Programming languages are for humans and machines. Machines do better with braces. (I've written lexers for both.)
Also, try to write an automatic formatter for a whitespace-delimited language; if your indentation is wrong, your logic is wrong, whereas with a brace language, if your indentation is wrong, you run a formatter, and it's all golden.
A parser is something you write once and then it's done, so it shouldn't be the deciding factor. And parsing indentation-based languages is not that hard, just replace indents and dedents with a special token.
In a brace-based language, if your braces are wrong, your logic is wrong. How is that better? At least a mistake in indentation is much easier for a human to spot.