Originally Posted by
Singu
Predictions isn't actually the main point, because you'd first need to know what to predict, or even what it is that you're supposed to be looking for.
Imagine that there's a computer that could magically predict anything, as long as you input what to predict. That would be the ideal Instrumentalist machine. But what are you supposed to be predicting...? If you want to predict whether a spaceship will liftoff or explode in mid-air, then you'd have to first build that spaceship, then test whether it will liftoff or explode. And in order to build a spaceship, it requires knowledge, which is not just knowledge about predicting things. And even if it does indeed predict that it will explode in mid-air, how would we build a spaceship that won't explode? We wouldn't know how without understanding how a spaceship works.
So what this magical computer is actually doing is that it's just replacing human experiments with perfect hypothetical experiments. It still requires the creativity and the imagination of human beings to come up with new theories to understand how things work, which will be done by understanding how reality actually is and using theories to uniquely solve problems. And if you can understand how things work, then you will automatically be able to predict things. So understanding comes first, then predictions later.
Instrumentalism is about using theories as an "instrument" to predict things. It says that its content doesn't matter, perhaps other than as some "useful fiction". Logical Positivism is an even stronger form of Instrumentalism that says that any statements that are not about predicting observations are meaningless. But then what is Logical Positivism itself...? It considers itself to be meaningless.