Comment by Chris_Newton

There are plenty of advantages to text formats, to be sure. Others have mentioned many already, so I won’t repeat them. But let’s also consider both the disadvantages and how many of the advantages are “accidental” rather than inherent benefits of using a textual format.

One disadvantage of text is its lack of expressive power. Try reading the equation shown in the article aloud. Now try giving a one hour lecture on advanced quantum mechanics without the aid of mathematical notation. We can often represent information far more concisely and accurately with a good notation than with text alone, particularly when there is some inherent underlying structure that goes beyond what we can conveniently represent with some linear sequence of a tiny set of symbols. Computers are good at that kind of thing, but we don’t read Shakespeare in binary, and we certainly don’t draw the Twitter icon from the article using nothing but 1s and 0s.

Another disadvantage of text is how much it relies on everyone to use the same conventions, even though in the real world they don’t. Go just about anywhere in the world and you can recognise what the little pictures of a man and a women on the two doors in the restaurant mean. Replace them with ‘M’ and ‘F’ and you’ll see people who don’t speak English waiting outside to see who comes out of which door. We use different languages. We use different alphabets. In technology, we use different encodings for glyphs and invent all kinds of other concepts in an attempt to standardise how we represent written text, and we still create numerous bugs and portability issues and lost-in-translation problems. We’ve been using computers for half a century and change, and we still haven’t standardised what the end of a line looks like. Or was it the end of a paragraph?

Now, certainly the simplicity of a text format has big advantages today in terms of things like searching for data and programmatic manipulation. But how much of that is just convention and historical accident? Right now, I’m typing this using an input device heavily optimised for text, because that’s what my computer comes with. If I want to input some graphical notation, say an equation, my choices are probably limited to using some awkward purely textual representation (TeX notation, etc.) or some even more awkward half-text, half-mouse graphical user interface. Neither is an appealing choice, which is why it takes those of us working in mathematical disciplines forever to type up a simple note or paper today.

Technology does exist that can interpret a much wider range of symbols drawn with a stylus or other pointing device as an alternative means of input, but usually as a niche tool or a demonstration of a concept. Until we routinely build user interfaces that parse freeform input and readily turn it into whatever graphical notation was intended, a lot of us are still going to reach for a pencil and paper whenever we want to draw some quick diagram to explain an idea. But I bet a lot of us still do draw that diagram instead of speaking for another five minutes to try to explain it.

Personally, I’m looking forward to the day when source control doesn’t show me a bunch of crude text-based edits to my code, but instead a concise, accurate representation of what I actually changed from a semantic point of view. But to do that sort of thing, we have to have more semantic information available in the first place, instead of relying on simplistic and sometimes error-prone textual approximations.