He's probably talking about a video that's twice the size of the SNES's resolution.
Due to the way video is compressed, double resolution will make it possible for the resulting video (that is downloaded, not streamed) to look almost exactly like the original video output. SD encodes, which are at the same resolution as the SNES, have bleeding pixels due to
chroma subsampling.
However, video cards normally perform bilinear smoothing when resizing videos, which means a double resolution encode will look fairly blurry when playing it in full screen on a high resolution monitor. Bilinear smoothing works well on films and drawn animation, but 2D games look best in their original pixelated form or when resized with a pixel art scaling algorithm (like hq2x, which is normally not available when playing back video).
Additionally, YouTube does not stream videos in their original resolution (unless the uploaded video's resolution is beyond HD), so the downsize from double resolution to 480p will suffer visibly from resize smoothing (plus some issues with chroma subsampling).
tl;dr: the first sentence in this post is the answer to the question above.