It's one thing to have the data around for centuries, it's another to be able to view it. Some content (Flash) is already very difficult to view just a few decades after it was created, let alone centuries.

Who is going to be able to view a JPEG file 500 years from now? It requires not only the data, but also all the software infrastructure to be able to interpret this data. To me it seems a stretch that all software required to view it is still usable in a century, let alone 5 centuries.

So even if you keep these home pages around "forever", it seems unlikely to me that anyone wants to invest time in transcoding everything when the world moves on to different storage formats. So what you end up might be little more than a bag of bytes? Though I guess the textual content within the HTML still has some chance of still being readable.

It will be fun times for Internet Historians in 2496 when they try to compile the historical copies of gcc, cmake, nasm and libjpeg.