1. Declaring a hypothesis causes confirmation bias.
2. There may not be such a thing as objectivity. So long there is an observer (a subject), there can be biases in observation. Peer review can help so long everyone is not biased the same way. Even machines and programs can be biased because they are made by humans. For example, cameras appear to emulate the human eye, not the eagle eye, not the cat eye, even though there is no way to prove that what we see is objectively truer than what other animals see.
3. The problem with inductive reasoning: even if a statement has always been true, we can’t say it will stay that way (Russel, ch. VI). For example, an animal that has a shorter life span than that of a chicken may not be able to guess that the chicken was once an egg. Having observed its growth, it would assume that it must have been a smaller chicken someday before which it used to be an even smaller chicken before which it used to be a dimensionless point in space.
4. The problem with deductive reasoning: all deductive conclusions depend on axiomatical first principles. The axiom is either assumed to be self-evident or is inductively produced. To call something self-evident can be considered philosophical laziness, so we mustn’t be blindly sure. And we just also saw the problem with induction.
5. The problem with observation itself: there is no way to really tell if the physical objects we see are any more real than shapes in clouds, which are nothing but projections of our own mind. A peer review would also reveal the same shapes. A camera would also show the same image. We may not ever know what the universe actually looks like before it is processed via the senses and the brain. Just how much of our perception is our own fooling?
6. Physical systems do not have to be isolated. Saying y= f(x) assumes an independent x, but no variable truly has to be independent. We can say f=ma, but mass (m) might be dependent on another variable, which does not change very often or does so only minutely, giving us the perception of a uniform entity called mass.
7. Not everything can be observed. Consider cosmic microwave radiations. Even though they are omnipresent, no one has ever seen them. They were discovered accidentally. Just wonder how many accidents never happened.
Russel, Bertrand. "On Induction". The Problems of Philosophy, edited by Andrew Churley, 1998.
Oxford University Press, 1959. Digital Text International, ditext.com/russell/russell.html.