Since you were a baby, you learn from observation. You'll mimic behaviors of your parents or siblings. You'll start to realize that if you cry, you get food. We learn all these patterns and behaviors from observing others and the environment around us. This is a vital skill and many great discoveries started from an observation. It is however extremely dangerous and rather stupid to rely solely on observation to find out about something. Take a look at these examples.
Assuming you have no knowledge of the mathematical symbols addition (+) and multiplication (x), what can you conclude from observing these 2 equations
2 + 2 = 4
2 x 2 = 4
From these 2 equations, you can conclude that + and x are the same and some might even go a step further and argue that x is just + written differently or wrongly
Now another similar example. What can you conclude from these 2 equations
4 - 2 = 2
4 ÷ 2 = 2
Similarly, you can observe from these 2 equations that - and ÷ gives the same result
Some of you might say, well those are rather simple and dumb examples. No one will conclude something simply based on a small observation sample. But sad to say, in my limited 20+ years of observation, there ARE people who draw conclusion from 1 or 2 observations.
Do you know anyone who decided not to buy something simply because they heard their friend's friend had some issue with it? Did they check if it's the same model? Was the issue due to faulty manufacturing or a user's mistake?
It becomes scarier when people start to use past observation to predict the future. Your observation is simply 1 possibility and there could be millions or billions of others. When you start to assume, you need to be aware of the possibility of being wrong.
Sometimes you try to draw a better conclusion by asking more people. But what if your friends all have similar demographics? Just because all your friends think something is good doesn't guarantee that it is. After all, peer pressure will tend to make individuals conform to their social norm.
I'm not saying that you shouldn't learn from observation but rather to not so easily conclude something based solely on observations. It's not enough to just increase the sample size. You need to consider different scenarios and always be open to other possibilities. You need to conduct experiments.
When someone tell you something is faulty, ask him exactly what is wrong. Ask other friends. Go online and do some research. If you can, get yourself a sample to test if the problem is an isolated case. Having done all that, you still must account for the possibility that it could be something else you missed.
An observation is merely the beginning of the learning process. You need to test and experiment your observation before you can conclude anything from it. Start by observing and make no assumptions. Test and experiment your findings. Then conclude but be ready to go through the entire process again.
Learning is a life-long process because a new observation can change what you already know. So start observing but don't forget to experiment.