6

I have been a member of Stack Overflow for almost a year and a half. Recently I've also registered for SuperUser, Programmers.SE and WebApps.SE. But now I am getting a bit confused maintaining so many different accounts and sites. And more "same-type" sites are coming through the Stack Exchange 2.0 process. All these made me feel like I want to be a member of a more generic site. Just ONE site. Maybe stackexchange.com. So, I have a proposal. You may call it Stack Exchange 3.0. Please go through my full proposal about this new SE 3.0:

  • There will be just one site where anyone can ask questions on any topic and anyone can answer any question. But there will be still some universal criteria for closing a question. As someone has said, we don't want another yahoo answers.

  • The reputation and other user-specific things will be managed for the whole site.

  • Questions will be tagged, just as they are on the present sites. But tagging will have more important roles.

  • When a tag will be used in a certain number of questions (say 20k), then some process can be started to define the tag as a "Super tag." It will not be an automatic process. The community will decide whether a tag will become a "Super tag" or not.

  • For each "Super tag," there will be a sub-domain of the main site like "dotnet.stackexchange.com" or "webapps.stackexchange.com". Or, "Super tags" could have their own domains. The sub-domain/domain name may or may not be same as the "Super tag" name. There will be specific designs for each "Super tag" as well. In other words, these "Super tags" will become new sites like present different Stack Exchange sites.

  • Any user profile/reputation can be queried against a "Super tag" to get a tag-specific profile/reputation. Visitors to "Super tag" subsites will see tag-specific user reputation. For some privileges like chat and retagging the overall reputation will be used while for some other like closing a question will require tag-specific reputation.

  • Some special rules can be applied to these "Super tags" if needed.

  • Normally a question can be asked under more than one "Super tag"s.

  • There may be something like "tag inheritance." There will be a parent tag for each tag. So, [c#] and [vb] will have [DotNet] as their parent. [google] and [yahoo] will belong to [WebApps]; [baking] will belong to [Cooking]. The top level tags will have null. If this cannot be managed, a tag may have more than one parent.

  • When a question is tagged with a child tag, it will be considered tagged with all its parent tags in upper levels. So, it will be easier to organize different tags of the same type under one umbrella. And these umbrella parent tags have higher chances to become "Super tags."

  • Each user will be able to customize his home page and other pages by subscribing to one or more "Super tags." So he will just see specific types of questions, and the other questions will not bother him, keeping this Stack Overflow ideology in mind: Isn't StackOverflow becoming too generic?

If this scheme is implemented, I think the following problems will be solved:

  • Users have to keep track of many similar sites. I am now tracking 5 but wishing to register with some other. But it will be tough I think.

  • Users get annoyed when questions are closed for being "off-topic". Sometimes even migration to the relevant site is not possible due to the unavailability or immaturity of that sub-site. But in this new system, simply editing tags of a question will suffice. There will be no hassle of migrating questions from one site to another.

  • Sometimes people do not know which is the exact place to ask a question. There are many similar types of sites. I have seen same question is asked in different sites and responses are very much different. With an universal community, we may get better answers.

  • Choosing new SE sites is relatively difficult. The process is quite lengthy. And after the lengthy process still possibility is there that the new sub-site will not succeed. Then all the questions-answers and other activities are wasted.

I would like to have Stack Exchange as a global community for knowledge-sharing through questions and answers.

Update:
I think most important flaw raised in the idea is "Ambiguous tag". This happened probably because of my flawed [baking]->[cooking] tag inheritance example. I think it will not be a very problematic issue if users go directly to the sub-domains or "Super-tags". But when someone is asking in the main site, sometimes it IS a problem. Though don't know how much severe. Does anyone has any ideas to solve this? I have already stated mine in the comments.

Update 2:
I assume people are thinking that this whole idea is a completely different pattern comparing with the present one. So, backward compatibily is almost impossible. But actually backward compatibility is very much possible only with a few tweakings in the tagging. We just have to add some more tags to the questions and that will include them in the relevant "Super tag" or sub-sites because of the tag-inheritance. For example-

  • [stackExchange] : [programming] [development]
  • [programmers] : [subjective] [suggestion] [discussion]
  • [SuperUser] : [feature] [usage]
  • [ServerFault] : [server] [database]

The relevant tags for a sub-site or super-tag can be viewed if someone goes to the Tags section of the present sites. And if any of the tags cannot be applied, the Super-tag itself can be applied easily. So, the migration from Stack Exchange 2.0 to Stack Exchange 3.0 may take time but still possible.

Update 3: These days SE has tag subscription and filters. We can query reputation specific to tags. Though not same, the reason for my tag-inheritance is same as filters. Though filters just group tags, no inheritance complexity. Now maybe my proposal is somewhat half implemented. But I still want an integrated interface for all SE sites. Hope someday we shall get it.

22
  • Your tag inheritance concept is flawed. Baking can refer to a car paintjob process. VB can refer to pre-.NET era VB6. Commented Oct 21, 2010 at 10:25
  • 4
    @Tadeusz A. Kadłubowski Actually the examples are flawed, not the concept IMHO. Probably this can be resolved by choosing distinct tag names like VB.net and VB6. And also less common scenarios can be overlooked.
    – Gulshan
    Commented Oct 21, 2010 at 10:40
  • 3
    -1 because I disagree with the overall idea. Having one 'super-site' seems like a bad idea to me and would lead to utter chaos (some tags are very ambiguous, f.e. baking might also mean to fixate particle animations in 3D-Modelling). Commented Oct 21, 2010 at 12:45
  • 2
    @Bobby Not everyone will agree. And you have the right to disagree and downvote. But better to express logics. The ambiguous tag problem can be resolved easily by adding prefixes with the tab names.
    – Gulshan
    Commented Oct 21, 2010 at 12:52
  • 3
    @Popular Demand I think the overall reputation can be used for privileges. The overall rep indicates someone contribution to the site, while the tag-specific rep indicates the knowledge and reliability of someone regarding the specific matter. I don't want recipes from a web developer. I can not think of any special rules for now. But I am quite sure some situation will come imposing special rules for a specific community. You can guess a "Super tag" means a different community.
    – Gulshan
    Commented Oct 21, 2010 at 13:04
  • 2
    @Gulshan, now that, I do have a problem with. Use of rep for privileges is predicated upon some degree of familiarity with the subject matter. On a site like the one you're proposing, it would be possible to earn a lot of rep answering questions about programming and then use it to incorrectly edit/close questions about horse breeding, which is not okay. I suppose that could be worked out, though.
    – Pops
    Commented Oct 21, 2010 at 13:09
  • 1
    @Popular Demand Then the privileges can be classified. Some will be available with overall reputation. And some will be with tag-specific reputation.
    – Gulshan
    Commented Oct 21, 2010 at 13:16
  • 1
    @Gulshan "The ambiguous tag problem can be resolved easily by adding prefixes with the tab names." Who is going to explain to new users that they can't use the generic "baking" tag, and must use the specially prefixed one, which won't be intuitive at all? Are you just shoving all the tagging work onto the community?
    – Pollyanna
    Commented Oct 21, 2010 at 13:45
  • 1
    @Pollyanna When tags will be categorized as said in "tag inheritance" user will not tag questions as we do now.May be they will choose tags from a tree. Or when a tag is added, all/immediate parents can be shown to choose the right one. Or there may be special notes for the ambiguous tags.
    – Gulshan
    Commented Oct 21, 2010 at 13:52
  • 1
    @Gulshan - You are coming up with quite complex (and certainly non-user-centric) "fixes" for problems your proposal raises. The current system is already better than your proposal even without the flaws people are bringing up, but the flaws and their fixes are making your proposal worse - much worse. You have come up with this large, complex proposal to fix one small problem: " I am getting a bit confused, maintaining so many different accounts and sites " This is a significant over-reaction to a relatively small problem. Why can't you fix this with a custom site using the API?
    – Pollyanna
    Commented Oct 21, 2010 at 14:05
  • 5
    @Pollyanna To be honest, I am not a web developer. Even I know very less about APIs. So, I tried to find a way everyone can use. Not just the programmers. The whole idea was just not for solving the small problem you stated. I felt like "it would be better to manage it in this way". And I shared my idea with the community so that others can also think about this.
    – Gulshan
    Commented Oct 21, 2010 at 14:24
  • 1
    Under this system, how would we identify the questions that would fit in Super User's community? Computer software and hardware is partly covered by a lot of other communities such as Gaming and AskUbuntu; how is Super User going to establish itself while still getting the tightly focused scope that it desires? Since those other sites will use [hardware] and [software] tags, it seems less than usable to try and focus on those tags... but then what is left for Super User?
    – Grace Note StaffMod
    Commented Oct 21, 2010 at 15:56
  • 2
    I'd +1 for a very nicely put and thought-through proposal; but -1 because I don't like merging the sites. SO should have "tag communities" to avoid splinter sites from getting created (see e.g. testing.SE). Other than that, I'm fine with the separation
    – Pekka
    Commented Oct 22, 2010 at 9:33
  • 1
    @Pekka I think still there will be many sites, not just one. Almost same as now. And this whole idea can be implemented in the core without even notifying the present users. I am feeling that the "One site" is a misleading name for this idea. And everyone is making the same mistake.
    – Gulshan
    Commented Oct 22, 2010 at 9:45
  • 1
    I really appreciate the fact that you have really thought this through. Commented Oct 26, 2010 at 20:21

4 Answers 4

30

I think that the current system plays to a few critical aspects of why StackOverflow is successful.

Each community has a laser sharp focus with significant depth, and narrow breadth.

This primarily benefits experts - the people who are able to answer those questions.

While users - the people who ask questions - may float from site to site as they seek information, the burden is on them to find the experts.

If you mash all the questions together then ultimately the experts are going to end up looking at far too many irrelevant questions, and will leave. This will occur even with your proposed advanced tagging, because remember that tagging is performed by the user - the person who, by definition, doesn't know enough about the subject to be expected to tag properly.

By mashing them all together you are giving users permission to poorly tag their questions, while increasing the workload of the experts who then MUST read every question that looks even a little related to their field in order to retag it so it'll be seen by those that can actually answer it.

Further, the reputation system must then be reconfigured. Right now I can look at someones collection of reputation scores across the sites and get a reasonably good picture of how skilled they are on each topic. I can think of a few ways to deal with reputation in your new one-site, super tag based system, and none of them will result in a good way for me to tell if the person who answered my question is an expert or not.

Lastly, there's already too many questions on the more popular site (2,000 a day for SO). Throwing them all together is going to disenfranchise even more experts, and once the experts are gone, the site is useless.

It's also worthwhile pointing out that a new user currently knows they are in the right place because the questions they see before pressing "ask question" look familiar. I will understand immediately that I'm not in the right place if I'm on an auto body website and about to post a question about baking pies.

There are APIs for the sites. Make this "One site to rule them all" using the APIs and demonstrate that it can work, and that it solves these (and all the other problems that will undoubtedly crop up) and perhaps you'll be more convincing. Until then, though, I don't see how a vibrant expert community will benefit from this change.

7
  • 1
    For proper tagging, we can just think of some new ways. I stated some in the comment addressing you. For reputation, you will get the super-tag specific reputation to judge whether this person is a python pro or a photography expert. Just like present reputation, with an overall one. And for last point, though there will be a LOT of questions, someone will see only his/her subscribed tags. IMHO, it will be more specific than the present stackoverflow. I see many questions here I don't want to. And plaese don't tell me to make a whole site JUST to demonstrate "It works".
    – Gulshan
    Commented Oct 21, 2010 at 14:02
  • @Gulshan - Maybe you've got it solved in your head, but as presented it looks like you've just hand-waived away the issues Pollyanna raised.
    – Brad Mace
    Commented Oct 21, 2010 at 14:12
  • 2
    One thing to add. There can be flaw. But please do not think it is just my duty to come up with all the solution. If the whole idea is nice, then everyone can try to solve the issues. Don't just try to find out a loophole. Be positive.
    – Gulshan
    Commented Oct 21, 2010 at 14:19
  • 2
    @Gulshan - You are correct - it's worth thinking about. I've thought about it, and here are the results of my thinking. Don't be the guy in the room that says, "Hey, I only want yes-men, if you don't think it's a good idea, go away." It's all well and good to explore the idea and think about it, but what I'm pushing you to understand is that if you don't resolve the major flaws, or convince others to resolve the flaws, then chances are good that it's not going to go anywhere. Every good idea needs a leader that either resolves the problems, or convinces someone else to resolve them.
    – Pollyanna
    Commented Oct 21, 2010 at 14:42
  • 1
    In the interest of full-disclosure, I too want something similar: meta.stackexchange.com/questions/29382/stack-aggregation - but this particular proposal leaves much to be desired.
    – Pollyanna
    Commented Oct 21, 2010 at 14:49
  • @Pollyanna I am feeling like some are convinced. If not fully, partially at least.
    – Gulshan
    Commented Oct 21, 2010 at 14:55
  • @Ivo Flipse You said you will give a feedback on the proposal. Waiting for it.
    – Gulshan
    Commented Oct 26, 2010 at 9:52
11

My first instinct on reading this question was like George's: "great, another one-site proposal from someone who's never heard of Yahoo! Answers." Now that I've read through the proposal, though, I kind of like it, even though it's far from perfect.

This hypothetical SE 3.0 actually isn't that different than SE 2.0. There are two major changes here that I can see. The first is the universal tagging system, which has obvious flaws but I think merits further thought. I particularly like how it's intended to act more like the labels used by GMail than the folders used by file systems.

The second major change is the centralized interface. That one, I think, could be implemented with less disruption to SE 2.0, would be helpful right off the bat and could be a good barometer for whether it's worth going down the road proposed by the OP. stackexchange.com already displays a list of hot questions from across the network, and I think expansion/customization of that list into a more general portal page could be really helpful.

stackexchange.com already knows who users are, because they're logged in. How about showing a customized question list there for each user? Users could choose to ignore questions from some sites altogether, and maybe choose some favorite sites to get a higher percentage of questions from that site. Just as importantly, the user's personal "interesting" and "ignored" tags from the specific sites could be taken into account.

mockup coming eventually

better mockup possibly coming eventually but don't hold your breath

alt text

13
  • 1
    Stick with the Paint, otherwise the answer gets it.
    – Mark C
    Commented Oct 21, 2010 at 14:28
  • You currently can remove sites from the stackexchange.com listing. Commented Oct 21, 2010 at 14:59
  • 2
    Nice work. Your mockup has, at a minimum, convinced me this isn't a terrible idea
    – Brad Mace
    Commented Oct 21, 2010 at 18:23
  • I think the next major thing that would need to be solved is what the global "Ask a Question" page looks like. How do you ensure people ask their question on the right site? Or would the site be selected based on the question's tags?
    – Brad Mace
    Commented Oct 21, 2010 at 18:29
  • 2
    There's too damned much stuff to look at on that page to determine what site it's on and whether or not I care about it. Not to mention that you're very much dependent on an active editor community to edit the invariably sucky, "Problem With Code" titles. Commented Oct 21, 2010 at 19:12
  • @George: Yeah, it's not a pretty page, but then again, it's just a ten-minute mockup. A lot of the obvious ugly can be toned down or removed entirely. As for "Problem With Code" titles, well, those are going to be an issue whether you're on the portal site or not.
    – Pops
    Commented Oct 21, 2010 at 19:30
  • @Pupular Demand What I had in head was something like Ubuntu Unity Left-hand bar. Icons of different sites or "Super tags" at the left. When clicked they will show that very site with respective theme.
    – Gulshan
    Commented Oct 22, 2010 at 4:15
  • @Gulshan, link?
    – Pops
    Commented Oct 22, 2010 at 4:28
  • @Pupular Demand A left hand bar like- static.arstechnica.com/unity-multi-selector.png To indicate which one is selected- news.softpedia.com/images/news2/…
    – Gulshan
    Commented Oct 22, 2010 at 4:49
  • +1 (if I could twice...) I like your mockup. Maybe the interesting/ignored tabs part could be merged with different per-site layouts as well. And this would definitely require the long-sought myTags tab (or better, myTagsOnly-Checkbox (TM)) Commented Oct 22, 2010 at 7:19
  • 1
    I proposed this type of idea a long time ago as a way to manage multiple niche sites (before they started getting closed down for being... well... too niche). However, what you're talking about is effectively an aggregator and does not require changes to SE itself. It can very easily be built on top of SE using the API. ("Easily" as in concept, not implementation)
    – Aarobot
    Commented Oct 26, 2010 at 21:13
  • @Aarobot, I wasn't aware of your earlier proposal, but the fact that this can be done with the existing SE setup is very much intentional.
    – Pops
    Commented Oct 26, 2010 at 23:05
  • Oh, sorry @Pop, I wasn't intending to imply that it was a formal proposal, it was more like various scattered comments and bits and pieces of answers. And of course it's intentional, that's the point of an API!
    – Aarobot
    Commented Oct 26, 2010 at 23:18
4

I've had the feeling we needed a different way to interact with all the separate Stack Exchange sites from the moment Jeff brought up Quora.

However, I think there are several disadvantages to the proposal that would need to be fixed before it would even be feasible to implement.

  • There can't be questions on any topic, because you need users to answer all those questions too. That's why limiting the scope to several global topics was a good idea and should be maintained even when there's is one global access point for questions.

  • Rep should be based on your knowledge of the sub-domains, though perhaps these could be broader than they currently are, since the overlap between a lot of topics would make the rep more transferable.

    Also, global rep (rep over all SE-sites) should be an indication of how 'trusted' you are on the network. So I think several community features such as editing and voting to close should be based on your global rep rather than your local rep, because you've already proven that you know how the system works.

    From the comments I see others disagree with my assumption, but I think the system is designed in such a way that it's unlikely for users to abuse their abilities without getting spotted. If you are spotted to abuse your powers, moderators are likely to find out and contact you. I have yet to see misuse of these abilities on Super User, Gaming, Web Apps and several other new SE sites. Perhaps these aren't the typical users to abuse their powers, but I like to keep a positive view of our community as a whole and assume this goes for the majority. The exceptions to the rule can always be smacked down!

  • To me this feature is more important for users who answer questions, than the users who ask questions. The askers benefits from clear distinction between subdomains, because the site's branding should make it clear what the questions can be about. The answerers benefit from an easy access point to all the questions that might be of interest for them. There's also a third audience which are lurkers, who find the question through Google and just read the answers. However, this doesn't work when you're looking for new questions to answer as you would almost need to know the question before finding it through Google.

As has been suggested, this problem can be solved through the API and is currently being improved with the new additions to StackExchange.com, where you now get streams with questions all over the network for active, new and unanswered questions. This already solves several of your problems:

  • Users will only need to look at one page to keep track off several sites.

    Though a downsite for me is that the UI isn't really intuitive (what tags can/should I add) and it doesn't make use of all the knowledge the network has about your interests. If only someone could come up with a better algorithm for suggesting questions (other than Stacked Odds, which again isn't network wide).

  • Your point regarding off-topic-ness is more a user issue, than a network issue. You'll have to be patient with all the new beta sites, once they launch they will no longer be unavailable or immature. Furthermore, we simply can't answer every question and therefore strive to limit the questions that are allowed to a subset that is more likely to be answered. This might anger some users, but in the end it's for their own good, since it's likely that it wouldn't get an answer in the first place.

  • By creating one big site, we still don't fix the problem with users not knowing where to ask their questions. Even worse, with your tagging scheme, new users would be overwhelmed with the concept of tagging which would be overly complicated. The new site branding, elevator pitch and tagline should (in the best case) make it a lot more clear what the site is about. In the end, I feel the best way to show what the site is about are exemplary questions that truly reflect the site's image! Sadly, not every site has nailed this.

  • I think the current choice for trying to let new SE sites grow up on their own is a very good choice, because at a certain point the original SO concept no longer scales. It becomes to hard to keep track of every new question and without strict rules, the quality of the questions is likely the first to suffer. By keeping the scope of new topics more strict, the topics have to try and grow their own community. If it works, good for them, if it doesn't well at least nice try! Allowing questions on every topic without having a community of answerers in place is a recipe for disaster (aka Yahoo! Answers).

  • As Pollyanna mentions when having one large feed, solely based on tags, an expert would have to go through every question to find something to answer. Something similar happens when you try to follow activity on several sites (and their meta's), where you simply get overwhelmed by the amount of questions. I don't think that there's currently a good solution for this other than looking for hot questions (which sadly are likely to already be answered before becoming hot).

  • While I agree with Pollyanna that the API might be the solution. The downside is that many non-programmer users are unable to use the API and therefore depend on others to cater their needs. But since most users have a regular job and lack sufficient time to create such a feature, we just have to wait for someone to be willing to take the effort...

In the end, I just hope the SO team will keep adding new features to Stack Exchange to cater to the needs of question answerers that try and keep track off multiple sites. And perhaps we could have another API contest for improving the recommendation of new questions.

1
  • Thank you for the answer. I was waiting. I know that no system will be perfect. And switching to a newer solution just few months after introducing a new one is not feasible either. Now, I don't even hope for that. But I think, despite the disadvantages, if you people ever go for a new stackexchange strategy, may be these ideas can help.
    – Gulshan
    Commented Oct 27, 2010 at 3:35
0

So basically, you're looking for Yahoo! Answers.

So how again is Babby formed?

3
  • 6
    -1, that was my first reaction too, but this proposal seems to be a bit more sophisticated than that.
    – Pops
    Commented Oct 21, 2010 at 13:05
  • 3
    It is true that you can find many all-in-one question-answer site. But I didn't think this from that side. I tried to follow the approach stackexchange is currently following. And then tried to propose a better solution. And with "Super tag" sub-domains, I tried to put the idea of many different communities. Now, stackoverflow is like a single communities dominating all stackexchange sites. I do not like it. And you will not find this thing in other all-in-one qa sites. Please read the proposal again and provide some constructive feedback.
    – Gulshan
    Commented Oct 21, 2010 at 13:13
  • do you really want users and mods to be wasting their time playing ping-pong migrating questions back and forth between sites? (Yes, some questions are receiving votes to be migrated more than once now that SE has become so ghettoised, e.g. superuser.com/questions/274444 .)
    – user136089
    Commented Apr 26, 2011 at 2:48

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .