Last week I watched again the highly amusing film "Kitchen Stories". It's hardly a mainstream affair - in fact I feel like editing Wikipedia's meagre entry on it. The scenario is amusing because it's so ridiculous - a group of Swedish scientists is sent off to Norway to observe single men use their kitchens, in order to optimize the kitchen layout for them. I detected a bit of Norway-poking-fun-at-Sweden in this film (or was it the other way around?) but also quite a lot of everyone-poking-fun-at-scientists-and-their-stupid-studies.
Observer Nilsson gets the short-straw and has to observe the intensely unco-operative Bjorvik. The observers have been instructed that it is of paramount importance that they do not communicate with their subjects or interact with them in any way. The research is from a positivist perspective - that means the subject 'does' and the observer 'observes'. Any mixing of the two would jepoardise the whole undertaking.
In protest at this stupidity Bjorvik refuses to do anything in his kitchen. He then decides that if he is being observed, he can observe back, and starts observing Nilsson. There are long periods in this film where nothing else happens. No dialogue, no interaction, just one observing the other.
What happens from then I'll leave you to find out. Be warned - this is not fast-paced Indiana Jones-style entertainment.
There are some interesting parallels with science here. An underlying assumption in the 'positivistic' approach is that there is a real reality out there that has no relation to what's looking at it. Doing an experiment on it doesn't change it. However, just as Nilsson's presence changed how Bjorvik behaved, so it's not always the case in science that we can do this. In some experiments, it is pretty hard to carry out the experiment without fundamentally changing what it is you are experimenting on. The extreme example is quantum mechanics, where it is impossible to observe a system without making a drastic change to it. The observer cannot be isolated from the system.
But even in more classical situations, the issue may still be there. I have a project in which a student is tackling the thorny problem of measuring the electrical conductivity of biological tissue. In order to keep the tissue 'alive' (in the sense that any isolated piece of tissue can be alive), it has to be bathed in a solution that contains what the tissue needs to function. But the solution has an appreciable electrical conductivity itself. If we measure the conductivity of the tissue on its own, it will be dead tissue - but if we measure that of the alive tissue it won't be the conductivity of the tissue alone. Tricky. To measure it, we have to change it.
It's an issue that's worth more than a passing thought in the design of a good many experiments.