> easier to write without using tags

...until you want to write any tabular data. The Markdown syntax for tables might look pretty when you read it, but the only time I've ever bothered to produce it has been with the help of a WYSIWYG Markdown editor (e.g. Typora). And I actively recoil from the idea of modifying one (since that could potentially change the width of a column, requiring carefully adding/removing spaces from the other cells in the column... ugh.)

Meanwhile, I've been hand-rolling + hand-modifying HTML tables with ease since the HTML4 days. HTML tables do not emit ugh-fields. (With the tradeoff that they do not visually read as tables in their source-code representation. But they still do read as cells grouped into records — and for most cases, that's almost as good.)

And yeah, sure, for embedding wholly-plaintext data tables, you could just drop some lines of CSV in a quote-fenced code block. But that doesn't work if you want that table to render with MD semantics itself, such that you can style the contents of the cells using MD syntax. Think "links to additional resources in Github READMEs."

IMHO, a format that's mostly just Markdown, but which deprecates MD's existing table syntax in favor of a much-simpler-to-write "sigil-annotated list-of-lists" syntax, would be a big hit.

(And, now that I think about it, would also probably make it simple for LLMs to spit actual headered tables at you, rather than relying on endless hierarchical bulleted lists.)