Evaluations differ from engine to engine - if you give a position to two different programs or to one program with different settings, you'll come up with different evaluations. This being the case, you'd better not worry about computer evaluations, as they're inherently subjective.
Should you however wish to get into an engine's reasoning, first try to find the program's evaluation formulas, or at least to get some info about that. It will help you know how are different kinds of positional features (including advantages and disadvantages) evaluated. Because this differs from engine to engine, you'd need those for the program you're using.
But for practical purposes, playable moves' evaluations can vary much more widely than +-0.30. Especially in the opening and in some endgames where material is not so important. When I analyze some game I've played with Black, often in the opening the positions are evaluated up to +0.80 to +0.90 favoring my opponent. (For example, 1.e4 d6 gets +0.78, believe it or not, while being quite reasonable and safe for Black.) That is certainly not credible because the lines in those games lead to equality (as human chess understanding sees it), and in the early middlegame, when all the developing moves have been played, the engine agrees with that too.
If you want to prove to yourself that computer evaluations are inaccurate, consider the following: a position's evaluation should by default be equal to the evaluation of a position that results from the first one with best play. Choose a position. Let the computer analyze it. Make the move the computer finds best. Repeat this several times. Analyze the final position. If its evaluation is different from the first one's, then either the engine doesn't find the "best play" line or doesn't evaluate accurately - both of which actually mean the same - that a move can't be accurately judged by such means.
It's a matter of calculation depth and of exactness of evaluation formulas. I guess it's necessary for computers to have some kind of formulas that don't work always exactly but are often accurate, or not to look deeply into the position so as to save time while achieving "enough" accuracy, however I'd not believe a computer evaluation in itself. You' better not too, because you can neither use an engine while playing so you needn't worry what their evaluations are, nor can you judge a move by a number without understanding why it is good or bad.
I wanted to start analyzing some games with a chess engine and have been trying to figure out how many points are considered "playable moves". I know 0.30 is considered by many as having a small advantage but should this number also be used when looking at a list of the best moves for a position. So another words any move within 0.30 is "playable".