Andrew at the "Statistical Modeling, Causal Inference, and Social Science" blog (that's a long name!) is often talking about how he think that graphs should always be used in favor of tables.
He has created a website called tables2graphs describing how this can be done and has developed some programs to do it.
I agree that tables are not as visually pleasing as graphs and I think that seeing data graphically can illuminate patterns that we might not otherwise see. The only advantage of tables that I can imagine is that with graphs you don't get the exact values, you just get eyeball guesses, but I'm sure they have some way of addressing that.