Hacker News Reader

Funny that we feel so different about Markdown.

The whole point of Markdown (in my opinion) is that the formatting is a function of text that you can see and edit. The entire state of the document is visible, and there are no formatting operations you have to learn, just regular text editing operations like backspace, copy, paste, and so on. For example, if you press backspace, you always know exactly what the result will be: it will delete the character to the left of the cursor.

This is not so in rich text editors; when you press backspace, it changes both the document and invisible internal state, and you can't easily predict what will happen simply by looking at the screen; you need to be aware of the invisible formatting structure, or perhaps a recent history of things you did, or it might be entirely impossible to tell because it depends on some invisible state that was created weeks ago and saved in the document file.

For this huge advantage in predictability and simplicity, Markdown sacrifices formatting power (you can only do a limited set of things like headings, lists, bold, and italic).

That's why Markdown tools that hide the markup are, in my opinion, the worst of both worlds: you sacrifice formatting power, and you sacrifice editor predictability. After the markup is gone, how do you edit it? How do you unbold something, or change the indentation of a list? If there are separate commands for all that, then there was no point in using Markdown.

Markdown isn't just text though, it also does images. Which is when the argument about rendering falls apart IMO. I really would like to see the image i am working on. Describing a chart/graph when all i see is ![Important](blah.jpg) isn't really that great.

> there are no formatting operations you have to learn

I see it as the opposite. If you give a 2 panel markdown editor to any non-developer human they will be baffled at how to do any formatting with it. Rich text editors are intuitive, that's why they've dominated computing for the last 30-40 years.

> you can't easily predict what will happen simply by looking at the screen

Huh? In rich text you see what everybody else will see. In 2-panel markdown you need the second panel to see how your final text will look to the rest of the world because you can't predict that easily when writing in an esoteric markup.

> For this huge advantage in predictability and simplicity, Markdown sacrifices formatting power

I wouldn't call requiring a second panel to preview how the actual text will look predictable or simple. It's a rube goldberg machine. On the other hand, I think the best thing about markdown is the restrictions to formatting. It forces you to write in semantic HTML.

IMO, Notion, for all its sluggishness, nails this part. They provide the intuitiveness of a rich text editor, while restricting the amount of styling to a markdown-esque set of properties so what you write will be easily portable to the web.

The mentioned Bear and Typora solve this by showing you the "raw" Markdown at least when your cursor is there. So anytime you edit, you're editing in text, as you mention. But you don't need a second "display" panel to do that.

Both tools do it differently. Typora shows it when your cursor is there, but otherwise hides the Markdown. Bear always shows the Markdown but applies the corresponding style to it.

I think you express the precise reason why I prefer markup formats like Markdown, ReDoc, and LaTeX. I get to write markup which completely eliminates the need to click around menus and buttons or do app-specific keyboard shortcuts for formatting.

Rich wysiwyg editors also seems to have issues where the visible editing and internal formatting state breaks or behaves in weird ways. Like when you insert links and it automatically formats it as a link in some cases, and not in others.

Edit: Personally I always though the split view was okay as long as the compile times weren't too bad. Having a markup editor and then a view of the result is what I do for coding, so why not when writing stuff as well.

Typora is great. But I also tend to use it in "source" view where I can she all the markup.

The point of markdown is literally that things don't hide. Wanna know if that semicolon is included in the italic styling? You will see it right away. Wanna know if a link has an old URL or is pointing to a weird domain? You see it.

This alone makes it worth it. Forcing the whole thing back into a WYSIWYG-editor will just mean that markdown becomes unreadable. A bit like when people used MS Word to write HTML.

> I can't be the only one who believes this...writing in a 2 panel markdown+preview side-by-side view is just horrible and backwards.

It's called markup and a matter of efficiency. With richtext I need to mark the Text and click some button/menu-item OR learn a gazillion shortcuts for every little syntax-function. With Markup I use the same keys that I use for writting text, because the syntax is internal to the text, not some hidden external thing.

Of course it's possible to make a hybrid, something that displays richtext instantly while still working with syntax. But this is hard and cumbersome. Nonthless, Obsidian has it on their list as a longterm-goal, so maybe in some years we will see it.

> With richtext I need to mark the Text and click some button/menu-item OR learn a gazillion shortcuts for every little syntax-function.

Cmd-B for bold is exactly the same number of keystrokes as shift-8, and is exactly the same amount of cognitive load to learn. (You can also apply styling both on selected text and for text about to be typed, unlike plain text environments like this textfield I'm using now where you can't do shift-8 on selected text.)

When you get into the more esoteric syntax, it's still a tie: look how often even Gruber gets the Markdown link syntax wrong, and he designed it.

Have you tried https://www.craft.do/ ? It’s iA Writter mixed with Roam mixed with Notion all Native and best in class writing experience

FWIW, Joplin allows you to have a single panel "raw markdown" view, and you can just assume it'll all render the way you want, and maybe double check at the end. I find that works well for me. As you say, previewing what bold looks like offers minimal value. :)

Joplin also has a beta single pane WYSIWG view which is like, I believe, Typora or Bear (markdown in the backend, but you type in what is effectively the preview window). I haven't really tried it out yet though. If it works well though, it might solve your concern though.