Most of a modern HTML page is dynamic scripting and tag gibberish. It would be cleverer to use AI to copy out the actual content X rendered at moment Y from place Z (we allow for location and geo tags), and store it in versioned, indexed Asciidoc, which is much more concise and directly human-readable, and then merely render it in HTML. Then you gain the transclusivity, for all webpages, without as much compression and storage being necessary, and with the content directly legible. But, it's a legal wormhole I don't want to enter. I'm aware that we've created some amalgamation of WikiLeaks, InternetArchive, and Wikipedia. Basically, a monster. I hardly sleep, anymore.