Theory
Modern use of the word "theory" is a little confused and hence confusing, and might benefit from looking back at ancient Greek usage. The word in its most elementary form means "sight" but with the connotation of sight from a distance -- which is why theoria was also used to refer to the sending of ambassadors (theoroi) abroad -- to scope something out. This why Croesus refers to Solon as a philosopher who has "gen pollen theories heineken" -- "traveled widely for the sake of seeing" (Herodotus, Histories, Bk. I, Ch. 30).
To theorize, one goes out into the world. To someone in archaic and classical Greece (and, as airstrip correctly pointed out last time I played around with this word, India), sight was something that went out from the eyes to meet the world -- which after all wasn't such a bad primitive metaphor to express the way your brain imposes order on sensory information, constructing perceptions based on probabilistic guesses about how the world is.* Your brain uses theory because theory is cheap: if it can construct reliable perception with minimal sensory processing, it can then filter out all the redundant bits, which saves attention for use on other things. It can also make better discriminations between ambiguous stimuli, and amplify weak signals to perceive things that would otherwise remain subliminal.
Theory in science functions the same way: it's a cognitive labor-saving device, but more than that it's a technology for extending and amplifying perception -- it allows you to see things you couldn't see without it. The world is really complicated, and being able to ignore most of it and pluck out only what's useful increases one's productive powers considerably. This is why my maxim that facts are a poor substitute for theory is a deliberate inversion of common sense: facts themselves are actually theory in the wild -- they're determined by a dense network of assumptions which are rarely examined or understood, are collectively incoherent as a rule, and take some discriminating judgment and a lot of brain power to make sense of.
What we usually think of as scientific theory is the domesticated kind -- you have some explicit grasp of what your assumptions are, and you try to make them as clear and simple as possible so that their implications can (ideally) be deduced by an electric abacus. One of the many virtues of this is that it leaves comparatively little wiggle room when the theory's predictions come up persistently false, simply by virtue of its paucity of parameters. Facts, on the other hand, have a lot of implicit parameters to twiddle, and Darwin was absolutely right that false facts are more dangerous than false theories on account of being harder to spot and squash.
This leads to a biased perception which gives rise to the more commonsensical distrust of theory -- it's easy to find instances of theories misleading, but harder to find instances of facts misleading because the process by which the latter happens is simply less transparent and obvious. It also tends to be forgotten -- one good theory quietly sweeps away ten thousand lousy facts, most of which are never documented for posterity because they're things every schoolboy knows. And the few that are documented get quietly swept under the rug and ignored because "now we know better", even though the theory that killed them may have encountered enormous resistance before being accepted. Being right had better be its own reward, because otherwise it's thankless.
But I digress. The point is that there is no dichotomy between theory and empiricism: theory is about seeing things out in the vast fog of possibilities surrounding us, and empiricism (i.e. experimentation) is a method to sort the real sights from the phantoms. Empiricists need theory more than anybody because it's the difference between using a nuclear weapon to wipe out a horde of wrong ideas versus knifing them all one by one from here to eternity.
* Actually, we don't even have to go back in time -- people still think about vision in those terms today. In a sense the directionality is a red herring -- what matters is that part of your brain is resonating with something beyond it. Empedocles, like many pre-Socratic philosophers, was ahead of his time.
To theorize, one goes out into the world. To someone in archaic and classical Greece (and, as airstrip correctly pointed out last time I played around with this word, India), sight was something that went out from the eyes to meet the world -- which after all wasn't such a bad primitive metaphor to express the way your brain imposes order on sensory information, constructing perceptions based on probabilistic guesses about how the world is.* Your brain uses theory because theory is cheap: if it can construct reliable perception with minimal sensory processing, it can then filter out all the redundant bits, which saves attention for use on other things. It can also make better discriminations between ambiguous stimuli, and amplify weak signals to perceive things that would otherwise remain subliminal.
Theory in science functions the same way: it's a cognitive labor-saving device, but more than that it's a technology for extending and amplifying perception -- it allows you to see things you couldn't see without it. The world is really complicated, and being able to ignore most of it and pluck out only what's useful increases one's productive powers considerably. This is why my maxim that facts are a poor substitute for theory is a deliberate inversion of common sense: facts themselves are actually theory in the wild -- they're determined by a dense network of assumptions which are rarely examined or understood, are collectively incoherent as a rule, and take some discriminating judgment and a lot of brain power to make sense of.
What we usually think of as scientific theory is the domesticated kind -- you have some explicit grasp of what your assumptions are, and you try to make them as clear and simple as possible so that their implications can (ideally) be deduced by an electric abacus. One of the many virtues of this is that it leaves comparatively little wiggle room when the theory's predictions come up persistently false, simply by virtue of its paucity of parameters. Facts, on the other hand, have a lot of implicit parameters to twiddle, and Darwin was absolutely right that false facts are more dangerous than false theories on account of being harder to spot and squash.
This leads to a biased perception which gives rise to the more commonsensical distrust of theory -- it's easy to find instances of theories misleading, but harder to find instances of facts misleading because the process by which the latter happens is simply less transparent and obvious. It also tends to be forgotten -- one good theory quietly sweeps away ten thousand lousy facts, most of which are never documented for posterity because they're things every schoolboy knows. And the few that are documented get quietly swept under the rug and ignored because "now we know better", even though the theory that killed them may have encountered enormous resistance before being accepted. Being right had better be its own reward, because otherwise it's thankless.
But I digress. The point is that there is no dichotomy between theory and empiricism: theory is about seeing things out in the vast fog of possibilities surrounding us, and empiricism (i.e. experimentation) is a method to sort the real sights from the phantoms. Empiricists need theory more than anybody because it's the difference between using a nuclear weapon to wipe out a horde of wrong ideas versus knifing them all one by one from here to eternity.
* Actually, we don't even have to go back in time -- people still think about vision in those terms today. In a sense the directionality is a red herring -- what matters is that part of your brain is resonating with something beyond it. Empedocles, like many pre-Socratic philosophers, was ahead of his time.