I really want the underlying files to be human readable and universally pareseable.

While Logseq uses Markdown files, block references are not meaningful in their pure Markdown form. There’s probably some sort of Markdown evolution we need to align how different applications that use Markdown files can share a protocol (thinking about Obsidian and Logseq here especially).

If all goes well, we are probably heading towards making Markdown files a “human-readable database”.