Always Bet on Text (2014)

Even if you include things like circuit diagrams as text I don't agree that you should always bet on text. Personally I think you should always pick text as a default, unless you have good reasons to chose something else and use text to elaborate on it.

Couple of examples:

- For a musician to convey a musical idea, the arguably most straight forward and effective way to do so is just to play/sing/record it. Any other means like musical notation, textual description, midi etc. require a lot of additional knowledge and work on both ends of the communication and come with their own drawbacks.

- The Wikipedia text about human rights in the article is a great example when written text is better than an image. But say I want to describe what the grand canyon looks like. Personally, I don't have the writing skills to really do it justice. On the other hand my phone has a camera so I can snap a couple of pictures that will get the other person a much better idea of how it is. In this case text is useful for additional information not visible in the image, like why the color of the rocks is how they are.

- If I'm living together with someone and want to discuss how to pain the walls, a pantone color swatch is going to be much more useful than textual description of the colors. Text is useful here after the decision to pass on the color code to the contractor.

- I reckon for most people it easier to follow instructions on how to cut up a chicken when they are in form of a video or image slide show over just written words.

I don't think the problem is whether text is better or not, but that often the choice of medium is not based on what makes most sense for the use case, but other considerations like what will yield better conversion numbers.

What does the display (or TTS) technology used to read it matter in this discussion?

Obviously you need ways to display text in the physical world, be it a screen or a piece of handwritten paper. It's still text.

> SVG is definitely human-readable.

No. A path tag with a 500-600 element `d` attribute is not human readable. It is not human understandable or parsable until it's converted into a visual format.

Ask a blind person whether they think SVG charts are accessible on their own without any additional markup or alternate explanations or controls. Ask them if they can read SVG line charts by stepping through the path property. Figuring out how to communicate complicated graphs/charts with lots of data without relying on any visual medium is a complicated problem that we are still trying to solve.

> You could just say "I'm sending you the stylised image of a blue bird, with the instructions on how to draw it using the SVG text standard". Or just, more clearly even, the following string of characters: "twitter_logo.svg"

By that logic, a PNG is also text, because I can do the same thing. All images can have alt tags. Even Javascript Canvas can have fallback text for when it doesn't render. My Twitter png can be transmitted as "twitter_logo.png".

What do people mean when they say text is preferable if not that the textual representation is more useful than the image representation? If you didn't feel like the SVG format was adding any useful information in its rendered form, you wouldn't waste the time sending information to render it. You would just send the alt text.

> I can read it to someone over the phone with a pen and graph paper, and he can draw it exactly like it should be represented, with zero loss of quality

Right, and when that person draws it, they will have converted it from a textual format into a visual one. I can send you the individual pixels of a small raster image over the wire in a textual format and you can sit down on a piece of grid paper and ink in squares until you come up with the final image. At which point you will no longer be looking at a piece of text.

To argue that the fact that we can encode a format over a UTF-8 stream or physically replicate your computer's drawing algorithm makes a thing text seems to me to be a misreading of the original article's point. The original article is not complimentary of Twitter's SVG, it explicitly brings it up as an example of a non-text medium.

If you expand the definition of text to include basically any storeable information that can be theoretically encoded into UTF-8, then sure, I would agree that text in that system is tremendously powerful. But that's not a useful category to talk about, it's so broad that it becomes meaningless.

Your reply is in text. Would you care to link a set of paintings to express your argument instead?