Think of TVersity. It's a software application that takes one source and sends it to another device as another source. It's not a very good re-encode. That's what Google does.
*facepalm*
The differences are a little bit more specific than you're trying to make it out to be.
A. TVersity has a different goal. It attempts to re-encode a file quickly enough so that you can play it in real-time. Thus, the quality is reduced in exchange for speed.
B. It's still a re-encode, the only difference is is what you're doing with the end result. TVersity's only keeping the encoded file around long enough to play it.
C. Google re-encodes for different reasons. They do value speed, but they value space far more. Thus, the reason for lower quality on YouTube is to keep the filesize down, both for themselves and for the user. TVersity doesn't need to worry about this since almost all use cases involve LAN connections.
You wouldn't re-encode a video that's already encoded in X format if you need it in Y format.
Yes you would, because you would have to. You can't just plop X into Y and expect it to conform to Y.
If you are trying to make a point with this, be far more specific with an example next time. Give me codec names and such.
Why avoid it? You're talking about something no one else is talking about, lol. I'm talking about the source video, not taking a re-encoded video made from source, then using that video to "stream" it. That sounds ******ed and isn't mentioned anywhere...
I don't even know what you're saying here but let me just make it clear, this IS what Google does:
A. User uploads a source video.
B. Google re-encodes the video (the source is already encoded in one of a bazillion different formats YouTube accepts) into 360p, the default YouTube format that is most compatible.
C. For every subsequent video format after that, be it 480/720p, mobile, HTML5/WebM, etc, they take the source video and re-encode it to the formats desired.
D. As Google desires new formats, they instruct their render farms to re-encode from the source files any additional formats they need. They do not touch the ones they've already encoded, as there's no reason to (except to remove if it's been obsoleted).