Markdown is fine, but I can't understand why you would want to render it in the browser. Why get the client to do the work with every request? Just render it once on the backend and deliver the html.

Sure, it's "fine" - it's just unnecessary, less portable, less future-proof, and more complex to render (as this article demonstrates). What's the point? Why opt for that complexity when you don't need to? Just to save a very low percentage of key-strokes? That strikes me as being penny-wise & pound foolish. And much as most text is fairly bare-bones, sometimes you do want extra layout control - and then markdown is really in your way.

I don't see why it's either less portable or less future proof.

It's a different markup. One that many people find easier and more natural to write. Of course, different people have different opinions on that.

For people who find writing and reading markdown nicer than HTML it makes sense to use markdown as the base format that you write everything in, and then convert to different formats as needed (html, pdf, whatever).

For you, HTML may play that role. Personally I find HTML ugly to read and write.

I agree that markup is easier to write - it's shorter. And typical html is much messier. But if you wrote html the way you did markdown the difference is much smaller. In fact - look at the OP's text. That article in html would be barely longer.

To summarize: markup is better to read+write. But the difference is almost zero. Anything other than trivial markup is a mess; exceptional layout often impossible (and I value the ability to creatively use layout to get my point across). It doesn't support wysiwym editing (very well). It isn't trivially statically hostable - and if you try it's fragile and flashes uglily. Many flavors exist that are slightly different.

I prefer flexibility, consistency and simplicity over micro-optimizing source readability.

My blog originally lived on fully static hosting. When I moved to github I made minimal changes.

I also like the idea that the "source" is right there on the page rather than a separate "API" - the dream of XHTML+XSLT. When I originally set it up I wanted the page to be readable using Telnet - which I think it is, more so than one rendered on the server would be. I couldn't tell you now why that seemed important though.