Right, the problem with CSV is that it's often generated incorrectly.
Commas do not need to be escaped inside quoted blocks. But now quoted blocks are not literal data, and you need to drop the enclosing quotes. And escape quotes inside them, or outside if the quote is literal. Also quotes must be balanced inside a delimited field. Etc.
Commas are just too common of a character to be a good delimiter.
Tabs are an improvement, because usually tabs are just whitespace, which is usually arbitrary, and never needs to be balanced.
But tabs are not perfect either. Just a vast improvement for most data, and a functional compromise for humans.