jim said:
>   Leading and following underscores are not plain text. 

sure they are.

indeed, the underscore even falls in the 7-bit range.

so it's as plain-text as plain-text can be, and it has
a long and glorious tradition of indicating emphasis.

There are many long and inglorious methods of indicating emphasis in plain-text including *asterix* and SHOUTING and _underscore_ and <i></i> and [i][/i] and they all suffer from the same problem: They are all not what the author wrote, at least not as implemented by the typically concurrently existing publisher.  Now say 100 years later PG says ignore those previous efforts we as the publisher of this day knows better than the original intent so we will substitute something else for what was actually printed.  Now if someone really only has a 7-bit teletype to print their PG on, then I can understand this.  I can also understand PG’s desire to continue to support such teletypists [[I tried using one when I was in college which tells you how old I am but it kept overheating and burning out based on my demands]]

What I don’t understand is why PG continues to be wedded to plain-text as an *input* encoding format demanded of people submitting texts to PG.  Plain-text is too constrained to do the job well.  HTML is too ambiguous, and too ill-matched to books to do well.  We need something else, something that CAN be correctly and automagically converted “correctly” to one or another formats including plain-text, and Unicode, and HTML, and  mobi, etc.  And something that allows the simple every day tasks of the encoder, including italics and m-dash and poetry, titles and chapters and subchapters, publisher info, dates, etc to be handled correctly and easily.

PS: Bit curious which blind reader handles _the underscore “convention”_ correctly – I’ve not seen _that_ one!