Building an archive of deleted questions

Question

Back when the first discussion about deleted questions came up, I promised to look into building an official unofficial read-only archive for good deleted SE content.

I had a plan to build something like this:

Each link would point to a 1:1 scraped Stack Overflow (or other SE network) deleted question (of course with a big fat disclaimer to the top, historical data, legal stuff, yada yada yada - and of course after asking Stack Exchange for permission, seeing as this would go far beyond what the CC license allows.)

I also registered a domain name that would (IMO) suit nicely, overflown.org.

However, when thinking about it, I realized doing this properly is a lot of work. You would have to:

• build a scraper that can use a 10k+er's OpenID to fetch the deleted question (using the API or the data dump is for wusses. If we go through all this trouble, we want a historical document of what the site looked like "back then", right?)

• edit the page's DOM to remove dynamic elements, remove the "deleted" CSS classes, remove all the voting abilities from the buttons, and add the "this is historical data" disclaimer

• fetch - this is the biggest part and applies even if you use the API or data dump - all linked image resources (even those from Imgur - even images used only in deleted questions could be purged one day), store them, and link them

• fetch all associated CSS style sheets and store them intelligently - preferably in a way that you don't fetch a copy of each file for each page (as normal page scrapers do), but also in a way that doesn't break if SE update their style sheets

long story short, I won't be in a position to build this any time in the foreseeable future - with old work, new work, and studying, I really have enough on my plate already.

Of course, if somebody else in the community feels like giving this a try, cool! But what I was really thinking is - maybe this is something Stack Exchange should build after all?

I know Jeff's official stance on these questions is used to be

@system I did think about a sweet place for them, in the bitbucket. – Jeff Atwood♦ Dec 29 '10 at 1:54

but come on. Think about it:

• Future deletion discussions and changes in on-topicness policy would become way more relaxed because the popular and loved stuff has a place to go.

• A site for "overflown" questions could be curated by selected community members, who decide whether a contribution deserves to be archived or not.

• A separate domain name (I'm happy to donate overflown.org for the purpose if desired) and the separate site design would create enough distance to the Stack Exchange network: it would be crystal clear that this read-only archive is for historical purposes only, and the existence of a question there does not mean you can ask something similar on one of the live sites.

• The pages would be 1:1 scrapes of what the site looked like at the time of the scraping, making it a general historical archive as well.

• These questions traditionally drive a lot of traffic. The possibility of continuing to show ads in the usual places would allow Stack Exchange to take advantage of that traffic without endangering content quality on the SE network.

• Hosting an archive like this is a job for a business, not us regular joes. A private project hosted by some community member can go down once they forget to pay a bill, or decide they want to password protect their private blog or photo album. The ideal place to host a site like this is at a company with professional admins that intends to stay online for a long time.

How about it?

Answer 1

No, we shouldn't do this.

Deletion means deletion, and needs to stay. That something was once "on-topic" shouldn't protect it, it's become a broken window.

Most of the cited examples aren't really "on-topic" for that matter, every single example in your screenshot wasn't on-topic back in the day it just wasn't closed/deleted. This was because moderation was still evolving, and so the rules weren't as stringently enforced; they still existed though.

See the very first FAQ on Stack Overflow.

What kind of questions should I not ask here?

Avoid asking questions that are subjective, argumentative, or require extended discussion. This is a place for questions that can be answered!

Also, try to refrain from asking questions about Stack Overflow itself unless you absolutely, positively have to. Most programmers don't come here to learn about the intricacies of Stack Overflow; they come here to get answers to their programming questions. Let's try to help them out by not cluttering up the system with navelgazing meta-discussion.

Most popular programming cartoon, first home computer, programmer's last words, etc. are all at least "subjective" and obviously had "extended discussion".

Restating because it's important, these should have been closed or deleted; that they weren't is a failure of early moderation, not of changing "on-topic"-ness.

We can look to similar sites for opinions on this as well, for instance Wikipedia says:

Deleted pages should be visible

Proposal: Make deleted pages visible to everybody (or to all logged-in users), not just administrators.

Reasons for previous rejection: That defies the whole point of deletion (which is to improve Wikipedia quality by getting rid of the worst parts); also, it would increase the workload of the Oversight body to ensure that copyviolations and libelous statements are not visible to everybody. This proposal has been explicitly vetoed by the Wikimedia Foundation legal counsel.

I honestly hadn't considered the legal arguments until this was brought up to me, but it is a valid concern. There are lots of instances of SSNs, emails, phone numbers, and so on getting hard deleted via flagging; who knows how many were only soft deleted. Copyright violations and libel are outside my expertise, but presumably do need to be addressed.

To address Shog's proposal in particular (though the above objections apply equally to it):

Direct links to Archived questions will always work, displaying them in the same form that they appear to 10K users right now.

A new search operator - archived:1 - will allow searching Archived questions.

These two bits heavily undermine the usefulness of deletion. It's a pipe dream to expect new users (or even moderately experienced ones) to really grasp the distinctions around archived questions. People still use closed questions to justify asking similarly off-topic ones after all. That the question is visible on a site that is semantically connected to Stack Overflow/Stack Exchange is enough to confuse (or let them willfully delude themselves, as my more cynical side sometime suspects) people.

I'll ignore the technical concerns, though they're non-trivial.

Answer 2

YES PLEASE!

I've created the Deleted section on StackPrinter and the Popular deleted question list here on meta because I really really love those questions.

But StackPrinter is just an ugly crappy patch and those questions deserve surely something better!

Regarding Jeff's position, look at the comments of this recent blog post:

Me:
The only big complaint I have is the absence of a proper place for all these fabulous deleted questions that are part of the history of this community and a remarkable pieces of culture on the Internet; in my humble romantic opinion, they deserve something better than a deleted flag in your beefy database (Pls,do not start with the datadump yadda yadda ^__^).

Jeff:
@system I agree with you, and I think these “classic” deleted posts need a place to live — either on a site specific blog or something else.

EDIT: If you don't want to do this, please implement this feature-request on StackApps; the Stack Exchange API should return also the deleted questions allowing the user to easily build an archive of deleted questions without using the heavy and hard to use data dump.

Answer 3

I've spent a fair bit of time thinking about this now, listening to suggestions here on MSO and elsewhere, and watching how the Historical Lock feature has been used. Two observations:

1. No one really wants to curate this stuff. Forget turning a "classic" thread into a blog post; truth is, few even care enough about most of this stuff to make a strong case for its preservation in comments or here on Meta. Building a "best of Classic Stack Overflow" gallery, full of hand-picked, lovingly-preserved examples is a waste of time - the historical lock function can serve those rare gems that do offer something, but the vast, vast majority of deleted questions - even highly-voted/viewed questions - will never make it in.

2. The primary value of preserving the past is in avoiding the same mistakes in the future. This is the opposite of the "broken window" problem: if we delete all the most egregious examples of how certain forms of questions can go terribly, tragically wrong, then there's nothing to point to when they come up again. This is particularly important in the case of questions that aren't actually bad but simply obsolete - Meta provides a wealth of examples here: bug reports on features that no longer exist, retag requests long ago satisfied, merge requests for tags that were later synonymized... There's value here if the topic ever comes up again, but most of the time it's just noise.

With these two factors in mind, an obvious solution presents itself... As Maxim Zaslavsky wrote a while back:

The Gmail archiving model can be applied here, such as for old bugs and feature-requests that are [status-completed]. There should be a viewing mode that hides them from search on request.

This makes a lot of sense, and not just on Meta. Therefore I propose the following form of "archive":

• Questions that sat on the site for at least 60 days, with a score of at least 3, deleted by someone other than the author, will be considered Archived.

• Direct links to Archived questions will always work, displaying them in the same form that they appear to 10K users right now.

• A new search operator - archived:1 - will allow searching Archived questions.

• Archived questions will not appear in normal question lists, search results, user profiles, or (by default) reputation histories. They will behave exactly like deleted questions apart from the behavior described in the previous two bullets (no voting, answering, commenting, etc).

• Archived questions will, so far as is possible, be hidden from external search engines, and scrapers.

• Archived questions will require either a separate flag in data dumps, or be released separately in their own data dumps to avoid polluting the primary dumps (this is a separate discussion).

This would preserve access to these questions for those who need or want them, without leaving broken windows, "framed garbage", or noise to pollute the site that the vast majority of members live and work on. It would also avoid adding maintenance costs for a separate archive.

Answer 4

For all the anti-fun purists out there who have an irrational hatred of all the silly, interesting questions that were asked and answered in the old days of Stack Overflow, I'd like to point out that this could potentially be a huge non-trivial moneymaker for Stack Overflow.

A single location with all the most fun content from Stack Overflow would likely attract a lot of eyes. Surely Jeff Shog and Joel could sell some ad real-estate here, right?

If you guys don't want to do this for the right reasons, at least do it out of old fashioned capitalist greed :)

Answer 5

One thing that the ineffable Jason Scott just pointed out to me on Twitter:

if you give the url of the page to http://liveweb.archive.org and wait five minutes, it will archive that page. How about that?

Also you can enter http://liveweb.archive.org/http://www.website.com/page to have it happen without visiting the page.

So if you want to ensure that a popular soon-to-be or may-possibly-be deleted question gets archived by the Internet Archive, manually feed them to the URL above.

I suppose for already deleted questions, we could also undelete, wait 5 minutes, let it archive, then re-delete.

Answer 6

I just implemented an unofficial deleted question archive, without knowing you had already suggested this.

My archive is based on the historical data dumps - it includes everything from previous data dumps that is not included in the latest one.

Answer 7

YES!

I've been using StackPrinter to view old deleted questions and it's really tedious, especially because they don't look much like SO, and half of them weren't on StackPrinter before they got deleted.

Some of these questions and their answers are absolute gold and deserve a good home. Please do this!!

Answer 8

There's a serious need to preserve the content, but I don't like this solution (at least as I grok it). Here's why.

Right now, deleted pages show "Page Not Found" to everyone except 10k+ users (IE, just about everyone). The site blogs, external sites, etc with links to now deleted questions are broken. SE doesn't look cleaner when that happens... it looks dumb.

Would the "Page Not Found" errors turn into redirects to this new archive location? ... if not, what's the point? The links are still broken. If so, it sounds like a lot of work for something with a very simple solution.

Right now, the problem is a semantic gap. SE is failing good content, contributions from the people who helped build the community, and it's dropping the ball on a social responsibility to not break people's links into the site.

This gap is caused by disappearing the old content because it has no defined place to fit in, that's it.

The solution seems simple to me... just don't break the links. A question flagged "Deleted" should disappears from the lists, search, rep is removed, etc but the LINKS themselves should still work. Have some banner showing it's deleted and why, and possibly create a special kind of archive page where the links are all dead.

Who cares if the link still works, if in every other possible way it's disconnected from the site?

This is purist nonsense. You're definitely pushing some evangelists away with this insanity.

EDIT I was shown Pekka's Show battled deleted questions to the public question. I like that one more than what I'm suggesting... of course spam questions still need to go away. Although, if you do that to J. Random Newbie, please remember to try to find some nice way to say it.

Answer 9

There is an extreme need for something like this.

We come across a lot of questions which are essentially Good questions only find the next day that it is close. The usual pattern about this is that almost always they come from some real experience and they also invite a great traction with other people pouring into insight. However, when moderator looks at it - and who need to look at the nature of the question, it may find off-topic based on definition. Now this is not to say that one should preserve the off-topic questions because that would destroy the quality in the long run; however, shooting off the good off topic questions we are destroying great value right away.

Overflown.org will indeed solve a great deal of this problem.

Please please please do it.

Answer 10

Yes, please

I've come across too many links on MSO, which lead to questions that are good examples and help make a point--but they're deleted. Aside from that, there are too many fun-looking question links that I see which are deleted.

I really feel that deletion is a bit too harshly applied--it should be carried out on worthless posts, not off-topic ones. The other posts do not follow the site rules, but they're good enough to be there.

I understand that a relaxation of deletion rules is not possible, so why not do this?

• Have two levels of delete, one for the questions that have no value at all, and one for questions which have value, but are off topic. Technically all questions have value as examples, but let's not consider that.

• The "soft delete" shouldn't give the deletion 404, it should instead say "this question was deleted", in a non-404 manner, and link to a stackprinter-style archive.

Answer 11

Other.SE

I proposed other.stackexchange.com here, which could house these types of questions, without 'reducing the quality' of the other SE sites (an argument with which I disagree). If moderators and the overall community is going to say 'this isn't welcome here', then let's make a place where it IS welcome. If you agree, join the proposal.

Answer 12

No, we should not have access to the history of anything deleted. It means deleted for a reason. Which means that it being deleted is more beneficial than being able to look back at it.