If we are talking about machine transfer, there are tons of formats that work, and many of them are binary. xls, ods, SYLK, and dta are examples (not sure about

If we are talking about machine transfer, there are tons of formats that work, and many of them are binary. xls, ods, SYLK, and dta are examples (not sure about UTF8 wit SYLK and dta). Then there are the "just write it in code" formats like SQL dumpfiles (some xml fits here, too) - which work well and carry really nice, readable type declarations. If we are talking about human readable, text editor safe formats, we have lots of challenges. ASCII delimiter characters don't display and aren't on the keyboard (and will sometimes have side effects with old wire protocols). Nearly any character that is on a keyboard (or international keyboards) eventually becomes a problem because you will have to escape them when they are not used as a field (, is the classic example) or record delimiter (For example, <CR> can be valid inside of a longer text field). Also, many of the first 31 ASCII characters (the control characters) are used in different and surprising ways (the classic being null) by hardware and software. BTW what makes CSV bad is that there is no escape character for , or ". Likewise, TSV has the same problem with tabs. JSON, YAML, and XML seem to be much better, but all have edge cases.