[dragontamer]

Clearly, a new "internet community" needs to be formed. We've got a lot of classical communities: forums, Q&A (aka: Stackoverflow), link-aggregators + comments (Hacker News, Reddit), mailing lists, IRC, Discord, Github, Wikis... to name a few.

StackOverflow is trying to build an FAQ database. The Q&A format isn't really about answering questions, its about collecting questions that are popular / drive traffic, and to focus the community efforts on those popular questions.

------

I've hypothesized before, and I'll hypothesize again, that StackOverflow needs an archival process. A long time ago, I used to play a webgame called Utopia (and its "sister" game Earth 2025)... every 6 months or so, the world would be paused indefinitely, and a "new game" would be started.

6-months is too quick for Q&A formats, but I think a rolling "pause" cycle would be great for the StackOverflow system. Every 2-years, the Q&A database would be paused, all unanswered questions would be wiped out... all solutions permanently archived as "The winner" for those years.

For example: 2010 through 2012 would be one "cycle" of StackOverflow questions / answers. The question would have to be re-asked / re-answered in the 2012 through 2014 years (and the community can "reask" important questions every 2 years to ensure their continued communication).

The cycle of life / death of questions has caused the StackOverflow database to become too large. I think it needs to be archived, torn down, and rebirthed every few years. Its too much to expect an answer from 2012 to really remain relevant today in 2019.

Its too much to expect newbies, today in 2019, to know that some question was answered in 2010. Its too much to expect old moderators to comb through old questions and "update" wiki answers with information in 2019 to "fix" problems from 2010.

Case in point: modern compilers use cmov and avoid the branch-prediction question. This question is no longer relevant on modern compilers: https://stackoverflow.com/questions/11227809/why-is-processi...

Yes, its important and tells an important tale about how compilers worked back in 2012, and yes it should be archived. But time has moved on, the world is different now and the Q&A Database is failing to keep up with the changes.

[xingped]

I'm not sure if this is what you're suggesting, but if you mean to say that non-accepted solution answers to questions should be wiped out, I don't think that's a good idea. Many times my question might be roughly related to the question I landed on, but not exactly and a non-accepted solution is actually the answer to my question rather than the accepted solution.

[dragontamer]

Oh, I mean the myriad of questions with no answers at all should be wiped out. If no one was able to answer a question within the 2-year timeframe, it needs to be "re-asked" for the next cycle. No point in archiving something like that. And its a lot of effort to expect newbies to search all of the unanswered questions before posting questions of their own.

The "wrong" answers (or unaccepted answers) serve a point in the discussion, and in many cases, "wrong" answers are more helpful to me. But the "voting period" for these answers should be closed at the "end of the game".

[gitgud]

Questions with no answers are still useful, if I solve a bug, I'll try and answer the unanswered question for future Googler's (including myself)

Information shouldn't be removed it's just ranked badly in SEO as it's not as useful.

Stack overflow is quite a fine tuned system, I'm not sure there's much you could do to make it better, without sacrificing something else on the platform.

[dragontamer]

> Questions with no answers are still useful, if I solve a bug, I'll try and answer the unanswered question for future Googler's (including myself)

Under the "2-year cycles" model, you'd simply recreate the question, and then self-answer it. There's no reason why you should be searching the archives to know if a question was useful some point in time in the past: if you think its useful, then self-ask and self-answer.

If someone thought that a question from 2011 was useful (but is up for deletion), then they can simply re-ask the question in 2013, 2015, or whatever future date. By "refreshing" a question, you will get better community opinion for which questions are popular enough to deserve more attention.

The good thing about the 2-year limit is that moderators will no longer delete your question for "duplicate" anymore. So you have better assurance that your efforts won't be wasted.

[gitgud]

Yes, but answering a unanswered 5 year old question still brings benefits to the community, I just don't think it warrents deletion, as it won't be ranked highly unless you specifically search for it.

Another reason to preserve the question is for context. Many of my problems have been solved by discovering that someone else had the same error message, but they had different circumstances which leads to clues and possibly discover the root cause, this then could allows you to answer the ancient question (even if it's no longer relavent to the asker).

There's nothing worse than googling an error and having zero results show up...

[bryanlarsen]

I've got both the accepted flag and a bunch of upvotes on answers I posted over 2 years after the question was originally answered. Often it was the question and the discussion that helped me find the actual solution that I was then able to post and snag a necromancer badge for.

[Franciscouzo]

Questions that have negative score and have no answers are automatically deleted after some time.

[ahpearce]

An additional feature suggestion: if you really wanted to add value, require version and software that the question relates to, and track those against release repositories. That way, when you're looking at some sort of bug resolution that applies to version 2.x of 'some package' from 2014, and you're 5 years in the future on 4.x, you can use that as a heuristic that "maybe this doesn't apply to me".

[jolmg]

> StackOverflow needs an archival process

https://archive.org/details/stackexchange

[cannonedhamster]

Man the feels. I played the heck out of Utopia. But yeah Utopia's internal kingdom forums would get wiped out every season. It would take a little while for the Crystal Ball app to update. I think an archive system for the Q and A format isn't a terrible idea but you could take it further and build actual knowledge dbs that collected the best info sources and paired them with relevant books and tutorials where appropriate.

[stevenwliao]

We can get the most relevant answers on top if the algorithm decays voting power for historical votes. Fresh votes will keep relevant answers at the top.

[dragontamer]

There's no way to automatically know if an answer from 4 years ago is still relevant today.

Maybe it is, maybe it isn't. Trying to tie an algorithm to "age decay" is the wrong solution, because it will decay too quickly for some questions, but not quickly enough for other questions.

[stevenwliao]

It's softer than the original proposal: instead of archiving an answer (reducing vote weight to 0) at an arbitrary time, we do it gradually at an arbitrary rate.

[BrandoElFollito]

I am not sure about the cycle you describe but purging some questions is a good idea.

I flagged for deletion some of my own questions which were bug discoveries and are not relevant anymore. Or ones which deal with a version long gone.

[ghostbrainalpha]

What you are describing would be insanely hard to get right, but its exactly what we need.

[dragontamer]

The best part about cycles is that you don't have to get it right on the first cycle. There's always the next cycle to improve the rules.

[yabadabadoes]

I think the precise problem of letting this version go into a new form of volumes, is giving the people on meta a task of "improving" one set of rules.

I think the current SE site questions/answers should be prefixed, new namespaces should register with new COC/TOCs and we should allow a free for all of linked questions, answers, comments, etc with people deciding their own web of trust for what portions of the Q&A web they want to see.