Previous discussion from 2 months ago: https://news.ycombinator.com/item?id=7579773

(honest question) is this a good thing? When would I want to use this?

To download an image asset. To download build that is checked in to github as part of the build process.

Doing it through the github web interface takes a lot of clicks.

Also useful when inserting scripts in a dev console.

In making demos on jsfiddle

the real question is, why Github doesn't do this directly?

To prevent using them as a file host. If you can just dump files (of whatever size, it doesn't really matter) onto github and serve them natively across the web then they become a file share or a CDN. They'd have to seriously lock down the rules on publicly available free repos if that happened (and they'd have a copyright nightmare on their hand as well). Whereas if they force git into the loop on the client end then it keeps them in that repo hosting butter zone.

In fact I wonder why they do not have the problem yet.

It's too much work and there are better alternatives. Big files just won't work and wouldn't be superior to torrents. Small files would work but compared to imgur, flickr, youtube, pastebin, tumblr, blogger, etc. it's much more hassle to use github with no advantages.

probably because they don't want to become a free CDN for people

That is answered in the FAQ.

>They do this because serving raw files from a git repo is relatively inefficient, so they want to discourage people from using their GitHub repos for static file hosting.