is a small one trying to get out.
Never try to repeat a successful experiment.
1. If it is difficult to reproduce the experiment, run it once.
2. If you need to draw a straight line, get it only by two points.
All great discoveries are made by mistake.
If you don't understand a word in a technical text, ignore it. The text will completely retain its meaning without it.
The progress of science is inversely proportional to the number of published scientific journals.
Scientists are so lost in their own head that they do not see a single phenomenon as a whole, including their own research.
There is always not enough time to get the job done right, but there is time to redo it.
1. If there is an unknown scale factor in the problem, assume that it obeys a power law with an exponent of 0.70.
2. All characteristic numbers in everyday life usually have a 25% spread, which only occasionally drops to 10%. The experimental data error is almost always greater than 1%.
3. Any truly useful classification contains three to six categories.
4. Whatever quality we want to evaluate, there will always be at least three conflicting criteria for evaluating it.