Another problem of the present-day WWW is, even archiving all the data is far from enough to preserve the history! That's because the Web has a dual-role, (a) a

Another problem of the present-day WWW is, even archiving all the data is far from enough to preserve the history! That's because the Web has a dual-role, (a) as a protocol, or a medium of communication, and (b) as the software, or the user-interface. A good history preservation should allow you to somehow "browse" it, as if the historical system is still alive. How the website worked, how it was used, that's all parts of the history. If old operating systems and programs are preserved, there is no reason not to preserve websites in this way. Back in the old days, many systems are federated and/or distributed, which means the software and the protocol are two separate entities. You use a newsreader, which talks the NNTP protocol to obtain news from a Usenet newsgroup. If you want to preserve history, you can (a) archive the newsreader program with source code, and (b) archive all the data on the NNTP server. That's exactly what has been done already, if you load a Usenet archive to your newsreader, pretty much you would have the experience similar to how Larry Wall browsed the Usenet back in the late 80s, at worst you need to rewrite a compatible "mock" server, but that's all. On the other hand, little of the early BBS systems have been preserved, once the server is gone, everything is gone. The transformation to the web, means now the platform (a web community) = protocol (backend database format) = user interface (HTML/CSS), they're all tightly coupled together. It creates several problems: (1) The "internal state" cannot be archived. A website is a system with constantly updating parameters, and often they are not stored. Simple examples: (a) On Hacker News, I cannot see what was shown on the frontpage yesterday retroactively, (b) A user changes his/her avatar, now we had no idea how the old avatar used to look like, and (c) an early user has been banned from the forum, now his/her personal profile is inaccessible, (d) on some social media platforms, sometimes a old post may be raised from the dead by renewed interests (look, how stupid this comment was!), and now suddenly it's flooded by new posts, leaving no trace of how it used to look like. (2) The "reader/user interface" cannot be archived. You must have seen something like this: You changed the website frontend, superficially, lots of "conservative" users complained, but the point is: now the old frontend and its "look-and-feel" is lost. If it was a simple CSS file, there are chances to bring it back, but if it was a major rewrite of frontend code, now history is gone forever. And in the lifetime of a website, the design and architecture is likely to be changed many times. As a result, even if a website and all its content is still alive, it may already be a shadow of its past for a long time, don't even mention to preserve it! And currently, there are two ways to archive the web, both are flawed: (1) Preserve the HTML at the surface. It's good for single pages, but you cannot browse a website in this way at all. None of the button on the website would work. (2) Preserve the database. For example, using the API to save posts, or dumping the database - the frontend and reader are not preserved. Using Hacker News as an example, now every single post is archived, but it's far from a full experience, at least you should be able to click someone's username and see all the posts. Now more and more websites are powered by JavaScript, makes the problem even worse. You are now literally running a program on your computer without any control over it. Once the platform is gone, no archive can save you. What is the solution? I guess there's no full solution, but there are some possibilities: (1) Wikipedia-like websites already have builtin version control, but it's very difficult to browse the historical version of the entire website. Systems like this can improve the frontend / user interface to allow a user to "lock on" a historical date. (2) When building an all-Javascript website, spend some energy to build a plain HTML version as well, it may help avoiding the upcoming digital dark age. (3) If you are going to close a website, perhaps it may be a good idea to make your internal backups of database and codebase at different years publicly available with sensitive information removed, and allow everyone to setup and run a replicated version. It's infeasible for a big website, but it may be a workable idea for a small community. And I can imagine the archeologist from the 22th century digging into the old backup tapes of Reddit and attempt to rerun the system. But ultimately, it's a problem that is needed to be addressed by the protocol and software with archive and preservation in mind. --- BTW: A few weeks ago, I've written a lengthy comment on the fundamental conflict of history preservation and personal privacy, using the Usenet as an example, you may find it interesting. * https://news.ycombinator.com/item?id=19562650 https://news.ycombinator.com/item?id=19562650