[marbletiles]

They realised that StackExchange was failing, but because they don't realise the reasons why their attempted correction is only making it worse.

Their software is a hideously complicated and over-engineered attempt to twist human relationships into math. It only works on StackOverflow because: a) The tech community was desperate for an alternative to hidebound mailing lists on one hand and expertsexchange on the other b) How to put this? A whole lot of nerds really would like to be able to reduce the complexity of human relationships to math, too, and willingly participated.

But without a userbase that's dying for a solution, any solution, and especially a userbase prepared to put up with convoluted ranking-rating-have-I-got-enough-points-to-change-my-profile-picture-yet point-scoring games the software is actually a millstone. You're not going to get a liberal arts Q&A site that takes off with those restrictions. This is why StackExchange was such a dud.

By not realising this, their solution is more of the same! "Sure, you can start a site, you just need pi+4 users to seed your initial contract bounding, then that will need to be ranked to 6 by a quorum of level 3 users, and after an initial 26-day period of zzzzzzzzzz <click>".

You want to create a good Q&A site, you need to have a community, and it needs to be well-tended by empathic people who know how and where to prune. The software is pretty much irrelevant. Look at http://ask.metafilter.com/ for a success story: totally flat, forum-esque, but answers are obvious, there's no chatter or bullshit, and it works on the most amorphous and wide-ranging types of questions.

There is no shortcut solution to this problem. There is no way to mathematically manage human connections like this that works in this space. The route to success is careful relationship management, not yet more programming.

[spolsky]

Sounds like you have a great business plan. In the meantime, I think that the Stack Overflow model has proven itself among nerds, and I think that your assertion that nerds are weird in some way and need completely different software than the rest of the world is not backed up by any evidence.

I disagree that the software is irrelevant. Discussion groups that don't allow voting have no way to distinguish answers that the community thinks are good from answers that the community thinks are bad. Discussion groups that don't allow editing have no way to change answers as the world changes, so wrong answers stick around. Discussion groups without tags are forced to splinter communities into smaller and smaller fragments because they have no way of dealing with overlapping communities. Discussion groups without reputation systems are overrun with spam.

I can't think of anything I disagree with MORE than the concept that "the software is irrelevant." The software DEFINES how the community works with each other and is absolutely critical.

[marbletiles]

I didn't say that nerds need completely different software: I implied that nerds were the only people who would put up with SO-style numbers games, because their need was so great and they can grok the system.

Nor did I say the software per se was irrelevant: I agree it's vastly important to how a community interacts. But when it comes to community building it's beside the point: phpBB is very bad software for discussion, yes, but some excellent communities have formed nonetheless.

Good software facilitates communication and -- crucially -- it enables readers to use their established social skills. It stays out of the way, in other words. Metafilter, for example, has none of your "requirements" apart from a basic tagging system, but avoids every one the problems you think will result -- and it does it from nothing more than good relationship management and community stewardship.

By trying to automate away (or disperse to the "crowd") the hard work of that relationship management, StackOverflow has boxed itself into a niche where only nerds-with-a-need could bear to live. The rest will turn away and keep on searching, as the experience of StackExchange has shown.

[edit: first attempt made no sense!]

[spolsky]

Maybe you have different goals than us. You're trying to build communities... we're trying to make the Internet a better place to get expert answers.

[marbletiles]

Surely you have to build communities to do this, though? Your earlier answer mentioned how you needed voting so the "community" could choose best answers, you needed tags to keep sub-communities separate and so on. And one of the rating criteria for the new SE sites seems to be how much of a community they manage to create.

If it is the case, I think this is the fundamental tension: You've built software geared to rating and creating answers, but to get those answers you need vibrant communities -- and the resulting software is so complex and strictured it is effectively anti-community.

[netcan]

I'm siding with spolsky here. The ultimate goal you have in mind is defining. Since building a community is hard and is the hurdle that kills most attempts, people will often see it as the goal. In many cases, it might be the goal.

Take Wikipedia as an example. Community is necessary but it is not the goal. The goal is encyclopaedia making. Most online communities do not produce a wikipedia.

What this adds up to, in theory, is sacrificing some community building ability (more sites will dies from under participation)for more Q&A ability. While more of the remaining will produce a good archive of useful answers.

*This doesn't directly answer your original claim that the software is good for SO specifically but cannot be widely applied. But, if what I suggest is true, then you would expect it to appear that way.

[stse]

I think your reading to much into this. To me it seems like the stack exchange sites just got too little traffic to reach critical mass. Ask Metafilter is an exception to a lot of other communities which became awful as they grew.

[richardw]

No matter how good the software is, it's a short-term advantage. It gets you a ticket to the game but confers no sustainable advantage because the good bits will get copied. What matters is what you do on top of that software to build sustainable advantage.

(To be clear: I'm not referring to something like Windows here. SE is obviously complex but it isn't as complex as Windows. Much easier to copy and innovate around.)

I see it as a race now. You have to get that question and answer up before the other platforms do, to lock-in the Google advantage. You'll be leveraging your existing expert network in a sideways drift. Those experts already answer questions madly (and there are very few on the web) and now they have to battle to get their idea off the ground. Good plan. :)

You are missing out those snowflake sites (sorry Patrick) where the expert has zero SE cred but is the world expert on model trains just as model trains become the next bing thing. Still, over time you should have a reasonable monopoly on people who desire karma.

Unless someone comes up with a mechanism that can a) leverage karma whores better and b) discover and resolve hard topics faster.

[billybob]

1) I don't think they're trying to model 'the complexity of human relationships.' That's what Facebook tries to do. The trust metrics are crude, but the real goal seems to be to get credible upvotes for answers.

I use SO all the time, have great success, and haven't made any friends there. Because that's not what it's for.

2) Maybe this model won't work for EnthusiasticCatBreeders.com, but it will probably work for a lot of sites. Maybe it will self-select for topics where the people interested are a bit nerdy.

That's OK. There is still a lot of room for nerdy growth. I can imagine sites about cell phones, economics, geomapping, and lots of other topics where the audience is a bit nerdy, there are right and wrong answers, and this will probably work.

[marbletiles]

Just to expand a bit -- when I talk about modelling the complexity of human relationships I don't mean making friends, I mean how they try to turn the complicated ways we have of judging and trusting other people into a number.

You can do that on eBay, because the metric is really simple -- "did this user deliver, or did they rip you off?" But when it comes to judging the value of technical advice, it's much muddier.

We have thousands of years of experience at these sorts of judgements though, and can size people up in the blink of an eye. It's much more difficult online, but we're learning -- and numbers aren't really a part of it. Certainly not the morass of numbers of StackOverflow.

[spolsky]

Still, when you look at how answers are ranked on Stack Overflow, it really seems to work. The top ranked answers really are the best... far more than on a traditional PHPBB type site.

[bmm6o]

I think most people here agree that SO is head-and-shoulders above its competition. I can point to specific things about it that I don't like, but in the end I come back and I appreciate the content. What I am curious to see (and I think this is roughly what bonaldi is talking about) is how well the SX software supports other less-technical communities.

My main concern is that SO had such a large and willing user community from day 1, that you could survive problems in converting visitors into active participants just through sheer numbers. For a nascent community, there are the karmic barriers to entry (e.g. I've been a casual SO user for over a year, and am still not able to fix someone else's spelling mistake!). Also, the question domain of SO lends itself to being able to ask a fairly concise question and getting a decent answer with a minimum of back-and-forth. This is important, because SX handles discussion so poorly (and I realize that's by design).

Anyway, I hope it works out for you - I do admire your goals and ambition. I just hope you keep an open mind about what the software needs to do to best serve the new communities, and adjust it accordingly.

[blasdel]

Applying your magic sort() encourages replies by users who want to play a game. That only works if those users exist and actually have useful contributions to make — both of those properties are much diminished outside the SO population.

You don't to rank answers unless you have unproductive bullshit in the replies. While programmers might love the karma stuff, everyone else prefers it if the bullshit was just not tolerated by the community and deleted by moderators.

[jasonwatkinspdx]

I think it has more to do with not being able to leverage their personal blog audiences against other topic areas. It was easy for them to seed Stack Overflow just by talking about it on their blogs.

[netcan]

That is true for any community site. Shouldn't disadvantage them in particular.

[10ren]

Google gave me metafilter answers for my non-technical questions that were really excellent a few times; but I hadn't made that connection til your comment. Time to try it directly!

A SO site might perform well as a support channel for a software library product. I've used a mailing list in the past, but it's not as a good, in many ways. If it's free for commercial use, I might give this a go. It seems ideal to deliver support for an open-source software product - anyone done this?

[proee]

Why is Stack Exchange considered a failure? Don't they have a ton of exchange sites (if not thousands). $100/month x 1k = $100k Month in revenue.

Seems they would be quite happy with that sort of recurring revenue...

[alanl]

I think a lot of the sites were dead or dying. I'm Looking forward to seeing the new more open model. I think it could produce some useful Q&A sites.

[icey]

AFAIK they haven't actually started charging for Stack Exchange sites yet.

[proee]

So why not flip the switch then?

[dustingetz]

hardly any SE sites have a community worth paying for, which is kind of the whole point

[proee]

Well there's plenty of sites owners (with credit cards) that might assume otherwise. It depends on the target audience, but if it was for say doctors or lawyers 1/10th the traffic or even 1/100th the traffic of a typical programming site would be a success.

Maybe they did a survey among all the sites to come to your conclusion, but it seems odd they never tried go paid.

[loikujhygthj]

The people that could afford a site aren't the people that answer questions for free.

The main problem was that there was no way of charging the people who asked questions - who were the only ones likely to want to pay.